A Robust Feature-Based Digital Image Watermarking Scheme

Size: px
Start display at page:

Download "A Robust Feature-Based Digital Image Watermarking Scheme"

Transcription

1 Chpter 6 A Robust Feture-Bsed Digitl Imge Wtermrking Scheme 6.1 Introduction Attcks hve been developed to destroy wtermrks. These ttcks on wtermrks cn roughly be clssified s geometric distortions nd noise-like signl processing. Geometric distortions re difficult to tckle. They cn induce synchroniztion errors between the extrcted wtermrk nd the originl wtermrk during the detection process, even though the wtermrk still exists in the wtermrked imge. Nowdys, severl pproches tht counterttck geometric distortions hve been developed. These schemes cn be roughly divided into invrint trnsform domin bsed, moment-bsed nd feture extrction bsed lgorithms. Wtermrk embedded in invrint-trnsform domins generlly mintin synchroniztion under rottion, scling nd trnsltion. Exmples of these trnsforms re given in subsection The wtermrk detection process is similr to the pttern recognition process in computer vision, but the originl imges my not be vilble to the wtermrk detector. Moments of objects hve been widely used in pttern recognition. The discussion of moment-bsed wtermrking techniques is found in subsection On the other hnd, the extrcted feture of imge content cn be used s reference points for both wtermrk embedding nd detection s 87

2 illustrted in subsection In this chpter, we develop robust wtermrking scheme. This scheme combines the dvntges of feture extrction nd imge normliztion to resist imge geometric distortion nd to reduce wtermrk synchroniztion problem t the sme time. Section 6. describes the feture extrction method used in the proposed scheme. In Section 6.3, the imge normliztion process developed for pttern recognition is briefly reviewed. Section 6.4 contins the description of our wtermrk embedding procedure. Section 6.5 covers the detils of the wtermrk detection procedure. Simultion results in Section 6.6 will show the performnce of our scheme. Finlly, Section 6.7 concludes this presenttion. 6. Feture Extrction To detect wtermrks without ccess to the originl imges, we look for reference points tht re perceptully significnt nd cn thus resist vrious types of common signl processing, such s JPEG compression nd geometric distortions. These reference points cn lso ct s mrks for (loction) synchroniztion between wtermrk embedding nd detection. In this pper, we will use the term feture points to denote these reference points. In our scheme, we dopt feture extrction method clled Mexicn Ht wvelet scle interction. This method ws originlly used in [6][66][84]. It determines the feture points by identifying the intensity chnges in n imge. Since significnt intensity chnges (edges) my occur t different scled versions of the sme imge, Mrr nd Hildreth suggested tht different opertors should be used t different scles for optimlly detecting significnt intensity chnges. The Mexicn Ht wvelet (Mrr wvelet) [33][34] is rottion invrint wvelet. It hs circulrly symmetric 88

3 frequency response. The computtionl cost is high becuse this wvelet is not seprble. In fct, it is the negtive Lplcin of Gussin function. The wvelet nlysis filter is loclized t different frequencies nd sptil scles (resolutions). The Mexicn-Ht mother wvelet t loction x r is defined by (6.1): r r r x / Ψ ( x) = ( x ) e, (6.1) r. The two-dimensionl Fourier trnsform of ( ) where 1/ x = ( x + y ) ψˆ x r is given by r r r k / ψˆ ( k ) = k e, (6.) where k v represents the -D sptil-frequency. The feture extrction method proposed in [6][66] uses the following quntities: r r r P ( x) M ( x) M ( x), ij = γ (6.3) i i i r M ( x) = ( Ψ( x)) A, i j r (6.4) r where (x) represents the response of the Mexicn Ht wvelet opertor t sptil M i loction x r of scle i, γ is scling prmeter, (x) P ij r is the scle interction between two different scles i nd j, A is the input imge, nd denotes the convolution opertion. Our scheme is designed for both color nd gry-level imges. For color imges, the Y component is extrcted for wtermrk embedding. The Mexicn Ht wvelet filtering is implemented in the frequency domin using the FFT. An input imge is first zero-pdded to in size. We void selecting feture points locted ner borders of n imge. Hence, prohibited zone long the imge border is predefined. Thus, border effects re negligible in extrcting the feture points. Exmples of filtered imges t two different scles re shown in Figs. 6.1() nd 6.1(b). The difference of these two filtered imges is the Mexicn Ht scle 89

4 interction imge (with γ =1), shown in Fig. 6.1(c). The two scles we choose re suggested by [6][66]; tht is, i = nd j =4. Feture points re defined s locl mxim inside disks in the scle interction imge. The disk rdius is chosen to be 45, which is determined experimentlly. Feture points locted in regions of smll vrince re discrded for reducing wtermrk visibility. A flowchrt of the feture extrction method is given in Fig. 6.. Among the mny feture extrction lgorithms proposed in the literture, we hve dopted the scheme proposed in [6][66] for severl resons. First, since the Mexicn Ht wvelet scle interction is formed by two scles, it llows different degrees of robustness (ginst distortion) by choosing proper scle prmeters. Second, since locl vritions such s cropping or wrping generlly ffect only few feture points in n imge, the unffected feture points cn still be used s references during the detection process. Third, this wvelet function is rottionlly-invrint. It mens tht most feture points my not chnge fter imge rottion. Fourth, since the Mexicn Ht wvelet is essentilly bnd-limited, the noise sensitivity problem in feture extrction cn be reduced. Finlly, the extrcted feture points do not shift their loctions much under high-qulity JPEG compression s discussed in [66]. These feture points re the centers of the disks tht re to be used for wtermrk embedding (s described in the next section). Exmples of disks re shown in Fig. 6.1(d). Since these disks should not interfere with ech other, we only select the feture points tht re wy from ech other to crete non-overlpped disk set. In our scheme, feture point hs higher priority for wtermrk embedding if it hs more neighboring feture points inside its disk. 90

5 () (b) (c) (d) Fig () Mexicn Ht wvelet filtered imge t scle i =. (b) Mexicn Ht wvelet filtered imge t scle i =4. (c) The difference imge between () nd (b). (d) The center of ech disk is feture point. 91

6 Originl Imge Zero Pdding -D FFT Extrcted Feture Points Mexicn Ht Wvelet Filtering t Scle i Mexicn Ht Wvelet Filtering t Scle j Identify Nonoverlpped Points -D IFFT -D IFFT Serch for Locl Mximum Feture Points Recover the Originl Imge Size Recover the Originl Imge Size Mexicn Ht Wvelet Scle Interction Fig. 6.. Feture extrction by Mexicn Ht wvelet scle interction. 6.3 Imge Normliztion The imge normliztion technique developed for pttern recognition cn be used for digitl wtermrking s suggested in [58]. Severl geometric centrl moments re computed to trnsform the input imge to its normlized form. The normlized imge (object) of rotted imge (object) is the sme s the normlized imge of the originl imge (if no pdding or cropping occurs). Since objects re rottionlly invrint in the normlized imge, the wtermrk detection process cn be much simplified when it is pplied to the normlized imge. On the other hnd, becuse imge normliztion is sensitive to locl imge vritions, detection is more ccurte when pplied to individul objects rther thn the entire imge. In our scheme, we pply the imge 9

7 normliztion process to ech non-overlpped locl disk seprtely. The centers of these disks re the extrcted feture points described in Section II. Imge normliztion technique is used for selecting the loction of the wtermrks. However, wtermrks re not embedded in the normlized imges. This is becuse sptil interpoltion is necessry for mpping the originl imge pixels to the normlized imge pixels, nd vice-vers. This interpoltion process induces significnt mount of distortions nd thus reduces wtermrk detectbility. The detils of the imge normliztion process cn be found in [85]. Here, we only briefly describe its computtionl steps. The prmeters below re computed once for ech imge disk. 1) Men vector [ C xc ] T y, where C x = Ω p( x, y) xf ( x, y) dxdy, C y = Ω yf ( x, y) dxdy, f ( x, y) =, p( x, y) dxdy Ω where p( x, y) denotes the gry-level vlue t loction ( x, y), ndω is the region of interest. u u 0 11 k r ) Covrince mtrix M =, where u = Ω ( x C ) ( y C ) f ( x, y dxdy. 11 3) Centrl moments u 30, u1, u1, u03 of the originl disk. u u 0 kr x y ) T 4) Eigenvlues λ1, λ nd their ssocited eigenvectors [ e 1x e 1y ] of M. 5) Two ffine trnsformtion coefficient mtrices 93

8 c λ e x e 1 1 1y = c 1 e y e 0 1 1x λ c c e1 x e1 y = λ1 λ1 c c e1 y e1 x λ λ c 0 b1 = λ e x e y C x b c e1 y e1 x C y 0 λ 11 1 Cx = 1 C y where 1/ 4 c = ( λ1 λ ). 6) Centrl moments for clculting rottionl invrint trnsformtion: u' u' u' u' = = = = u u u u ( + ( 1 111u u u 03 1 u ) u 1 ) u ( + ( 3 u ) u ) u u u ) Tensors: t u' 1 + u' 30, t = u' 03 + u' 1 1 t Angle: α = tn ( ) t =. 1 8) Tensor t = t sin α + t cosα 9) If < 0 then α = α + π t. 1 Finlly, the normlized imge is computed from the originl imge bsed on the following coordinte trnsformtion: x cosα = y sinα sinα cosα c λ e 1 c e λ x 1y e 1y e 1x x C y C x y, where ( x, y) is the originl disk coordintes, nd ( x, y) is the normlized disk 94

9 coordintes. The normlized imge object is insensitive to trnsltion, scling nd rottion of the originl imge object [85]. After coordinte trnsformtion, ech disk becomes disk. Rectngulr windows used to hold wtermrks in the originl imge disks re constructed s follows. Two 3 3 blocks in ech (originl) imge disk re chosen for wtermrk embedding. The loctions of these 3 3 blocks re determined through the use of the normlized imge disk. Two ordered points A nd B re chosen t integer coordintes inside the normlized imge (disk) s shown in Fig. 6.3(). The loctions of these two points re chosen secretly but re known to the wtermrk detector. The loctions of A nd B re chosen close to the boundry of the normlized disk, nd the distnce between these two points is 3. Points nd b locted in the originl imge re the inverse mpping of A nd B (on the normlized imge) s shown in Fig. 6.3(b). Usully, points of the inverse mpping of A nd B do not hve integer coordintes, nd thus, points nd b re quntized to integers. They re connected to form line segment b. Although the distnce between points A nd B is 3, the distnce between nd b is generlly different due to the normliztion process. Therefore, b is shortened or extended to the line segment b ', which hs length 3. Usully, point b ' is not the sme s point b, but these two points re close. Then, 31 line segments prllel to b ' re creted running towrds the center of the disk. Finlly, b ' nd its 31 prllel line segments of length 3 form 3 3 block in the originl imge s shown in Fig. 6.3(c). Since the 3 points tht line segment psses through do not lwys hve integer coordintes, we choose 3 integer-coordinte pels nerest to the line segment to form the discrete-grid line segment s shown in Figs. 6.4() nd 6.4(b). The crossing points of grid represent integer-coordinte pels in the originl imge (disk). If the bsolute vlue of the slope of line segment is less thn 1, its discrete-grid pproximtion is 95

10 constructed long the horizontl direction s shown in Fig. 6.4(). Otherwise, the verticl direction is used s shown in Fig. 6.4(b). Two 3 3 blocks re selected for ech disk s shown in Fig. 6.3(d). To reduce the impct of feture point shift due to wtermrk embedding, these blocks should not contin the disk center (feture point). All the loction informtion of these two blocks is determined on the normlized imge (disk). After the coordintes of A nd B re determined s described bove, the coordintes of C nd D will be the symmetric pels with respect to the symmetric center C e (Fig. 6.3()). C e is not necessry the center of the disk. Point E is the middle point of A nd B. AB is perpendiculr to EC e. The distnce between points A nd C e is less thn 45 but greter thn 3. The distnce between E nd C e hs to be greter thn 3. Next, the corresponding pels c nd d in the originl imge disk re computed by the inverse normliztion trnsformtion. A shortened or extended line segment of cd is c' d, which contins 3 pels. The blocks selected for the imge Len re shown in Fig Occsionlly, tiny corner (very few pels) of 3 3 block my be outside the originl imge disk. If this hppens, these pels re not wtermrked. Another potentil problem is tht lthough the extrcted feture points (center of the disk) re locted in high-contrst regions, the two 3 3 selected blocks my be prtly locted in smooth regions. Therefore, to keep wtermrk imperceptibility, such disk is not wtermrked if the vrince of one 3 3 block in n originl imge disk is smll. In our experiment, there re only 8 qulified disks (Fig. 6.5) for wtermrk embedding lthough there re 11 feture points (disk centers) re extrcted on the Len imge (Fig. 6.1(d)). 96

11 C A D C e E B c b d () (b) c' d b' b' (c) (d) Fig () Two ordered points A nd B in the normlized imge (disk). (b) Two corresponding points nd b in the originl imge (disk). (c) A 3 3 block is constructed in the originl imge disk. (d) Two symmetric 3 3 blocks in the originl imge disk re formed. () (b) Fig The crossing points of the grid represent the integer pel loctions on the originl disk. () If the slope (bsolute vlue) of line segment is less thn or equl to 1, the integer pels closest to the line segment horizontlly re chosen to form the dt line segment. (b) If the slope (bsolute vlue) of line segment is greter thn 1, the integer pels closest to the line segment verticlly re chosen to form the dt line segment. 97

12 Fig Ech disk contins two 3 3 blocks for wtermrk embedding (Len). 6.4 DFT Domin Wtermrk Embedding Our wtermrk is designed for copyright protection. We view ll blocks s independent communiction chnnels. To improve the robustness of trnsmitted informtion (wtermrk bits), ll chnnels crry the sme copy of the chosen wtermrk. The trnsmitted informtion pssing through ech chnnel my be disturbed by different types of trnsmission noise due to intentionl nd unintentionl ttcks. During the detection process, we clim the existence of wtermrk if t lest two copies of the embedded wtermrk re correctly detected. Originl Imge Extrct Feture Points Compute Imge Normliztion Prmeters on Disks in the Originl Imge Trnsform Coordintes of Selected Points from Normlized Imge bck to the Originl Imge Construct Two 3x3 Blocks in Ech Qulified Disk in the Originl Imge Wtermrked Imge Replce the Disks in Originl Imge by the Wtermrked Ones -D IFFT Wtermrk Embedding -D FFT on 3x3 Blocks Wtermrk nd Secret Key K Fig Wtermrk embedding scheme. 98

13 The wtermrk embedding process is outlined in Fig First, the feture extrction method genertes reference centers of disks for wtermrk embedding nd detection. We then perform the imge normliztion technique on disks in the originl imge. The coordinte trnsformtion coefficients between the originl imge disks nd the normlized disk imges re generted. The loction of blocks in the originl imge for wtermrk embedding is determined from the normlized imge. Then, coordintes of selected points re trnsformed from normlized imge bck to the originl imge. As result, the wtermrk synchroniztion problem during the detection process is reduced. Next, -D FFT is pplied to these 3 3 blocks on ech qulified disk in n originl imge. The wtermrk is embedded in the trnsform domin. Lst, the wtermrked blocks re -D IFFT converted bck to the sptil domin to replce the originl imge blocks. The procedure of selecting nd modifying the mgnitude of DFT coefficients for wtermrk embedding is illustrted below. First, the FFT is pplied to ech 3 3 selected block. Then, severl middle DFT coefficients re selected ccording to the secret key K. Middle frequency components re generlly more robust in resisting compression ttcks. A modified version of [6] is used to embed wtermrk bits into DFT coefficients. Selected pirs, x, y ) nd y i, x ) ( i i ( i upper hlf DFT plne (Fig. 6.7) re modified to stisfy, o F '( xi, yi ) F '( yi, xi ) α if wmi F '( x, y ) F '( y, x ) α if wm i i i i i 90 prt, locted on the = 1 = 0, where F '( x i, y ) nd F' ( y i, x ) re the mgnitudes of the ltered coefficients t i i loctions x, y ) nd y i, x ) in the DFT trnsform domin, α is the wtermrk ( i i ( i strength, nd wm i is the binry wtermrk bit, which is either 0 or 1. The phse of the selected DFT coefficients is not modified. If the wtermrk bit is 1 nd the originl mplitude difference between points x, y ) nd y i, x ) is greter thn α, 99 ( i i ( i

14 no chnge is needed. Also, to produce rel-vlued imge fter DFT spectrum modifiction, the symmetric points on the lower hlf DFT plne hve to be ltered to the exct sme vlues, too. A lrger vlue of α nd longer wtermrk sequence length would increse the robustness of the wtermrking scheme. Becuse the 3 3 blocks re selected in the high vrince imge regions, typiclly the embedded wtermrk is less visible for smller α. Hence, there is trdeoff between robustness nd trnsprency. In our cse, we embed 16 bits in ech 3 3 block. ( i y i, x ) 0 y ( x i, yi ) x Fig Two points x, y ) nd y i, x ) ( i i ( i used for embedding one wtermrk bit., o 90 prt, on the upper hlf DFT plne re The secret key K shown in Fig. 6.6 is lso known to the wtermrk detector. This secret key is used s the seed for generting rndom numbers to specify the frequencies of the DFT coefficients used to hide wtermrk bits. 6.5 Wtermrk Detection The block digrm of our wtermrk detection scheme is shown in Fig The wtermrk detector does not need the originl imge. The feture (reference) points re first extrcted. The feture extrction process is similr to tht used in the wtermrk embedding process. All the extrcted feture points re cndidte loctions of embedded bits. Since imge contents re ltered slightly by the embedded mrks 100

15 nd perhps by ttcks too, the loctions of extrcted feture points my be shifted. In ddition, some of the originl feture points my fil to show up during the detection process. If the feture point shift is smll, the embedded wtermrk blocks cn still be extrcted correctly. Received Imge Feture Extrction Compute Imge Normliztion Prmeters on Ech Disk in the Received Imge Trnsform Coordintes of Selected Points from the Normlized Imge bck to the Received Imge Construct Two 3x3 Blocks in Ech Disk of the Received Imge Wtermrk nd Secret KeyK -D FFT Wtermrk Detection Wtermrk Detection Decision Fig Wtermrk detection scheme. Imge normliztion is pplied to ll the disks centered t the extrcted cndidte reference points. Two 3 3 blocks re extrcted in ech disk. The loctions of these 3 3 blocks re the sme s those specified t wtermrk embedding. The coordinte trnsformtion coefficients between the originl imge disk nd the normlized disk imge re generted. Thus, the loction of blocks in the received imge is determined from the normlized imge, nd the coordintes of the selected points re trnsformed from normlized imge bck to the received imge. In ech 3 3 DFT block, 16 wtermrk bits re extrcted from the DFT components specified by the secret key. For n extrcted pir of DFT coefficients, ( i i x, y ) nd y i, x ) formul, ( i, the embedded wtermrk bit is determined by the following 101

16 1 if F"( xi, yi ) F"( yi, xi ) 0 wm =, i 0 if F"( xi, yi ) F"( yi, xi ) < 0 where F ( x i, y ) nd F ( y i, x ) re the mgnitudes of the selected coefficients t " i " i loctions x, y ) nd y i, x ). The extrcted 16-bit wtermrk sequence is then ( i i ( i compred to the originl embedded wtermrk for deciding success detect. Two kinds of errors re possible in the detector: the flse-lrm probbility (no wtermrk embedded but detected hving one) nd the miss probbility (wtermrk embedded but detected hving none). There is trde-off between these two error probbilities in selecting detector prmeters. Typiclly, reducing one will increse the other. It is rther difficult to hve exct probbilistic models of these two kinds of errors. Simplified models re thus ssumed in choosing the detector prmeters s shown below. We first exmine the flse-lrm probbility. For n unwtermrked imge, the extrcted bits re ssumed to be independent rndom vribles (Bernoulli trils) with the sme success probbility, P success. It is clled success or mtch if the extrcted bit mtches the embedded wtermrk bit. We further ssume tht the success probbility, P success, is ½. Let r 1 nd r be the numbers of mtching bits in the two blocks on the sme disk, nd n the length of the wtermrk sequence. Then, bsed on the Bernoulli trils ssumption, r 1 nd r re independent rndom vribles with binomil distribution, nd P 1 = n n! r1!( n r 1 r 1 )! 1 = The men vlues of r 1 nd r re both n/. P n n! r!( n 10 )!. r r A block is climed wtermrked if the number of its mtching bits is greter thn

17 threshold. The thresholds for the two blocks on the sme disk re denoted by T 1 nd T. Clerly, T 1 nd T should be greter thn n/, the men vlues of r 1 nd r. The flse-lrm error probbility of disk is, therefore, the cumultive probbility of the cses tht r1 T1 nd r T. In order to control the level of flse-lrm probbility by one djustble prmeter, third threshold T is introduced. More precisely, the vrible pirs, r 1 nd r, shll stisfy the following two criteri simultneously: (1) r1 T1, r T nd () r 1 + r T. Tht is, P Flse lrm on one disk = r1 = n, r = n 1 n n! 1 n!. (6.5) r1!( n r )! r!( n r )! r1 = T1, r = T 1 r1 + r T n Furthermore, n imge is climed wtermrked if t lest m disks re detected s success. Under this criterion, the flse-lrm probbility of one imge is P Flse lrm on one im ge = N i N i N ( PFlse lrm on one dis k ) (1 PFlse lrm on one dis k ), (6.6) i= m i where N is the totl number of disks in n imge. We cn plot P ginst vrious T vlues s shown in Fig. 6.9 Flse lrm on one mge using (6.6). The other prmeters re chosen bsed on our experiences: n = 16, N = 10, m = 3, T 10 nd T 10. The curve in Fig. 6.9 drops shrply for T > 3. It is often 1 = = desirble to hve very smll P. However, the selection is ppliction Flse lrm on one mge dependent. We ssume tht 5 P should be less thn 10. In this cse, Flse lrm on one mge T should be greter thn or equl to 4 nd t T = 4, 6 P is Flse lrm on one mge We next exmine the miss probbility. In n ttcked wtermrked imge, we gin ssume tht the mtching bits re independent Bernoulli rndom vribles with equl success probbility, P success. This my not be very ccurte model but it seems to be sufficient for the purpose of selecting the detector prmeters. The 103

18 success detection probbility of r 1 bits in block of n wtermrked bits is Similrly, for the second block r! 1 n r n 1 P = ( P ) (1 P ) 1 r1!( n r. r success success 1)! r! n r n P = ( P ) (1 P ) r!( n r. r success success )! The success detection probbility of disk is the cumultive probbility of ll the cses tht r1 T1, r T nd r 1 + r T. Tht is, P Success on one dis k = r = n, r = n 1 r1 r1 = T1, r = T r + r T 1 P P r. (6.7) Recll tht n imge is climed wtermrked if t lest m disks wtermrk detected. Under this criterion, the miss probbility of n imge is P Miss on on e imge = N i N i N 1 ( PSucesss on one disk ) (1 PSucesss on one disk ). (6.8) i= m i It is difficult to evlute the success detection probbility of wtermrked bit, P success. It depends on the ttcks. For exmple, the distortion induced by JPEG compression cnnot be modeled by simple dditive white Gussin source. However, typicl success detection probbility my be estimted from the experiments on rel imges with ttcks. Becuse we like to see the detector performnce under geometric distortion, modertely difficult cse is chosen from Tble imge Len under combined distortions of 1 degree rottion, cropping nd JPEG compression t qulity fctor of 70. The simultion is done using 10 wtermrked imges Len imposed with (rndomly generted) different wtermrks. The selected vlue of P success is the totl number of mtching bits divided by the totl number of embedded bits. In this experiment, we obtin P success = Bsed on 104

19 this P success vlue, we plot the miss probbility of n imge for vrious T s shown in Fig In this experiment, we set gin n = 16, N = 10, m = 3, T 10 1 = nd T = 10. The curve goes up shrply for T > 3. For T = 4, P Miss on one imge is less thn 0.4. Clerly, from Figs. 6.9 nd 6.10 we cn see the trde-off in selecting T. Suppose tht lower flse-lrm probbility is our higher priority in the simultions in Section VI, T is therefore chosen to be 4 so tht P is less thn Flse lrm on o ne mge Fig The flse-lrm probbility of n unwtermrked imge. The probbility is generted for r 1 nd r stisfying the following two conditions: (1) r 1 10, r 10, nd () r 1 + r T. (n =16, m=3, nd N=10.) 105

20 Fig The miss probbility on imge Len. The plot is generted for r 1 nd r stisfying the following two conditions: (1) r 1 10, r 10, nd () r 1 + r T. (n = 16, m=3, nd N=10.) 6.6 Simultion Results We test the proposed wtermrking scheme on the populr test imges Len, Bboon nd Peppers. We use StirMrk 3.1[78] to test the robustness of our scheme. The StirMrk 3.1 ttcks cn roughly be clssified into two ctegories: common signl processing nd geometric distortions. The difference imges between the originl imges nd the wtermrked imges in the sptil domin re mgnified by fctor of 30 nd shown in Figs (), (b), nd (d). The PSNR vlues between the originl nd the wtermrked imges re 49.4 db, db, nd db for Len, Bboon nd Peppers, respectively. Becuse of their smll mplitudes, the embedded wtermrks re invisible by subjective inspection. Recll tht the rdius of ech disk in the normlized imges is 45, nd tht two 3 3 blocks re chosen in ech disk for wtermrk embedding. In ech 3 3 squre, the embedded 16 frequencies (of the DFT coefficients) re locted within the shded re of Fig

21 All blocks re embedded with the sme 16 bits wtermrk. The wtermrk strength α is set to 0, 15 nd 10 in Bboon, Len nd Peppers, respectively, for compromise between robustness nd invisibility. Since Bboon imge hs more texture, strong wtermrk is less visible thn in Len nd Peppers. The number of wtermrked imge disks is 11, 8 nd 4 in Bboon, Len nd Peppers, respectively. The more textured the imge is, the more extrcted feture points the imge hs. Simultion results for geometric distortions nd common signl processing ttcks re shown in Tbles 6.1 nd 6., respectively. The tbles show the number of correctly detected wtermrked disks nd the number of originl embedded wtermrked disks. As shown in Tble 6.1, our scheme cn resist JPEG compression up to qulity fctor of 30. The JPEG compression quntiztion step size used in StirMrk is defined by Scle = 5000 / qulity 00 qulity, if qulity <, otherwise. 50 QunStepSize[ i] = ( BsicQunMtrix[ i] Scle + 50) /100. Our scheme performs well under other common signl processing ttcks such s medin filtering, color quntiztion, 3 3 shrpening, nd Gussin filtering. It cn lso resist combined signl processing nd JPEG compression ttcks t qulity fctor of 90. Some of the signl processing opertions used in StirMrk 3.1 re detiled below. Color quntiztion is similr to tht in GIF compression. The 3 3 Gussin filter mtrix is The 3 3 sptil shrpening filter mtrix is The wtermrk robustness ginst common signl processing is much improved with stronger wtermrk strength, but there is the trdeoff between wtermrk robustness nd invisibility. An dditive noise ttck ws lso pplied to the wtermrked imge. The ttcked 107

22 imge is: L' ( x, y) = L( x, y) (1 + β n( x, y)), where L ( x, y) is the luminnce pixel vlue of n input imge t ( x, y), β is prmeter tht controls the strength of the dditive noise, n ( x, y) is noise with uniform distribution, zero men, nd unit vrince, nd L '( x, y) is the luminnce pixel vlue of the ttcked imge t ( x, y). In our experiment, the dditive noise is visible especilly in the imges Len nd Peppers, when β is greter thn 0.1. The wtermrk cn be detected when β is less thn 0.. As stted in Section 6., the noise sensitivity problem in feture extrction is reduced due to the essentilly bnd-limited property of Mexicn Ht scle interction scheme with proper prmeter settings. The PSNR vlue (comprison between the wtermrked imge nd the ttcked imges) in Tble 6.3 is computed by N mxi X i PSNR = 10log10, N ( X i X ' i ) i= 1 where N is the imge size, i is the index of ech pixel, nd levels of the originl nd the processed pixels. 108 X i nd X ' i re the gry The performnce of the proposed scheme under geometric distortions is shown in Tble 6.. Our scheme survives row nd column removl, 10% centered cropping, nd up to 5% shering in x or y direction. Combintion of smll rottions with cropping does not cuse our scheme to fil. But, it is still sensitive to globl imge spect rtio chnges due to the feture loction shifts. It cn lso survive combined geometric nd high qulity JPEG compression ttcks, s shown in Tble 6.. In fct, the correctness of wtermrk detection under geometric distortions strongly depends on the disk loctions. For exmple, if the reference point of n imge disk is locted t the border of n imge, this point might be removed due to cropping ttcks. As result, this

23 disk loction cnnot be correctly identified. Rottion with cropping cn hve to similr effect. The Bboon imge hs deeper nd lrger textured res thn Len nd Peppers. In the cse of Bboon, mny fke reference points (feture points) my show up, nd the true reference points my shift quite significntly fter ttcks. On the other hnd, Peppers hs less texture. Its true feture points my dispper following ttcks. In ddition to the geometric distortions in StirMrk 3.1, we hve pplied locl wrping on the eyes nd mouth of Len, s shown in Fig. 6.13(). The extrcted disks t detector re shown in Fig. 6.13(b). Since locl vritions generlly ffect only few feture points extrcted by the Mexicn Ht wvelet scle interction scheme, the feture points cn still be correctly extrcted for wtermrk detection. The wtermrk cn still be detected quite relibly. 109

24 () (b) (c) () (b) (c) Fig The difference imge between the originl imge nd the wtermrked imge. The mgnitudes in disply re mplified by fctor of 30. () Len, (b) Bboon, nd (c) Peppers. f y f x Fig The wtermrked coefficients re chosen from the shded re. 110

25 () (b) Fig () Locl wrping is pplied to wtermrked imge Len, in the eyes nd mouth re (b) Wtermrk detection result for (). Seven wtermrked disks re correctly detected mong the originl eight. Tble 6.1. Frction of correctly detected wtermrk disks under common signl processing ttcks. Attcks Len Bboon Pepper Wtermrked imge 7/8 10/11 4/4 Medin filter 1/8 6/11 1/4 Medin filter 3 3 1/8 /11 1/4 Shrpening 3 3 4/8 4/11 4/4 Color quntiztion 7/8 4/11 1/4 Gussin filtering 3 3 5/8 8/11 1/4 Additive uniform noise (scle=0.1) 5/8 6/11 4/4 Additive uniform noise (scle=0.15) 4/8 4/11 /4 Additive uniform noise (scle=0.) 1/8 5/11 1/4 JPEG 80 6/8 9/11 3/4 JPEG 70 7/8 11/11 3/4 JPEG 60 6/8 7/11 1/4 JPEG 50 5/8 7/11 3/4 JPEG 40 3/8 5/11 1/4 JPEG 30 /8 4/11 0/4 Medin filter + JPEG90 /8 6/11 0/4 Medin filter JPEG90 1/8 1/11 1/4 Shrpening JPEG90 4/8 /11 4/4 Gussin filtering JPEG90 5/8 8/11 /4 111

26 Tble 6.. Frction of correctly detected wtermrk disks under geometric distortion ttcks. Attcks Len Bboon Pepper Removed 1 row nd 5 columns 3/8 6/11 3/4 Removed 5 rows nd 17 columns 0/8 3/11 1/4 Centered cropping 5% off /8 /11 /4 Centered cropping 10% off /8 /11 /4 Shering-x-1%-y-1% 4/8 5/11 1/4 Shering-x-0%-y-5% /8 3/11 1/4 Shering-x-5%-y-5% 1/8 /11 0/4 Rottion 1+Cropping+Scle 0/8 4/11 /4 Rottion 1+Cropping 3/8 3/11 /4 Rottion +Cropping 0/8 1/11 1/4 Rottion 5+Cropping 0/8 0/11 0/4 Liner geometric trnsform 5/8 4/11 1/4 (1.007,0.01,0.01,1.01) Liner geometric trnsform 4/8 4/11 1/4 (1.010,0.013,0.009,1.011) Liner geometric trnsform 4/8 5/11 0/4 (1.013,0.008,0.011,1.008) Removed 1 rows 5 columns + JPEG70 4/8 6/11 3/4 Removed 5 rows 17 columns + JPEG70 1/8 3/11 1/4 Centered cropping 5% + JPEG70 /8 /11 /4 Centered cropping 10% + JPEG70 3/8 /11 /4 Shering-x-1%-y-1%+JPEG70 /8 4/11 1/4 Shering-x-0%-y-5%+JPEG70 /8 3/11 0/4 Shering-x-5%-y-5%+JPEG70 1/8 0/11 0/4 Rottion 1+Cropping+Scle+JPEG70 0/8 4/11 0/4 Rottion 1+Cropping+JPEG70 4/8 3/11 1/4 Rottion +Cropping+JPEG70 1/8 1/11 1/4 Rottion 5+Cropping+JPEG70 1/8 0/11 0/4 Liner geometric trnsform 4/8 3/11 1/4 (1.007,0.01,0.01,1.01) +JPEG70 Liner geometric trnsform 4/8 5/11 3/4 (1.010,0.013,0.009,1.011) +JPEG70 Liner geometric trnsform (1.013,0.008,0.011,1.008) +JPEG70 3/8 5/11 0/4 11

27 Tble 6.3. PSNR vlues. Attcks Len Bboon Pepper Medin filter Medin filter Shrpening Color quntiztion Gussin filtering Additive uniform noise (scle=0.1) Additive uniform noise (scle=0.15) Additive uniform noise (scle=0.) JPEG JPEG JPEG JPEG JPEG JPEG (noise = difference between the wtermrked imge nd the ttcked imges) 6.7 Summry In this chpter, digitl imge wtermrking scheme ws designed to survive both geometric distortion nd signl processing ttcks. There re three key elements in our scheme: relible imge feture points, imge normliztion, nd DFT domin bits embedding. No reference imges re needed t the detector. Geometric synchroniztion problem between the wtermrk embedding nd detection is overcome by using visully significnt points s reference points. In ddition, the invrince properties of the imge normliztion technique cn gretly reduce the wtermrk serch spce. The simultion results show tht the proposed wtermrking scheme performs well under mild geometric distortion nd common signl processing ttcks. Furthermore, the embedded wtermrk cn resist composite ttcks of high qulity JPEG compression together with geometric distortions/signl processing. 113

28 The performnce of our scheme could be further improved if the feture points were even more robust. Thus, one direction of future reserch cn be the serch of more stble feture points nd/or more relible extrction lgorithms under severe geometric distortions. 114

Chapter 4 Contravariance, Covariance, and Spacetime Diagrams

Chapter 4 Contravariance, Covariance, and Spacetime Diagrams Chpter 4 Contrvrince, Covrince, nd Spcetime Digrms 4. The Components of Vector in Skewed Coordintes We hve seen in Chpter 3; figure 3.9, tht in order to show inertil motion tht is consistent with the Lorentz

More information

NUMERICAL INTEGRATION. The inverse process to differentiation in calculus is integration. Mathematically, integration is represented by.

NUMERICAL INTEGRATION. The inverse process to differentiation in calculus is integration. Mathematically, integration is represented by. NUMERICAL INTEGRATION 1 Introduction The inverse process to differentition in clculus is integrtion. Mthemticlly, integrtion is represented by f(x) dx which stnds for the integrl of the function f(x) with

More information

8 Laplace s Method and Local Limit Theorems

8 Laplace s Method and Local Limit Theorems 8 Lplce s Method nd Locl Limit Theorems 8. Fourier Anlysis in Higher DImensions Most of the theorems of Fourier nlysis tht we hve proved hve nturl generliztions to higher dimensions, nd these cn be proved

More information

Physics 116C Solution of inhomogeneous ordinary differential equations using Green s functions

Physics 116C Solution of inhomogeneous ordinary differential equations using Green s functions Physics 6C Solution of inhomogeneous ordinry differentil equtions using Green s functions Peter Young November 5, 29 Homogeneous Equtions We hve studied, especilly in long HW problem, second order liner

More information

Recitation 3: More Applications of the Derivative

Recitation 3: More Applications of the Derivative Mth 1c TA: Pdric Brtlett Recittion 3: More Applictions of the Derivtive Week 3 Cltech 2012 1 Rndom Question Question 1 A grph consists of the following: A set V of vertices. A set E of edges where ech

More information

State space systems analysis (continued) Stability. A. Definitions A system is said to be Asymptotically Stable (AS) when it satisfies

State space systems analysis (continued) Stability. A. Definitions A system is said to be Asymptotically Stable (AS) when it satisfies Stte spce systems nlysis (continued) Stbility A. Definitions A system is sid to be Asymptoticlly Stble (AS) when it stisfies ut () = 0, t > 0 lim xt () 0. t A system is AS if nd only if the impulse response

More information

5.2 Volumes: Disks and Washers

5.2 Volumes: Disks and Washers 4 pplictions of definite integrls 5. Volumes: Disks nd Wshers In the previous section, we computed volumes of solids for which we could determine the re of cross-section or slice. In this section, we restrict

More information

Numerical integration

Numerical integration 2 Numericl integrtion This is pge i Printer: Opque this 2. Introduction Numericl integrtion is problem tht is prt of mny problems in the economics nd econometrics literture. The orgniztion of this chpter

More information

Tests for the Ratio of Two Poisson Rates

Tests for the Ratio of Two Poisson Rates Chpter 437 Tests for the Rtio of Two Poisson Rtes Introduction The Poisson probbility lw gives the probbility distribution of the number of events occurring in specified intervl of time or spce. The Poisson

More information

CBE 291b - Computation And Optimization For Engineers

CBE 291b - Computation And Optimization For Engineers The University of Western Ontrio Fculty of Engineering Science Deprtment of Chemicl nd Biochemicl Engineering CBE 9b - Computtion And Optimiztion For Engineers Mtlb Project Introduction Prof. A. Jutn Jn

More information

Numerical Integration

Numerical Integration Chpter 5 Numericl Integrtion Numericl integrtion is the study of how the numericl vlue of n integrl cn be found. Methods of function pproximtion discussed in Chpter??, i.e., function pproximtion vi the

More information

Median Filter based wavelet transform for multilevel noise

Median Filter based wavelet transform for multilevel noise Medin Filter bsed wvelet trnsform for multilevel noise H S Shuk Nrendr Kumr *R P Tripthi Deprtment of Computer Science,Deen Dyl Updhy Gorkhpur university,gorkhpur(up) INDIA *Deptrment of Mthemtics,Grphic

More information

Student Activity 3: Single Factor ANOVA

Student Activity 3: Single Factor ANOVA MATH 40 Student Activity 3: Single Fctor ANOVA Some Bsic Concepts In designed experiment, two or more tretments, or combintions of tretments, is pplied to experimentl units The number of tretments, whether

More information

Monte Carlo method in solving numerical integration and differential equation

Monte Carlo method in solving numerical integration and differential equation Monte Crlo method in solving numericl integrtion nd differentil eqution Ye Jin Chemistry Deprtment Duke University yj66@duke.edu Abstrct: Monte Crlo method is commonly used in rel physics problem. The

More information

13.3 CLASSICAL STRAIGHTEDGE AND COMPASS CONSTRUCTIONS

13.3 CLASSICAL STRAIGHTEDGE AND COMPASS CONSTRUCTIONS 33 CLASSICAL STRAIGHTEDGE AND COMPASS CONSTRUCTIONS As simple ppliction of the results we hve obtined on lgebric extensions, nd in prticulr on the multiplictivity of extension degrees, we cn nswer (in

More information

Probability Distributions for Gradient Directions in Uncertain 3D Scalar Fields

Probability Distributions for Gradient Directions in Uncertain 3D Scalar Fields Technicl Report 7.8. Technische Universität München Probbility Distributions for Grdient Directions in Uncertin 3D Sclr Fields Tobis Pfffelmoser, Mihel Mihi, nd Rüdiger Westermnn Computer Grphics nd Visuliztion

More information

W. We shall do so one by one, starting with I 1, and we shall do it greedily, trying

W. We shall do so one by one, starting with I 1, and we shall do it greedily, trying Vitli covers 1 Definition. A Vitli cover of set E R is set V of closed intervls with positive length so tht, for every δ > 0 nd every x E, there is some I V with λ(i ) < δ nd x I. 2 Lemm (Vitli covering)

More information

Math 520 Final Exam Topic Outline Sections 1 3 (Xiao/Dumas/Liaw) Spring 2008

Math 520 Final Exam Topic Outline Sections 1 3 (Xiao/Dumas/Liaw) Spring 2008 Mth 520 Finl Exm Topic Outline Sections 1 3 (Xio/Dums/Liw) Spring 2008 The finl exm will be held on Tuesdy, My 13, 2-5pm in 117 McMilln Wht will be covered The finl exm will cover the mteril from ll of

More information

USA Mathematical Talent Search Round 1 Solutions Year 21 Academic Year

USA Mathematical Talent Search Round 1 Solutions Year 21 Academic Year 1/1/21. Fill in the circles in the picture t right with the digits 1-8, one digit in ech circle with no digit repeted, so tht no two circles tht re connected by line segment contin consecutive digits.

More information

A Blind Image Watermarking Based on Dual Detector *

A Blind Image Watermarking Based on Dual Detector * JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 25, 1723-1736 (2009) A Blind Imge Wtermrking Bsed on Dul Detector * CHIN-PAN HUANG 1, CHI-JEN LIAO 2 AND CHAUR-HEH HSIEH 1 1 Deprtment of Computer nd Communiction

More information

Chapter 5 : Continuous Random Variables

Chapter 5 : Continuous Random Variables STAT/MATH 395 A - PROBABILITY II UW Winter Qurter 216 Néhémy Lim Chpter 5 : Continuous Rndom Vribles Nottions. N {, 1, 2,...}, set of nturl numbers (i.e. ll nonnegtive integers); N {1, 2,...}, set of ll

More information

Properties of Integrals, Indefinite Integrals. Goals: Definition of the Definite Integral Integral Calculations using Antiderivatives

Properties of Integrals, Indefinite Integrals. Goals: Definition of the Definite Integral Integral Calculations using Antiderivatives Block #6: Properties of Integrls, Indefinite Integrls Gols: Definition of the Definite Integrl Integrl Clcultions using Antiderivtives Properties of Integrls The Indefinite Integrl 1 Riemnn Sums - 1 Riemnn

More information

Math 1B, lecture 4: Error bounds for numerical methods

Math 1B, lecture 4: Error bounds for numerical methods Mth B, lecture 4: Error bounds for numericl methods Nthn Pflueger 4 September 0 Introduction The five numericl methods descried in the previous lecture ll operte by the sme principle: they pproximte the

More information

Chapter 14. Matrix Representations of Linear Transformations

Chapter 14. Matrix Representations of Linear Transformations Chpter 4 Mtrix Representtions of Liner Trnsformtions When considering the Het Stte Evolution, we found tht we could describe this process using multipliction by mtrix. This ws nice becuse computers cn

More information

Multiscale Fourier Descriptor for Shape Classification

Multiscale Fourier Descriptor for Shape Classification Multiscle Fourier Descriptor for Shpe Clssifiction Iivri Kunttu, een epistö, Juhni Ruhm 2, nd Ari Vis Tmpere University of Technology Institute of Signl Processing P. O. Box 553, FI-330 Tmpere, Finlnd

More information

13: Diffusion in 2 Energy Groups

13: Diffusion in 2 Energy Groups 3: Diffusion in Energy Groups B. Rouben McMster University Course EP 4D3/6D3 Nucler Rector Anlysis (Rector Physics) 5 Sept.-Dec. 5 September Contents We study the diffusion eqution in two energy groups

More information

A recursive construction of efficiently decodable list-disjunct matrices

A recursive construction of efficiently decodable list-disjunct matrices CSE 709: Compressed Sensing nd Group Testing. Prt I Lecturers: Hung Q. Ngo nd Atri Rudr SUNY t Bufflo, Fll 2011 Lst updte: October 13, 2011 A recursive construction of efficiently decodble list-disjunct

More information

Math 426: Probability Final Exam Practice

Math 426: Probability Final Exam Practice Mth 46: Probbility Finl Exm Prctice. Computtionl problems 4. Let T k (n) denote the number of prtitions of the set {,..., n} into k nonempty subsets, where k n. Argue tht T k (n) kt k (n ) + T k (n ) by

More information

Operations with Polynomials

Operations with Polynomials 38 Chpter P Prerequisites P.4 Opertions with Polynomils Wht you should lern: How to identify the leding coefficients nd degrees of polynomils How to dd nd subtrct polynomils How to multiply polynomils

More information

Lecture 13 - Linking E, ϕ, and ρ

Lecture 13 - Linking E, ϕ, and ρ Lecture 13 - Linking E, ϕ, nd ρ A Puzzle... Inner-Surfce Chrge Density A positive point chrge q is locted off-center inside neutrl conducting sphericl shell. We know from Guss s lw tht the totl chrge on

More information

Continuous Random Variables

Continuous Random Variables STAT/MATH 395 A - PROBABILITY II UW Winter Qurter 217 Néhémy Lim Continuous Rndom Vribles Nottion. The indictor function of set S is rel-vlued function defined by : { 1 if x S 1 S (x) if x S Suppose tht

More information

Non-Linear & Logistic Regression

Non-Linear & Logistic Regression Non-Liner & Logistic Regression If the sttistics re boring, then you've got the wrong numbers. Edwrd R. Tufte (Sttistics Professor, Yle University) Regression Anlyses When do we use these? PART 1: find

More information

KINEMATICS OF RIGID BODIES

KINEMATICS OF RIGID BODIES KINEMTICS OF RIGID ODIES Introduction In rigid body kinemtics, e use the reltionships governing the displcement, velocity nd ccelertion, but must lso ccount for the rottionl motion of the body. Description

More information

Summary: Method of Separation of Variables

Summary: Method of Separation of Variables Physics 246 Electricity nd Mgnetism I, Fll 26, Lecture 22 1 Summry: Method of Seprtion of Vribles 1. Seprtion of Vribles in Crtesin Coordintes 2. Fourier Series Suggested Reding: Griffiths: Chpter 3, Section

More information

2008 Mathematical Methods (CAS) GA 3: Examination 2

2008 Mathematical Methods (CAS) GA 3: Examination 2 Mthemticl Methods (CAS) GA : Exmintion GENERAL COMMENTS There were 406 students who st the Mthemticl Methods (CAS) exmintion in. Mrks rnged from to 79 out of possible score of 80. Student responses showed

More information

Acceptance Sampling by Attributes

Acceptance Sampling by Attributes Introduction Acceptnce Smpling by Attributes Acceptnce smpling is concerned with inspection nd decision mking regrding products. Three spects of smpling re importnt: o Involves rndom smpling of n entire

More information

SUMMER KNOWHOW STUDY AND LEARNING CENTRE

SUMMER KNOWHOW STUDY AND LEARNING CENTRE SUMMER KNOWHOW STUDY AND LEARNING CENTRE Indices & Logrithms 2 Contents Indices.2 Frctionl Indices.4 Logrithms 6 Exponentil equtions. Simplifying Surds 13 Opertions on Surds..16 Scientific Nottion..18

More information

Matching patterns of line segments by eigenvector decomposition

Matching patterns of line segments by eigenvector decomposition Title Mtching ptterns of line segments y eigenvector decomposition Author(s) Chn, BHB; Hung, YS Cittion The 5th IEEE Southwest Symposium on Imge Anlysis nd Interprettion Proceedings, Snte Fe, NM., 7-9

More information

Vyacheslav Telnin. Search for New Numbers.

Vyacheslav Telnin. Search for New Numbers. Vycheslv Telnin Serch for New Numbers. 1 CHAPTER I 2 I.1 Introduction. In 1984, in the first issue for tht yer of the Science nd Life mgzine, I red the rticle "Non-Stndrd Anlysis" by V. Uspensky, in which

More information

A Signal-Level Fusion Model for Image-Based Change Detection in DARPA's Dynamic Database System

A Signal-Level Fusion Model for Image-Based Change Detection in DARPA's Dynamic Database System SPIE Aerosense 001 Conference on Signl Processing, Sensor Fusion, nd Trget Recognition X, April 16-0, Orlndo FL. (Minor errors in published version corrected.) A Signl-Level Fusion Model for Imge-Bsed

More information

Exam 2, Mathematics 4701, Section ETY6 6:05 pm 7:40 pm, March 31, 2016, IH-1105 Instructor: Attila Máté 1

Exam 2, Mathematics 4701, Section ETY6 6:05 pm 7:40 pm, March 31, 2016, IH-1105 Instructor: Attila Máté 1 Exm, Mthemtics 471, Section ETY6 6:5 pm 7:4 pm, Mrch 1, 16, IH-115 Instructor: Attil Máté 1 17 copies 1. ) Stte the usul sufficient condition for the fixed-point itertion to converge when solving the eqution

More information

Recitation 3: Applications of the Derivative. 1 Higher-Order Derivatives and their Applications

Recitation 3: Applications of the Derivative. 1 Higher-Order Derivatives and their Applications Mth 1c TA: Pdric Brtlett Recittion 3: Applictions of the Derivtive Week 3 Cltech 013 1 Higher-Order Derivtives nd their Applictions Another thing we could wnt to do with the derivtive, motivted by wht

More information

MATH34032: Green s Functions, Integral Equations and the Calculus of Variations 1

MATH34032: Green s Functions, Integral Equations and the Calculus of Variations 1 MATH34032: Green s Functions, Integrl Equtions nd the Clculus of Vritions 1 Section 1 Function spces nd opertors Here we gives some brief detils nd definitions, prticulrly relting to opertors. For further

More information

Energy Bands Energy Bands and Band Gap. Phys463.nb Phenomenon

Energy Bands Energy Bands and Band Gap. Phys463.nb Phenomenon Phys463.nb 49 7 Energy Bnds Ref: textbook, Chpter 7 Q: Why re there insultors nd conductors? Q: Wht will hppen when n electron moves in crystl? In the previous chpter, we discussed free electron gses,

More information

7.1 Integral as Net Change and 7.2 Areas in the Plane Calculus

7.1 Integral as Net Change and 7.2 Areas in the Plane Calculus 7.1 Integrl s Net Chnge nd 7. Ares in the Plne Clculus 7.1 INTEGRAL AS NET CHANGE Notecrds from 7.1: Displcement vs Totl Distnce, Integrl s Net Chnge We hve lredy seen how the position of n oject cn e

More information

Math 113 Exam 2 Practice

Math 113 Exam 2 Practice Mth 3 Exm Prctice Februry 8, 03 Exm will cover 7.4, 7.5, 7.7, 7.8, 8.-3 nd 8.5. Plese note tht integrtion skills lerned in erlier sections will still be needed for the mteril in 7.5, 7.8 nd chpter 8. This

More information

2. VECTORS AND MATRICES IN 3 DIMENSIONS

2. VECTORS AND MATRICES IN 3 DIMENSIONS 2 VECTORS AND MATRICES IN 3 DIMENSIONS 21 Extending the Theory of 2-dimensionl Vectors x A point in 3-dimensionl spce cn e represented y column vector of the form y z z-xis y-xis z x y x-xis Most of the

More information

A027 Uncertainties in Local Anisotropy Estimation from Multi-offset VSP Data

A027 Uncertainties in Local Anisotropy Estimation from Multi-offset VSP Data A07 Uncertinties in Locl Anisotropy Estimtion from Multi-offset VSP Dt M. Asghrzdeh* (Curtin University), A. Bon (Curtin University), R. Pevzner (Curtin University), M. Urosevic (Curtin University) & B.

More information

The steps of the hypothesis test

The steps of the hypothesis test ttisticl Methods I (EXT 7005) Pge 78 Mosquito species Time of dy A B C Mid morning 0.0088 5.4900 5.5000 Mid Afternoon.3400 0.0300 0.8700 Dusk 0.600 5.400 3.000 The Chi squre test sttistic is the sum of

More information

ADVANCEMENT OF THE CLOSELY COUPLED PROBES POTENTIAL DROP TECHNIQUE FOR NDE OF SURFACE CRACKS

ADVANCEMENT OF THE CLOSELY COUPLED PROBES POTENTIAL DROP TECHNIQUE FOR NDE OF SURFACE CRACKS ADVANCEMENT OF THE CLOSELY COUPLED PROBES POTENTIAL DROP TECHNIQUE FOR NDE OF SURFACE CRACKS F. Tkeo 1 nd M. Sk 1 Hchinohe Ntionl College of Technology, Hchinohe, Jpn; Tohoku University, Sendi, Jpn Abstrct:

More information

Chapter 0. What is the Lebesgue integral about?

Chapter 0. What is the Lebesgue integral about? Chpter 0. Wht is the Lebesgue integrl bout? The pln is to hve tutoril sheet ech week, most often on Fridy, (to be done during the clss) where you will try to get used to the ides introduced in the previous

More information

APPROXIMATE INTEGRATION

APPROXIMATE INTEGRATION APPROXIMATE INTEGRATION. Introduction We hve seen tht there re functions whose nti-derivtives cnnot be expressed in closed form. For these resons ny definite integrl involving these integrnds cnnot be

More information

The Regulated and Riemann Integrals

The Regulated and Riemann Integrals Chpter 1 The Regulted nd Riemnn Integrls 1.1 Introduction We will consider severl different pproches to defining the definite integrl f(x) dx of function f(x). These definitions will ll ssign the sme vlue

More information

Summary Information and Formulae MTH109 College Algebra

Summary Information and Formulae MTH109 College Algebra Generl Formuls Summry Informtion nd Formule MTH109 College Algebr Temperture: F = 9 5 C + 32 nd C = 5 ( 9 F 32 ) F = degrees Fhrenheit C = degrees Celsius Simple Interest: I = Pr t I = Interest erned (chrged)

More information

Goals: Determine how to calculate the area described by a function. Define the definite integral. Explore the relationship between the definite

Goals: Determine how to calculate the area described by a function. Define the definite integral. Explore the relationship between the definite Unit #8 : The Integrl Gols: Determine how to clculte the re described by function. Define the definite integrl. Eplore the reltionship between the definite integrl nd re. Eplore wys to estimte the definite

More information

DEFINITION OF ASSOCIATIVE OR DIRECT PRODUCT AND ROTATION OF VECTORS

DEFINITION OF ASSOCIATIVE OR DIRECT PRODUCT AND ROTATION OF VECTORS 3 DEFINITION OF ASSOCIATIVE OR DIRECT PRODUCT AND ROTATION OF VECTORS This chpter summrizes few properties of Cli ord Algebr nd describe its usefulness in e ecting vector rottions. 3.1 De nition of Associtive

More information

SOLUTIONS FOR ADMISSIONS TEST IN MATHEMATICS, COMPUTER SCIENCE AND JOINT SCHOOLS WEDNESDAY 5 NOVEMBER 2014

SOLUTIONS FOR ADMISSIONS TEST IN MATHEMATICS, COMPUTER SCIENCE AND JOINT SCHOOLS WEDNESDAY 5 NOVEMBER 2014 SOLUTIONS FOR ADMISSIONS TEST IN MATHEMATICS, COMPUTER SCIENCE AND JOINT SCHOOLS WEDNESDAY 5 NOVEMBER 014 Mrk Scheme: Ech prt of Question 1 is worth four mrks which re wrded solely for the correct nswer.

More information

Duality # Second iteration for HW problem. Recall our LP example problem we have been working on, in equality form, is given below.

Duality # Second iteration for HW problem. Recall our LP example problem we have been working on, in equality form, is given below. Dulity #. Second itertion for HW problem Recll our LP emple problem we hve been working on, in equlity form, is given below.,,,, 8 m F which, when written in slightly different form, is 8 F Recll tht we

More information

HW3, Math 307. CSUF. Spring 2007.

HW3, Math 307. CSUF. Spring 2007. HW, Mth 7. CSUF. Spring 7. Nsser M. Abbsi Spring 7 Compiled on November 5, 8 t 8:8m public Contents Section.6, problem Section.6, problem Section.6, problem 5 Section.6, problem 7 6 5 Section.6, problem

More information

Genetic Programming. Outline. Evolutionary Strategies. Evolutionary strategies Genetic programming Summary

Genetic Programming. Outline. Evolutionary Strategies. Evolutionary strategies Genetic programming Summary Outline Genetic Progrmming Evolutionry strtegies Genetic progrmming Summry Bsed on the mteril provided y Professor Michel Negnevitsky Evolutionry Strtegies An pproch simulting nturl evolution ws proposed

More information

3.4 Numerical integration

3.4 Numerical integration 3.4. Numericl integrtion 63 3.4 Numericl integrtion In mny economic pplictions it is necessry to compute the definite integrl of relvlued function f with respect to "weight" function w over n intervl [,

More information

UNIT 1 FUNCTIONS AND THEIR INVERSES Lesson 1.4: Logarithmic Functions as Inverses Instruction

UNIT 1 FUNCTIONS AND THEIR INVERSES Lesson 1.4: Logarithmic Functions as Inverses Instruction Lesson : Logrithmic Functions s Inverses Prerequisite Skills This lesson requires the use of the following skills: determining the dependent nd independent vribles in n exponentil function bsed on dt from

More information

New data structures to reduce data size and search time

New data structures to reduce data size and search time New dt structures to reduce dt size nd serch time Tsuneo Kuwbr Deprtment of Informtion Sciences, Fculty of Science, Kngw University, Hirtsuk-shi, Jpn FIT2018 1D-1, No2, pp1-4 Copyright (c)2018 by The Institute

More information

A New Receiver for Chaotic Digital Transmissions: The Symbolic Matching Approach

A New Receiver for Chaotic Digital Transmissions: The Symbolic Matching Approach A New Receiver for Chotic Digitl Trnsmissions: The Symbolic Mtching Approch Gilles BUREL nd Stéphne AZOU LEST, Université de Bretgne Occidentle CS 93837, 29238 BREST cede 3, Frnce Abstrct : Chotic digitl

More information

MORE FUNCTION GRAPHING; OPTIMIZATION. (Last edited October 28, 2013 at 11:09pm.)

MORE FUNCTION GRAPHING; OPTIMIZATION. (Last edited October 28, 2013 at 11:09pm.) MORE FUNCTION GRAPHING; OPTIMIZATION FRI, OCT 25, 203 (Lst edited October 28, 203 t :09pm.) Exercise. Let n be n rbitrry positive integer. Give n exmple of function with exctly n verticl symptotes. Give

More information

Name Solutions to Test 3 November 8, 2017

Name Solutions to Test 3 November 8, 2017 Nme Solutions to Test 3 November 8, 07 This test consists of three prts. Plese note tht in prts II nd III, you cn skip one question of those offered. Some possibly useful formuls cn be found below. Brrier

More information

Preparation for A Level Wadebridge School

Preparation for A Level Wadebridge School Preprtion for A Level Mths @ Wdebridge School Bridging the gp between GCSE nd A Level Nme: CONTENTS Chpter Removing brckets pge Chpter Liner equtions Chpter Simultneous equtions 6 Chpter Fctorising 7 Chpter

More information

Scientific notation is a way of expressing really big numbers or really small numbers.

Scientific notation is a way of expressing really big numbers or really small numbers. Scientific Nottion (Stndrd form) Scientific nottion is wy of expressing relly big numbers or relly smll numbers. It is most often used in scientific clcultions where the nlysis must be very precise. Scientific

More information

fractions Let s Learn to

fractions Let s Learn to 5 simple lgebric frctions corne lens pupil retin Norml vision light focused on the retin concve lens Shortsightedness (myopi) light focused in front of the retin Corrected myopi light focused on the retin

More information

5.7 Improper Integrals

5.7 Improper Integrals 458 pplictions of definite integrls 5.7 Improper Integrls In Section 5.4, we computed the work required to lift pylod of mss m from the surfce of moon of mss nd rdius R to height H bove the surfce of the

More information

Theoretical foundations of Gaussian quadrature

Theoretical foundations of Gaussian quadrature Theoreticl foundtions of Gussin qudrture 1 Inner product vector spce Definition 1. A vector spce (or liner spce) is set V = {u, v, w,...} in which the following two opertions re defined: (A) Addition of

More information

Review of Calculus, cont d

Review of Calculus, cont d Jim Lmbers MAT 460 Fll Semester 2009-10 Lecture 3 Notes These notes correspond to Section 1.1 in the text. Review of Clculus, cont d Riemnn Sums nd the Definite Integrl There re mny cses in which some

More information

p-adic Egyptian Fractions

p-adic Egyptian Fractions p-adic Egyptin Frctions Contents 1 Introduction 1 2 Trditionl Egyptin Frctions nd Greedy Algorithm 2 3 Set-up 3 4 p-greedy Algorithm 5 5 p-egyptin Trditionl 10 6 Conclusion 1 Introduction An Egyptin frction

More information

Indefinite Integral. Chapter Integration - reverse of differentiation

Indefinite Integral. Chapter Integration - reverse of differentiation Chpter Indefinite Integrl Most of the mthemticl opertions hve inverse opertions. The inverse opertion of differentition is clled integrtion. For exmple, describing process t the given moment knowing the

More information

Lecture 3 Gaussian Probability Distribution

Lecture 3 Gaussian Probability Distribution Introduction Lecture 3 Gussin Probbility Distribution Gussin probbility distribution is perhps the most used distribution in ll of science. lso clled bell shped curve or norml distribution Unlike the binomil

More information

13.4 Work done by Constant Forces

13.4 Work done by Constant Forces 13.4 Work done by Constnt Forces We will begin our discussion of the concept of work by nlyzing the motion of n object in one dimension cted on by constnt forces. Let s consider the following exmple: push

More information

Chapter 3 Polynomials

Chapter 3 Polynomials Dr M DRAIEF As described in the introduction of Chpter 1, pplictions of solving liner equtions rise in number of different settings In prticulr, we will in this chpter focus on the problem of modelling

More information

APPLICATIONS OF THE DEFINITE INTEGRAL

APPLICATIONS OF THE DEFINITE INTEGRAL APPLICATIONS OF THE DEFINITE INTEGRAL. Volume: Slicing, disks nd wshers.. Volumes by Slicing. Suppose solid object hs boundries extending from x =, to x = b, nd tht its cross-section in plne pssing through

More information

Chapter 5 Bending Moments and Shear Force Diagrams for Beams

Chapter 5 Bending Moments and Shear Force Diagrams for Beams Chpter 5 ending Moments nd Sher Force Digrms for ems n ddition to illy loded brs/rods (e.g. truss) nd torsionl shfts, the structurl members my eperience some lods perpendiculr to the is of the bem nd will

More information

A REVIEW OF CALCULUS CONCEPTS FOR JDEP 384H. Thomas Shores Department of Mathematics University of Nebraska Spring 2007

A REVIEW OF CALCULUS CONCEPTS FOR JDEP 384H. Thomas Shores Department of Mathematics University of Nebraska Spring 2007 A REVIEW OF CALCULUS CONCEPTS FOR JDEP 384H Thoms Shores Deprtment of Mthemtics University of Nebrsk Spring 2007 Contents Rtes of Chnge nd Derivtives 1 Dierentils 4 Are nd Integrls 5 Multivrite Clculus

More information

In-Class Problems 2 and 3: Projectile Motion Solutions. In-Class Problem 2: Throwing a Stone Down a Hill

In-Class Problems 2 and 3: Projectile Motion Solutions. In-Class Problem 2: Throwing a Stone Down a Hill MASSACHUSETTS INSTITUTE OF TECHNOLOGY Deprtment of Physics Physics 8T Fll Term 4 In-Clss Problems nd 3: Projectile Motion Solutions We would like ech group to pply the problem solving strtegy with the

More information

set is not closed under matrix [ multiplication, ] and does not form a group.

set is not closed under matrix [ multiplication, ] and does not form a group. Prolem 2.3: Which of the following collections of 2 2 mtrices with rel entries form groups under [ mtrix ] multipliction? i) Those of the form for which c d 2 Answer: The set of such mtrices is not closed

More information

1 Online Learning and Regret Minimization

1 Online Learning and Regret Minimization 2.997 Decision-Mking in Lrge-Scle Systems My 10 MIT, Spring 2004 Hndout #29 Lecture Note 24 1 Online Lerning nd Regret Minimiztion In this lecture, we consider the problem of sequentil decision mking in

More information

The use of a so called graphing calculator or programmable calculator is not permitted. Simple scientific calculators are allowed.

The use of a so called graphing calculator or programmable calculator is not permitted. Simple scientific calculators are allowed. ERASMUS UNIVERSITY ROTTERDAM Informtion concerning the Entrnce exmintion Mthemtics level 1 for Interntionl Bchelor in Communiction nd Medi Generl informtion Avilble time: 2 hours 30 minutes. The exmintion

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Physics Department Statistical Physics I Spring Term Solutions to Problem Set #1

MASSACHUSETTS INSTITUTE OF TECHNOLOGY Physics Department Statistical Physics I Spring Term Solutions to Problem Set #1 MASSACHUSETTS INSTITUTE OF TECHNOLOGY Physics Deprtment 8.044 Sttisticl Physics I Spring Term 03 Problem : Doping Semiconductor Solutions to Problem Set # ) Mentlly integrte the function p(x) given in

More information

An instructive toy model: two paradoxes

An instructive toy model: two paradoxes Tel Aviv University, 2006 Gussin rndom vectors 27 3 Level crossings... the fmous ice formul, undoubtedly one of the most importnt results in the ppliction of smooth stochstic processes..j. Adler nd J.E.

More information

Physics 9 Fall 2011 Homework 2 - Solutions Friday September 2, 2011

Physics 9 Fall 2011 Homework 2 - Solutions Friday September 2, 2011 Physics 9 Fll 0 Homework - s Fridy September, 0 Mke sure your nme is on your homework, nd plese box your finl nswer. Becuse we will be giving prtil credit, be sure to ttempt ll the problems, even if you

More information

Jackson 2.7 Homework Problem Solution Dr. Christopher S. Baird University of Massachusetts Lowell

Jackson 2.7 Homework Problem Solution Dr. Christopher S. Baird University of Massachusetts Lowell Jckson.7 Homework Problem Solution Dr. Christopher S. Bird University of Msschusetts Lowell PROBLEM: Consider potentil problem in the hlf-spce defined by, with Dirichlet boundry conditions on the plne

More information

Best Approximation. Chapter The General Case

Best Approximation. Chapter The General Case Chpter 4 Best Approximtion 4.1 The Generl Cse In the previous chpter, we hve seen how n interpolting polynomil cn be used s n pproximtion to given function. We now wnt to find the best pproximtion to given

More information

Numerical Analysis: Trapezoidal and Simpson s Rule

Numerical Analysis: Trapezoidal and Simpson s Rule nd Simpson s Mthemticl question we re interested in numericlly nswering How to we evlute I = f (x) dx? Clculus tells us tht if F(x) is the ntiderivtive of function f (x) on the intervl [, b], then I =

More information

Intensity transformations

Intensity transformations Intensity trnsformtions Stefno Ferrri Università degli Studi di Milno stefno.ferrri@unimi.it Methods for Imge Processing cdemic yer 2017 2018 Sptil domin The sptil domin of n imge is the plne tht contins

More information

The Properties of Stars

The Properties of Stars 10/11/010 The Properties of Strs sses Using Newton s Lw of Grvity to Determine the ss of Celestil ody ny two prticles in the universe ttrct ech other with force tht is directly proportionl to the product

More information

New Expansion and Infinite Series

New Expansion and Infinite Series Interntionl Mthemticl Forum, Vol. 9, 204, no. 22, 06-073 HIKARI Ltd, www.m-hikri.com http://dx.doi.org/0.2988/imf.204.4502 New Expnsion nd Infinite Series Diyun Zhng College of Computer Nnjing University

More information

Physics 202H - Introductory Quantum Physics I Homework #08 - Solutions Fall 2004 Due 5:01 PM, Monday 2004/11/15

Physics 202H - Introductory Quantum Physics I Homework #08 - Solutions Fall 2004 Due 5:01 PM, Monday 2004/11/15 Physics H - Introductory Quntum Physics I Homework #8 - Solutions Fll 4 Due 5:1 PM, Mondy 4/11/15 [55 points totl] Journl questions. Briefly shre your thoughts on the following questions: Of the mteril

More information

Quantum Mechanics Qualifying Exam - August 2016 Notes and Instructions

Quantum Mechanics Qualifying Exam - August 2016 Notes and Instructions Quntum Mechnics Qulifying Exm - August 016 Notes nd Instructions There re 6 problems. Attempt them ll s prtil credit will be given. Write on only one side of the pper for your solutions. Write your lis

More information

A. Limits - L Hopital s Rule ( ) How to find it: Try and find limits by traditional methods (plugging in). If you get 0 0 or!!, apply C.! 1 6 C.

A. Limits - L Hopital s Rule ( ) How to find it: Try and find limits by traditional methods (plugging in). If you get 0 0 or!!, apply C.! 1 6 C. A. Limits - L Hopitl s Rule Wht you re finding: L Hopitl s Rule is used to find limits of the form f ( x) lim where lim f x x! c g x ( ) = or lim f ( x) = limg( x) = ". ( ) x! c limg( x) = 0 x! c x! c

More information

CMDA 4604: Intermediate Topics in Mathematical Modeling Lecture 19: Interpolation and Quadrature

CMDA 4604: Intermediate Topics in Mathematical Modeling Lecture 19: Interpolation and Quadrature CMDA 4604: Intermedite Topics in Mthemticl Modeling Lecture 19: Interpoltion nd Qudrture In this lecture we mke brief diversion into the res of interpoltion nd qudrture. Given function f C[, b], we sy

More information

Bernoulli Numbers Jeff Morton

Bernoulli Numbers Jeff Morton Bernoulli Numbers Jeff Morton. We re interested in the opertor e t k d k t k, which is to sy k tk. Applying this to some function f E to get e t f d k k tk d k f f + d k k tk dk f, we note tht since f

More information

Natural examples of rings are the ring of integers, a ring of polynomials in one variable, the ring

Natural examples of rings are the ring of integers, a ring of polynomials in one variable, the ring More generlly, we define ring to be non-empty set R hving two binry opertions (we ll think of these s ddition nd multipliction) which is n Abelin group under + (we ll denote the dditive identity by 0),

More information

7 - Continuous random variables

7 - Continuous random variables 7-1 Continuous rndom vribles S. Lll, Stnford 2011.01.25.01 7 - Continuous rndom vribles Continuous rndom vribles The cumultive distribution function The uniform rndom vrible Gussin rndom vribles The Gussin

More information