Shifting and adding images

Size: px
Start display at page:

Download "Shifting and adding images"

Transcription

1 Preparaton for the examnaton 5LIN0 Vdeo processng Gerard de Haan Avalable materal: Lectures x h durng 7 weeks Book (Dgtal Vdeo Post Processng) Verson Dec. 04 Except Chapter 6 Questons n every chapter, to exercse for exam Avalable from Marja de Mol, Flux 4.3 (Eu.50) Demo software (VdProc) Downloadable from bookssoftware (password ) Sldes: Your notes (hardly necessary when you learn from the book and do the exercses) You may brng the book to the exam! VdProc 3 Schedule lectures 5LIN0 4 Week Week Week 3 Week 4 Bascs (Ch, 3) Vdeo dsplays (Ch 9) Flterng (Ch 4) Pcture-Rate Converson (Ch 7) Week 5 Week 6 Week 7 Week 8 De-nterlacng (Ch 8) Moton Estmaton (Ch 0) Object Detecton (Ch ) X Shftng and addng mages 5 What f we average an mage wth a shfted verson? 6 What f we subtract an mage wth a shfted verson? = Orgnal mage pel shfted mage average mage A frequency s suppressed f we shft over 80 degree. Therefore, the frequency has a complete cycle of 4 pxels In other words, suppresson occurs at a quarter of the samplng frequency pel + = Averagng wth a Orgnal mage pel dagonally shfted mage Subtracton suppresses DC (coarse spatal detal A frequency wth a complete cycle of 4 pxels doubles n ampltude In other words, enhancement occurs at a quarter of the samplng frequency Subtracton wth a pel dagonally shfted mage pel - =

2 7 How do we shft an mage? (scannng assumed) 8 f v (c/ph) Delay f t (c/pp) Image scanned from top-left to bottom rght pxel horzontal shft lne vertcal shft pcture same poston, temporal shft Lnear flterng of dscrete mages f h (c/pw) 9 What s a lnear flter? A lnear flter s a system wth an nput and an output that: Contans s, multplers and adders = weghted sum of ed copes of nput/output sgnal 0 Example lnear flter x[k-3] x[k-] x[k-] x[k] w 3 w w w ww3 ww4 0 y [ k] w x[ k n] n sum n y[k] Meanng of : Smple notaton From horzontal to vertcal flterng Defnton: (,, ) flter weghted sum of horzontally neghbourng pxels Defnton: flter weghted sum of vertcally neghbourng pxels Vertcally shftng requres lne-s Orders of magntude more expensve than a pxel- Temporal flterng requres pcture-s Agan orders of magntude more expensve Change pxel-s lne-s Lne- can be mplemented wth a lne-memory ¼ 0 Lne Lne Lne ½ sum ¼

3 out Vdeo course: Bascs: Flterng 3 3 What f you flter an mage both n H- and V-dmenson? Flter support w w w3 w4 w5 w6 4 What f we re-use output samples? Recursve flterng ww 3 ww ww3 ww4 0 w7 w8 w9 n sum wr4 wr3 wr w 3r w r w r wr w 0r mage mage y [ k] w. x[ k n] w. y[ k m] n n m mr 5 6 Some mpulse responses (,,) Cascadng flters (,) Sngle whte pxel nput (,0,) 0 7 The cascadng of two (, ) flters A 8 Why a trangle? (convoluton!) tme A ½ ½ A ½ ¼ ¼ tme tme Example cascade: Cascade of (,) and (, ):

4 out Vdeo course: Bascs: Flterng 4 9 Smple calculaton of mpulse response of cascade 0 Smple calculaton of mpulse response of cascade Cascade of and 0 Cascade of (,) and (, ): Cascade of and : ¼ f s Cascade of and (cascade of and and and ): ¼ f s ½ f s ½ f s Smple calculaton of mpulse response of cascade Cascade of (horzontal) and a(vertcal) Cascade of (horzontal) and a (vertcal) 4 Qualtatve analyss of lnear flters 3 What s mage flterng? (D convoluton) 4 Effect of the box flter on a natural mage Flter support w w w3 w4 w5 w6 n Assgn weghted sum to output mage Orgnal (nput) Box-flter output w7 w8 w9 mage mage Move the flter-support (=coeffcent-matrx,.e. mpulse response) over each pxel n the mage, multply the entres by the pxels, sum together, and assgn result to central pxel of output mage E.g. 3x3 box -flter 9 Effect s averagng

5 5 5 Box Flter Smoothng due to averagng neghborng pxels Frequency doman: leaves LF unaltered and attenuates HF So, box flter s a low-pass flter A 0 Spatal: Box pos 9 Frequency doman: snc-functon H 0 f 6 Not only low-pass flters the frequency doman Transfer of flter Spectrum nput mage Spectrum of output mage A H A = Low-pass 0 f 0 f 0 f A H A = Hgh-pass 0 f 0 f 0 f A H A = Band-pass 0 f 0 f 0 f 7 Hgh-pass flters 8 Effect of the hgh-pass flter on a natural mage Turn a LP-flter nto a HP-flter Subtract smoothed mage from orgnal = remove LF Result: only HF Orgnal (nput) Hgh-pass flter output +8 Impulse-response found wth matrx subtracton: Example: LPF = 3x3 Box: Edge Enhancement 30 Effect of the edge-enhance, or peakng, flter HP-flters: Hgh values at edges, Low values n constant regons Addng hgh frequences back to the mage enhances edges Possble approach: Image = Image + [Image smooth(image)] Low-pass Hgh-pass [- - ]- - - x/

6 6 3 Edge enhancement, peakng, or unsharp maskng 3 Negatve flter coeffcents may lead to problems! Negatve coeffcents can lead to negatve mage values Most mage formats don t support ths No negatve lght. Also, boostng HF may gve values above peak-whte Solutons: Clppng: Chop off values below mn or above max Offset: Add a constant to move the mn value to 0 Re-scale: Rescale the mage values to fll the range (0,max) Non-lnearty Spols black Spols contrast 33 What about the mage boundares? At mage poston (0,0), you need e.g. pxel (-,-), whch doesn t exst Opton : Make output mage smaller Opton : Replcate edge pxels Opton 3: Reflect mage about edge 5LIN0 Vdeo processng Gerard de Haan Purpose of flters The purpose of flters Removal of spectral components E.g. for alas preventon, or remove nterference Enhancement of spectral components Edge/feature enhancement Removal of nose Balance between suppresson of nose and relevant mage components Interpolaton Image scalng, geometrcal deformatons (warpng)

7 7 37 Nose removal 38 Alas preventon H-LPF V-LPF The smple way. Nearest pxel Flterng and mage re-szng (D case) Orgnal pxels: Requred lower densty grd: Requred hgher densty grd: Repeated pxels! Dropped pxels! 4 Spatal scalng 4 Bt better: tap lnear nterpolaton Scaled usng nearest pxel nterpolaton Trangular shape of mpulse response (weght lnearly deceases wth dstance) Length mpulse response s 3 (always pxels n support), Suppose we need a sample at x=4.75: Result s: 0.5F(4) F(5) Scaled wth proper flters x

8 8 43 Longer flter? 44 What are the effects of the flter length? Trangular shape of mpulse response (weght lnearly decreases wth dstance) Now flter support sze s 4 (max 3 pxels n support) Suppose we need a sample at x=4.75: Length = Assume the flter sze s 3, and we need a sample at x=4.75 We need the mage samples, F(x), from x=4, x=5 and x=6 x We need the flter value, H(s), at s=-0.75, s=0.5 and s=.5 We compute: F(4)*H(-0.75)+F(5)*H(0.5)+F(6)*H(.5) Of course: The samplng theorem These were example attempts Does theory provde any requrements for the nterpolaton flter? 0 f s f s We have a contnuous sgnal We sample t to obtan a dscrete representaton Samplng theorem: we can reconstruct the contnuous sgnal from ts dscrete representaton, provded t contaned no frequences above half the samplng frequency We can re-sample the contnuous sgnal at any rate! T s tme 47 The mpulse response of such flter looks lke ths 48 Re-samplng It s a sync-functon The flter has to be approxmated as t has an nfnte length Makng an mage larger s lke re-samplng the orgnal sgnal at a hgher densty Approxmaton means removng smaller coeffcents Example for BW = f s / [ ] [ ] [ ]... Orgnal More samples Orgnal spacng = bgger Reducng an mage n sze s lke re-samplng at lower densty In theory, steps: Reconstructon of the contnuous sgnal Followed by re-samplng wth the new samplng frequency In practce: sample rate converson n the dscrete doman

9 9 49 The general prncple of sample rate converson 50 Spectra n samplng rate converson (down-samplng) Intermedate nterpolator decmator grd grd grd Samplng-rate converter As a consequence of the tme-dscrete nature of the processng, the nput and output samplng frequency have a ratonal relaton. results from an nteger up-samplng and an nteger sub-samplng 0 frequency f s f s 0 frequency f s Alas! 0 frequency f s Repeat on output rate wll overlap wth base-band unless the nterpolatng LPF suppresses part of spectrum 5 Sgnals n sample rate converson (down-samplng 3/4) 5 Spectra n sample rate converson (up-samplng) nput 0 frequency f s f s 3f s Interp. Intermed. Repeats at nput rate must be suppressed wthout suppressng base-band 0 frequency f s output to prevent alas or blur at output rate 0 frequency f s3 53 Sgnals n sample rate converson (up-samplng, 3/) 54 Requrements for the nterpolaton (scalng) flter nput F(nT) sgnal sgnal Interpolaton flter Interp. output 3 4 nt nt Up-samplng: LPF passes base-band sgnal and suppresses the nput repeat spectra where possble at ntermedate rate Down-samplng: LPF suppresses part of the base-band and repeat spectra to avod foldng nto the new base-band output spectrum

10 0 55 Poly-phase flterng: the flter s desgned at the ntermedate frequency, at the output frequency a varyng set of coeffcents s used 56 Poly-phase flterng In practce, we do not use an ntermedate samplng rate. Orgnal pxels: C-6 C-5 C-4 C-3 C- C- C0 C C C3 C4 C5 C6 [ ] a X b X c X d X e X f X C-6 C-5 C-4 C-3 C- C- C0 C C C3 C4 C5 C6 Requred pxel on other densty grd: 4 n 5 The flter coeffcents, a b c d e f.., depend on the poston of the requred pxel n 57 Applcaton: Aspect rato converson 58 4:3 mage on a wde-screen (6:9) pcture tube Lnear scalng optons: Accept sde-panels Accept geometrcal dstorton 59 4:3 mage on a wde-screen (6:9) pcture tube 60 Wde-screen mage on a 4:3 pcture tube Non-lnear scalng optons: Lnear scalng optons: Accept geometrcal dstorton Panorama vew Accept letter-box Geometrcal dstorton Partal mage loss

11 magnfcaton Vdeo course: Bascs: Flterng 6 Wde-screen mage on a 4:3 pcture tube 6 Optons n wde-screen converson Non-lnear scalng optons: Dstortons concentrated evenly dstrbuted n sde panels Usually magnfcaton between 0.5 and.0.33 Constant horzontal stretch Panorama.0 Amaronap vew 0.75 Constant horzontal compress Amaronap Horzontal poston 64 5LIN0 Vdeo processng Recaptulaton Gerard de Haan 65 output, adders, multplers and s w w w3 w4 sum wr4 wr3 wr wr y [ k] w. x[ k n] w. y[ k m] n n m mr 66 Purpose of flters Removal of spectral components E.g. for alas preventon, or removal of nterference sgnal Enhancement of spectral components Edge/feature enhancement Removal of nose Balance between nose suppresson and suppresson of relevant mage components Interpolaton Image up- and down-scalng, geometrcal deformatons

12 67 The general prncple of sample rate converson 68 Poly-phase flterng In practce, we do not use an ntermedate samplng rate. grd nterpolator Intermedate grd decmator Samplng-rate converter grd Orgnal pxels: a X b X c X d X e X f X As a consequence of the tme-dscrete nature of the processng, the nput and output samplng frequency have a ratonal relaton. results from an nteger up-samplng and an nteger sub-samplng Requred pxel on other densty grd: 4 n 5 The flter coeffcents, a b c d e f.., depend on the poston of the requred pxel n Dfferent flter types Non-lnear flters Lnear flters Rank-order flters Hybrd flters Morphologcal flterng Adaptve flters 7 Lnear flters not very effectve aganst shot nose Lnear flters fne n case small nose values are more frequent than hgh nose values E.g. Gaussan nose Shot nose (salt and pepper nose,..) s characterzed by relatvely few extreme nose values and (almost) no small nose values E.g. mpulses from gnton of combuston engnes, or flm defects, defectve pxels To know f a pxel s extreme we have to rank adjacent pxel values 7 Lnear and rank-order flters The dfference Ths s what a lnear flter outputs: y w x w x... w n x n Let the flter support (vector) defne the nput pxels used: T S( x) ( x, x,..., xn) and the coeffcent vector defnes the weghtng: W w, w,..., w ) ( n Now we can brefly defne the fltered output pxel as: y( x) W. S( x)

13 3 73 Lnear and rank-order flters The dfference 74 Lnear and rank-order flters The dfference So, the lnear flter s defned as: y( x) W. S( x) If we now defne the ordered (ranked) support: T Sr ( x) ( x( ), x(),..., x( n) ) wth: x x x ( ) ()... ( n) Then the rank-order flter s defned by: y( x) W r. S ( x) r The pxel-weghts n a lnear flter are determned by the spatotemporal poston of the pxel relatve to the output poston The pxel-weghts n a rank-order flter are determned by the rank number of the pxel after orderng all values n the support. Examples: mnmum, maxmum, mdpont = (max + mn)/ medan, α-trmmed mean 75 Effectveness aganst shot nose 76 Less extreme dstrbutons: The α-trmmed-mean flter Orgnal 3x3 box 3x3 medan The general rank-order flter s defned by: y( x) W r. S ( x) r The medan flter, especally effectve for shot nose: W med (0,...,...,0) The α-trmmed-mean flter, long-tal dstrbuted nose, but less extreme: (0,...0,,,,,,0...,0) W Medan flter, especally effectve for shot nose: y( x) W med. Sr ( x) W med (0,...,...,0) α-central pxels are averaged, extremes gnored In an alternatve termnology, α dentfes the percentage of pxels that s trmmed-off 77 Shot nose + Gaussan nose reducton 78 The max and mn flter (morphologcal flterng) Orgnal sgnal 5x5 mn flter (dlaton) 5x5 mn- followed by 5x5 max-flter (closng) Orgnal 3x3 medan Orgnal sgnal 5x5 max flter (eroson) 5x5 max- followed by 5x5 mn-flter (openng) 3x3 box 3x3 -trmmed mean

14 4 79 Relevance: moton detecton sgnals 80 Combnaton of lnear and rank-order: The hybrd flter Orgnal sgnal 5x5 max flter (dlaton) 5x5 max- cascaded wth 5x5 mn-flter (closng) If we concatenate the lnear and ranked supports: S T h( x) ( x, x,..., xn, x(), x(),..., x( n) ) and also the coeffcent vectors defnes the weghtng: Wh ( w, w,..., wn, wr, wr,..., wrn) Orgnal sgnal 5x5 mn flter (eroson) 5x5 mn- cascaded wth 5x5 max-flter (openng) Then the hybrd flter s defned by: y( x) W. S h ( x) h 8 Combnaton of lnear and rank-order: The blateral flter 8 Blateral flter example: edge-preservng nose flter In the lnear flter, weghts depend on poston relatve to centre: w k k f( c k) In a rank-order flter, they may depend on the smlarty wth current pxel: w f ( x x ) k c In the blateral flter the weght s defned by: w N f ( c k). f ( x k. k c x ) Where N s selected such that the sum of the coeffcents s Functons f and f may be e.g. Gaussan or trangular functons Optmzaton s often a problem Flter optmzaton Brefly back to lnear flters to prepare for adaptve flterng Store vdeo Program enhancement algorthm Play enhanced vdeo Subjectve assessment Subjectve assessment s tme consumng Huge desgn space requres automatc optmzaton

15 5 85 It seems straghtforward 86 Bottleneck elmnated? Desred output Processng output Store vdeo Play enhanced vdeo (a-b) /N.S MSE Program enhancement algorthm MSE optmzaton Mnmze a sum of squared pxel dfferences by varyng all parameters 87 MSE-optmal (non-recursve) lnear flterng Cross-correlaton matrx The lnear flter s defned by: Auto-correlaton matrx y w x w x w3 x3... w n x n Compare the output wth orgnal mage to know error: e y o y For a mnmal MSE, the frst dervatve should be zero: e e e xe 0 w w X X... X n w Y If we now defne: X X... X n w Y X j x x j Y x y o We then get: X n X n... X nnwn Yn 88 MSE-optmal (non-recursve) lnear flterng We have to solve the followng equaton: X X... X n X X... X We thereto wrte: n X n w Y X n w Y X nnwn Yn X.W Y And conclude upon the followng optmal weghts: W X X.W X Y 89 We can use ths for nose reducton Orgnal nosy mage 3-taps flter support 90 We may also use t for de-blurrng nosy mages Orgnal blurred nosy mage Concluson: LMSE-flter enhances nose less Therefore, also reduces the blur less MSE-best compromse 3-taps LMSE-flter 3-taps averagng flter 3x3LMSE-flter 3x3 peakng-flter

16 6 9 How does ths compare wth nverse flterng? 9 Possble to calculate an nverse flter for perfect deblurrng. Typcally a recursve flter Infntely hgh gan results, f blurrng flter has a zero response for a gven frequency Consequence s that nose wll be amplfed unlmted Orgnal blurred nosy mage De-blurrng wth nverse flter Problem wth LMSE-flters 93 Images wth dentcal MSE Nosy Blurred Nosy Compressed Blurred Sharpened 5LIN0 Vdeo processng Compressed Sharpened Gerard de Haan Adaptve flterng Traned Flters (optmzng adaptve flters)

17 7 97 Images wth dentcal MSE Nosy Nosy Blurred Compressed Blurred Sharpened 98 What s wrong wth MSE? The mstake s n the averagng! Some pcture parts get a lot better wth a method that s poor on the average! characterze mage parts that can be treated dentcally Compressed Sharpened and ndvdually optmze (LMSE!) parts wth the same character! 99 Classfcaton based flterng 00 Support wth nput pxels x x x 3 x 4 x 5 x 6 x 7 x 8 x 9 Classfcaton LUT Lnear flterng y Classfcaton examples y w x w x... w c c c x nc n 0 Local edge drecton classfcaton 0 Local contrast classfcaton Image Edge drectons (colour) Image Local contrast (colour) 0 o 50 o 90 o 30 o 70 o Low Hgh

18 8 03 Local sharpness classfcaton Image Local sharpness (colour) 04 Codng the local structure to classfy support data Classfcaton code s concatenaton of pxels reduced to sngle bt:,( x xav ) ADRC( x) 0,( x xav ) wth x av n n x Low Hgh Archtecture of the hybrd flter Not only lnear flterng Classfcaton LUT Non-lnear flterng Concept can be extended to traned rank-order and blateral flters Traned Mxng of processng optons Not only flterng metrcs Classfcaton Processng opton Processng opton LUT Mxer Processng opton

19 Sharpenng and de-nosng Applcaton examples Nose flter wthout classfcaton Traned Flter output PxelPlus, PerfectPxel, Dgtal Realty Creaton, Up-scalng combned wth sharpenng Interlace artfact removal Standard up-scalng Traned Flter output Lne-average nput Traned Flter 3 Enhancement of dgtal vdeo Traned Flter output 4 Mcroscopy resoluton enhancement : RED : RED GREEN BLUE

20 0 5 Up-scalng of MRI-mages 6 Conclusons MSE (%) 00 Classfcaton LUT 80 In Flter Out MSE Smlar archtecture for many functons 0 0 HD BLn Fourer TF Desgn methodology replaces heurstcs 7 The value of traned flterng 8 Automatc optmzaton Desgn methodology replaces heurstcs for tunng No thnkng faster Desgner creatvty stll for fndng the relevant classes Neghbourhood Selecton 9 Neghbourhood selecton 0 Adaptve flterng: Neghbourhood Selecton So far, we assumed that ALL pxels n the support are combned under ALL crcumstances We may also propose to adapt the set of pxels that are combned to expectatons about local correlaton, usng a socalled neghbourhood selecton technque A neghbourhood shall be defned as a sub-set of the support Varous optons exst to exclude pxels from the support K-nearest, sgma nearest, symmetrcal nearest, Pxels n the neghbourhood are then combned Weghted averagng, rank-order, etc. support Select neghbourhood from support E.g. only pxels smlar to current pxel neghbourhood Combne pxels n neghbourhood Lnear flter Weghted average Rank-order flter Mn, max, medan

21 Support and neghborhood examples Case Case Non-adaptve: Standard neghborhood Case Case Example 3x3 support contans all 9 pxels Center pxel s the current pxel For dfferent technques the neghborhood pxels wll be shown wth a red border In a standard neghborhood all pxels from the support end up n the neghborhood 3 Effect of the standard neghbourhood selecton (5x5) 4 Neghbourhood selecton: K () nearest neghbours Case Case In a K () nearest neghborhood, pxels plus the central pxel, end up n the neghborhood 5 Effect of the k () nearest neghbourhood selecton (5x5) 6 Neghbourhood selecton: Sgma nearest neghbours Case Case In a Sgma nearest neghborhood, the number of [pxels n the neghborhood depends on how smlar the pxels are

22 Grey-level Grey-level Grey-level Vdeo course: Bascs: Flterng 7 Effect of the sgma nearest selecton (5x5) 8 Neghbourhood selecton: Symmetrc nearest neghbours Case Case In a Symmetrc nearest neghborhood, half of the pxels plus the central pxel, end up n the neghborhood 9 Neghbourhood selecton: Symmetrc nearest neghbours 30 Neghbourhood selecton: Symmetrc nearest neghbours 3 Pxel,, and 3 n neghborhood E.g. averagng leads to an output value hgher than the central pxel Pxel 4, 5, and 6 n neghborhood E.g. averagng leads to an output value lower than the central pxel 4 Poston Poston 3 Neghbourhood selecton: Symmetrc nearest neghbours 3 The effect of symmetrc nearest neghbour flterng (5x5) Result s a steeperedge In ths case symmetrc neghborhood selecton, wth pxel averagng leads to a sharper edge! Poston

23 3 33 The effect of symmetrc nearest neghbour flterng (5x5) 34 Adaptve flterng: Neghbourhood Selecton Select neghbourhood Combne pxels n from support neghbourhood support E.g. only pxels close to current pxel neghbourhood Lnear flter Weghted average Rank-order flter Mn, max, medan 35 Prepare yourself for the exam Before: Chapter, 3, and 9 Today: Chapter 4 I recommend you read the text Book avalable at Flux 4:3 Book may be used durng exam And try the exercses n the book: Chapter, 3 and 9 Chapter 4 You have to download VdProc (w3.es.ele.tue.nl/~dehaan/ ) Send me e-mal for password (G.d.Haan@tue.nl) 36 Quanttatve analyss of lnear flters 37 Frequency response of a lnear flter 38 Frequency response of a lnear flter H(0) H() H(N) sum y [ k] h( ) x[ k ] Gven a system wth mpulse response h() Gven an nput sgnal = complex snusod transfer jk x( k) e The output sgnal s: y( k) h( ). x( k ) And the frequency response:, k h( ). e j H ( e ) h( ). e j( k) jw e jk h( ). e j

24 4 39 Example: the flter 40 For all flters wth a symmetrc mpulse response Followng the defnton: j jw jw 0 jw H( e ) h( ). e e e e e jw cos t jsnt Substtutng: We get: H( ) cos( t ) jsn( t ) cos t jsnt Whch smplfes to: 4 H( ) cos t Symmetrc: h()=h(-), =..L L The frequency response s L j j. H( e ) h( ). e... a.cos(. ) L 0 Real-valued (=zero-phase) transfer functon Ths translates nto a constant for all frequences L ½.f s f 4 Example: the ( ) - flter 4 Example: the flter Followng the defnton: j jw 0 jw H( e ) h( ). e e e Substtutng: e jw cos t jsnt We get a complex frequency transfer (non-zero phase): H( ) cos t jsnt We can calculate the magntude of the transfer: H( ) R I cos t cos t sn t So, we have: Ths smplfes to: And we remember from modulaton: So, we conclude: H( ) cos t cos t sn t H( ) cos t cos cos cos( ) cos( ) 4cos ( t / ) cost H( ) cos( t / ) ½.f s f 43 What f we re-use output samples? Recursve flterng h(0) h() h(n) sum h r (M) h r () y ( k) h( ). x( k ) h ( m). y( k m) m r 44 What about a recursve flter? Now the output sgnal s: y( k) or: M N 0 h( ). x( k ) M h ( m). y( k m) r m ( h r ( m)). y( k) h( )). x( k ) m or: h( n) n y( k). x( k) hr ( m) m jn h( n). e n and fnally: H ( j) j h ( m). e m r m x( k) e jk

25 5 45 Inverse flterng Assume a sgnal has been fltered wth a (½ ½) - flter Can we nvert the operaton (de-blur)? j jw 0 jw H( ) ( e ) h( ). e 0.5e 0. 5e We are lookng for H - (e jω ),.e.: j H () e ) jw 0 jw h() ( ). e e e ( e Ths s a recursve flter!: ( jw ) 46 The nverse of a (½ ½) - flter H(0)= H() H(N) sum h r (M) h r ()=- 47 So ths s equvalent to a copper wre: 48 Consequences of nfnte gan nverse flter Result of cascade: Same nose, but nverse of ½ 0 ½ ½ ½ sum - + ½.f s f But f there was some nose: Worse than the nosy blurred: f ½.f s Snce the (½ ½) flter has a zero transfer for the Nyqust frequency, the nverse flter has an nfnte gan there! 49 Solutons for de-blurrng nosy mages 50 Prepare yourself for the exam The value of the nverse flter s lmted because of nose amplfcaton The nose easly domnates the percepton f blurrng process suppresses parts of the spectrum sgnfcantly Norbert Wener ( ) proposed the MSE-optmal soluton by addng a nose-dependng term n the denomnator The result s commonly referred to as the Wener Flter MSE-best soluton modellng blur AND nose Last week: Chapter and Chapter 3 Today: Chapter 4 I recommend you read the text Book avalable at Pt9:4 And try the exercses n the book: Chapter -3 Chapter 4 You have to download VdProc (w3.cs.ele.tue.nl/~dehaan/ ) Send me e-mal for password (G.d.Haan@tue.nl)

Tutorial 2. COMP4134 Biometrics Authentication. February 9, Jun Xu, Teaching Asistant

Tutorial 2. COMP4134 Biometrics Authentication. February 9, Jun Xu, Teaching Asistant Tutoral 2 COMP434 ometrcs uthentcaton Jun Xu, Teachng sstant csjunxu@comp.polyu.edu.hk February 9, 207 Table of Contents Problems Problem : nswer the questons Problem 2: Power law functon Problem 3: Convoluton

More information

Fourier Transform. Additive noise. Fourier Tansform. I = S + N. Noise doesn t depend on signal. We ll consider:

Fourier Transform. Additive noise. Fourier Tansform. I = S + N. Noise doesn t depend on signal. We ll consider: Flterng Announcements HW2 wll be posted later today Constructng a mosac by warpng mages. CSE252A Lecture 10a Flterng Exampel: Smoothng by Averagng Kernel: (From Bll Freeman) m=2 I Kernel sze s m+1 by m+1

More information

Digital Signal Processing

Digital Signal Processing Dgtal Sgnal Processng Dscrete-tme System Analyss Manar Mohasen Offce: F8 Emal: manar.subh@ut.ac.r School of IT Engneerng Revew of Precedent Class Contnuous Sgnal The value of the sgnal s avalable over

More information

Report on Image warping

Report on Image warping Report on Image warpng Xuan Ne, Dec. 20, 2004 Ths document summarzed the algorthms of our mage warpng soluton for further study, and there s a detaled descrpton about the mplementaton of these algorthms.

More information

TOPICS MULTIPLIERLESS FILTER DESIGN ELEMENTARY SCHOOL ALGORITHM MULTIPLICATION

TOPICS MULTIPLIERLESS FILTER DESIGN ELEMENTARY SCHOOL ALGORITHM MULTIPLICATION 1 2 MULTIPLIERLESS FILTER DESIGN Realzaton of flters wthout full-fledged multplers Some sldes based on support materal by W. Wolf for hs book Modern VLSI Desgn, 3 rd edton. Partly based on followng papers:

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Lossy Compression. Compromise accuracy of reconstruction for increased compression.

Lossy Compression. Compromise accuracy of reconstruction for increased compression. Lossy Compresson Compromse accuracy of reconstructon for ncreased compresson. The reconstructon s usually vsbly ndstngushable from the orgnal mage. Typcally, one can get up to 0:1 compresson wth almost

More information

Errors for Linear Systems

Errors for Linear Systems Errors for Lnear Systems When we solve a lnear system Ax b we often do not know A and b exactly, but have only approxmatons  and ˆb avalable. Then the best thng we can do s to solve ˆx ˆb exactly whch

More information

Linear Feature Engineering 11

Linear Feature Engineering 11 Lnear Feature Engneerng 11 2 Least-Squares 2.1 Smple least-squares Consder the followng dataset. We have a bunch of nputs x and correspondng outputs y. The partcular values n ths dataset are x y 0.23 0.19

More information

For now, let us focus on a specific model of neurons. These are simplified from reality but can achieve remarkable results.

For now, let us focus on a specific model of neurons. These are simplified from reality but can achieve remarkable results. Neural Networks : Dervaton compled by Alvn Wan from Professor Jtendra Malk s lecture Ths type of computaton s called deep learnng and s the most popular method for many problems, such as computer vson

More information

Generalized Linear Methods

Generalized Linear Methods Generalzed Lnear Methods 1 Introducton In the Ensemble Methods the general dea s that usng a combnaton of several weak learner one could make a better learner. More formally, assume that we have a set

More information

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could

More information

GEMINI GEneric Multimedia INdexIng

GEMINI GEneric Multimedia INdexIng GEMINI GEnerc Multmeda INdexIng Last lecture, LSH http://www.mt.edu/~andon/lsh/ Is there another possble soluton? Do we need to perform ANN? 1 GEnerc Multmeda INdexIng dstance measure Sub-pattern Match

More information

NON-LINEAR CONVOLUTION: A NEW APPROACH FOR THE AURALIZATION OF DISTORTING SYSTEMS

NON-LINEAR CONVOLUTION: A NEW APPROACH FOR THE AURALIZATION OF DISTORTING SYSTEMS NON-LINEAR CONVOLUTION: A NEW APPROAC FOR TE AURALIZATION OF DISTORTING SYSTEMS Angelo Farna, Alberto Belln and Enrco Armellon Industral Engneerng Dept., Unversty of Parma, Va delle Scenze 8/A Parma, 00

More information

Instance-Based Learning (a.k.a. memory-based learning) Part I: Nearest Neighbor Classification

Instance-Based Learning (a.k.a. memory-based learning) Part I: Nearest Neighbor Classification Instance-Based earnng (a.k.a. memory-based learnng) Part I: Nearest Neghbor Classfcaton Note to other teachers and users of these sldes. Andrew would be delghted f you found ths source materal useful n

More information

Chapter 3 Describing Data Using Numerical Measures

Chapter 3 Describing Data Using Numerical Measures Chapter 3 Student Lecture Notes 3-1 Chapter 3 Descrbng Data Usng Numercal Measures Fall 2006 Fundamentals of Busness Statstcs 1 Chapter Goals To establsh the usefulness of summary measures of data. The

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

ADVANCED MACHINE LEARNING ADVANCED MACHINE LEARNING

ADVANCED MACHINE LEARNING ADVANCED MACHINE LEARNING 1 ADVANCED ACHINE LEARNING ADVANCED ACHINE LEARNING Non-lnear regresson technques 2 ADVANCED ACHINE LEARNING Regresson: Prncple N ap N-dm. nput x to a contnuous output y. Learn a functon of the type: N

More information

Problem Set 9 Solutions

Problem Set 9 Solutions Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem

More information

Department of Electrical & Electronic Engineeing Imperial College London. E4.20 Digital IC Design. Median Filter Project Specification

Department of Electrical & Electronic Engineeing Imperial College London. E4.20 Digital IC Design. Median Filter Project Specification Desgn Project Specfcaton Medan Flter Department of Electrcal & Electronc Engneeng Imperal College London E4.20 Dgtal IC Desgn Medan Flter Project Specfcaton A medan flter s used to remove nose from a sampled

More information

Section 8.3 Polar Form of Complex Numbers

Section 8.3 Polar Form of Complex Numbers 80 Chapter 8 Secton 8 Polar Form of Complex Numbers From prevous classes, you may have encountered magnary numbers the square roots of negatve numbers and, more generally, complex numbers whch are the

More information

The Concept of Beamforming

The Concept of Beamforming ELG513 Smart Antennas S.Loyka he Concept of Beamformng Generc representaton of the array output sgnal, 1 where w y N 1 * = 1 = w x = w x (4.1) complex weghts, control the array pattern; y and x - narrowband

More information

Lecture 12: Discrete Laplacian

Lecture 12: Discrete Laplacian Lecture 12: Dscrete Laplacan Scrbe: Tanye Lu Our goal s to come up wth a dscrete verson of Laplacan operator for trangulated surfaces, so that we can use t n practce to solve related problems We are mostly

More information

The Geometry of Logit and Probit

The Geometry of Logit and Probit The Geometry of Logt and Probt Ths short note s meant as a supplement to Chapters and 3 of Spatal Models of Parlamentary Votng and the notaton and reference to fgures n the text below s to those two chapters.

More information

CONTRAST ENHANCEMENT FOR MIMIMUM MEAN BRIGHTNESS ERROR FROM HISTOGRAM PARTITIONING INTRODUCTION

CONTRAST ENHANCEMENT FOR MIMIMUM MEAN BRIGHTNESS ERROR FROM HISTOGRAM PARTITIONING INTRODUCTION CONTRAST ENHANCEMENT FOR MIMIMUM MEAN BRIGHTNESS ERROR FROM HISTOGRAM PARTITIONING N. Phanthuna 1,2, F. Cheevasuvt 2 and S. Chtwong 2 1 Department of Electrcal Engneerng, Faculty of Engneerng Rajamangala

More information

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl RECURSIVE SPLINE INTERPOLATION METHOD FOR REAL TIME ENGINE CONTROL APPLICATIONS A. Stotsky Volvo Car Corporaton Engne Desgn and Development Dept. 97542, HA1N, SE- 405 31 Gothenburg Sweden. Emal: astotsky@volvocars.com

More information

Lecture 10 Support Vector Machines II

Lecture 10 Support Vector Machines II Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed

More information

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011

Stanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011 Stanford Unversty CS359G: Graph Parttonng and Expanders Handout 4 Luca Trevsan January 3, 0 Lecture 4 In whch we prove the dffcult drecton of Cheeger s nequalty. As n the past lectures, consder an undrected

More information

Pulse Coded Modulation

Pulse Coded Modulation Pulse Coded Modulaton PCM (Pulse Coded Modulaton) s a voce codng technque defned by the ITU-T G.711 standard and t s used n dgtal telephony to encode the voce sgnal. The frst step n the analog to dgtal

More information

Chapter 8 SCALAR QUANTIZATION

Chapter 8 SCALAR QUANTIZATION Outlne Chapter 8 SCALAR QUANTIZATION Yeuan-Kuen Lee [ CU, CSIE ] 8.1 Overvew 8. Introducton 8.4 Unform Quantzer 8.5 Adaptve Quantzaton 8.6 Nonunform Quantzaton 8.7 Entropy-Coded Quantzaton Ch 8 Scalar

More information

EEE 241: Linear Systems

EEE 241: Linear Systems EEE : Lnear Systems Summary #: Backpropagaton BACKPROPAGATION The perceptron rule as well as the Wdrow Hoff learnng were desgned to tran sngle layer networks. They suffer from the same dsadvantage: they

More information

Transform Coding. Transform Coding Principle

Transform Coding. Transform Coding Principle Transform Codng Prncple of block-wse transform codng Propertes of orthonormal transforms Dscrete cosne transform (DCT) Bt allocaton for transform coeffcents Entropy codng of transform coeffcents Typcal

More information

Meshless Surfaces. presented by Niloy J. Mitra. An Nguyen

Meshless Surfaces. presented by Niloy J. Mitra. An Nguyen Meshless Surfaces presented by Nloy J. Mtra An Nguyen Outlne Mesh-Independent Surface Interpolaton D. Levn Outlne Mesh-Independent Surface Interpolaton D. Levn Pont Set Surfaces M. Alexa, J. Behr, D. Cohen-Or,

More information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng

More information

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography CSc 6974 and ECSE 6966 Math. Tech. for Vson, Graphcs and Robotcs Lecture 21, Aprl 17, 2006 Estmatng A Plane Homography Overvew We contnue wth a dscusson of the major ssues, usng estmaton of plane projectve

More information

Feb 14: Spatial analysis of data fields

Feb 14: Spatial analysis of data fields Feb 4: Spatal analyss of data felds Mappng rregularly sampled data onto a regular grd Many analyss technques for geophyscal data requre the data be located at regular ntervals n space and/or tme. hs s

More information

Tracking with Kalman Filter

Tracking with Kalman Filter Trackng wth Kalman Flter Scott T. Acton Vrgna Image and Vdeo Analyss (VIVA), Charles L. Brown Department of Electrcal and Computer Engneerng Department of Bomedcal Engneerng Unversty of Vrgna, Charlottesvlle,

More information

A linear imaging system with white additive Gaussian noise on the observed data is modeled as follows:

A linear imaging system with white additive Gaussian noise on the observed data is modeled as follows: Supplementary Note Mathematcal bacground A lnear magng system wth whte addtve Gaussan nose on the observed data s modeled as follows: X = R ϕ V + G, () where X R are the expermental, two-dmensonal proecton

More information

IMAGE DENOISING USING NEW ADAPTIVE BASED MEDIAN FILTER

IMAGE DENOISING USING NEW ADAPTIVE BASED MEDIAN FILTER Sgnal & Image Processng : An Internatonal Journal (SIPIJ) Vol.5, No.4, August 2014 IMAGE DENOISING USING NEW ADAPTIVE BASED MEDIAN FILTER Suman Shrestha 1, 2 1 Unversty of Massachusetts Medcal School,

More information

CSE4210 Architecture and Hardware for DSP

CSE4210 Architecture and Hardware for DSP 4210 Archtecture and Hardware for DSP Lecture 1 Introducton & Number systems Admnstratve Stuff 4210 Archtecture and Hardware for DSP Text: VLSI Dgtal Sgnal Processng Systems: Desgn and Implementaton. K.

More information

Statistical Inference. 2.3 Summary Statistics Measures of Center and Spread. parameters ( population characteristics )

Statistical Inference. 2.3 Summary Statistics Measures of Center and Spread. parameters ( population characteristics ) Ismor Fscher, 8//008 Stat 54 / -8.3 Summary Statstcs Measures of Center and Spread Dstrbuton of dscrete contnuous POPULATION Random Varable, numercal True center =??? True spread =???? parameters ( populaton

More information

Internet Engineering. Jacek Mazurkiewicz, PhD Softcomputing. Part 3: Recurrent Artificial Neural Networks Self-Organising Artificial Neural Networks

Internet Engineering. Jacek Mazurkiewicz, PhD Softcomputing. Part 3: Recurrent Artificial Neural Networks Self-Organising Artificial Neural Networks Internet Engneerng Jacek Mazurkewcz, PhD Softcomputng Part 3: Recurrent Artfcal Neural Networks Self-Organsng Artfcal Neural Networks Recurrent Artfcal Neural Networks Feedback sgnals between neurons Dynamc

More information

Nice plotting of proteins II

Nice plotting of proteins II Nce plottng of protens II Fnal remark regardng effcency: It s possble to wrte the Newton representaton n a way that can be computed effcently, usng smlar bracketng that we made for the frst representaton

More information

Negative Binomial Regression

Negative Binomial Regression STATGRAPHICS Rev. 9/16/2013 Negatve Bnomal Regresson Summary... 1 Data Input... 3 Statstcal Model... 3 Analyss Summary... 4 Analyss Optons... 7 Plot of Ftted Model... 8 Observed Versus Predcted... 10 Predctons...

More information

Impulse Noise Removal Technique Based on Fuzzy Logic

Impulse Noise Removal Technique Based on Fuzzy Logic Impulse Nose Removal Technque Based on Fuzzy Logc 1 Mthlesh Atulkar, 2 A.S. Zadgaonkar and 3 Sanjay Kumar C V Raman Unversty, Kota, Blaspur, Inda 1 m.atulkar@gmal.com, 2 arunzad28@hotmal.com, 3 sanrapur@redffmal.com

More information

), it produces a response (output function g (x)

), it produces a response (output function g (x) Lnear Systems Revew Notes adapted from notes by Mchael Braun Typcally n electrcal engneerng, one s concerned wth functons of tme, such as a voltage waveform System descrpton s therefore defned n the domans

More information

Outline and Reading. Dynamic Programming. Dynamic Programming revealed. Computing Fibonacci. The General Dynamic Programming Technique

Outline and Reading. Dynamic Programming. Dynamic Programming revealed. Computing Fibonacci. The General Dynamic Programming Technique Outlne and Readng Dynamc Programmng The General Technque ( 5.3.2) -1 Knapsac Problem ( 5.3.3) Matrx Chan-Product ( 5.3.1) Dynamc Programmng verson 1.4 1 Dynamc Programmng verson 1.4 2 Dynamc Programmng

More information

MEM 255 Introduction to Control Systems Review: Basics of Linear Algebra

MEM 255 Introduction to Control Systems Review: Basics of Linear Algebra MEM 255 Introducton to Control Systems Revew: Bascs of Lnear Algebra Harry G. Kwatny Department of Mechancal Engneerng & Mechancs Drexel Unversty Outlne Vectors Matrces MATLAB Advanced Topcs Vectors A

More information

Lecture Topics VMSC Prof. Dr.-Ing. habil. Hermann Lödding Prof. Dr.-Ing. Wolfgang Hintze. PD Dr.-Ing. habil.

Lecture Topics VMSC Prof. Dr.-Ing. habil. Hermann Lödding Prof. Dr.-Ing. Wolfgang Hintze. PD Dr.-Ing. habil. Lecture Topcs 1. Introducton 2. Sensor Gudes Robots / Machnes 3. Motvaton Model Calbraton 4. 3D Vdeo Metrc (Geometrcal Camera Model) 5. Grey Level Pcture Processng for Poston Measurement 6. Lght and Percepton

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

Numerical Heat and Mass Transfer

Numerical Heat and Mass Transfer Master degree n Mechancal Engneerng Numercal Heat and Mass Transfer 06-Fnte-Dfference Method (One-dmensonal, steady state heat conducton) Fausto Arpno f.arpno@uncas.t Introducton Why we use models and

More information

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

Comparison of Wiener Filter solution by SVD with decompositions QR and QLP

Comparison of Wiener Filter solution by SVD with decompositions QR and QLP Proceedngs of the 6th WSEAS Int Conf on Artfcal Intellgence, Knowledge Engneerng and Data Bases, Corfu Island, Greece, February 6-9, 007 7 Comparson of Wener Flter soluton by SVD wth decompostons QR and

More information

PHYS 705: Classical Mechanics. Calculus of Variations II

PHYS 705: Classical Mechanics. Calculus of Variations II 1 PHYS 705: Classcal Mechancs Calculus of Varatons II 2 Calculus of Varatons: Generalzaton (no constrant yet) Suppose now that F depends on several dependent varables : We need to fnd such that has a statonary

More information

ADAPTIVE IMAGE FILTERING

ADAPTIVE IMAGE FILTERING Why adaptve? ADAPTIVE IMAGE FILTERING average detals and contours are aected Averagng should not be appled n contour / detals regons. Adaptaton Adaptaton = modyng the parameters o a prrocessng block accordng

More information

CS 468 Lecture 16: Isometry Invariance and Spectral Techniques

CS 468 Lecture 16: Isometry Invariance and Spectral Techniques CS 468 Lecture 16: Isometry Invarance and Spectral Technques Justn Solomon Scrbe: Evan Gawlk Introducton. In geometry processng, t s often desrable to characterze the shape of an object n a manner that

More information

SIO 224. m(r) =(ρ(r),k s (r),µ(r))

SIO 224. m(r) =(ρ(r),k s (r),µ(r)) SIO 224 1. A bref look at resoluton analyss Here s some background for the Masters and Gubbns resoluton paper. Global Earth models are usually found teratvely by assumng a startng model and fndng small

More information

arxiv:cs.cv/ Jun 2000

arxiv:cs.cv/ Jun 2000 Correlaton over Decomposed Sgnals: A Non-Lnear Approach to Fast and Effectve Sequences Comparson Lucano da Fontoura Costa arxv:cs.cv/0006040 28 Jun 2000 Cybernetc Vson Research Group IFSC Unversty of São

More information

Video Data Analysis. Video Data Analysis, B-IT. Lecture plan:

Video Data Analysis. Video Data Analysis, B-IT. Lecture plan: Vdeo Data Analss Image eatures Spatal lterng Lecture plan:. Medan lterng. Derental lters 3. Image eatures -> > mage edges 4. Edge detectors usng rst-order dervatve 5. Edge detectors usng second-order order

More information

princeton univ. F 13 cos 521: Advanced Algorithm Design Lecture 3: Large deviations bounds and applications Lecturer: Sanjeev Arora

princeton univ. F 13 cos 521: Advanced Algorithm Design Lecture 3: Large deviations bounds and applications Lecturer: Sanjeev Arora prnceton unv. F 13 cos 521: Advanced Algorthm Desgn Lecture 3: Large devatons bounds and applcatons Lecturer: Sanjeev Arora Scrbe: Today s topc s devaton bounds: what s the probablty that a random varable

More information

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

More information

Lecture 21: Numerical methods for pricing American type derivatives

Lecture 21: Numerical methods for pricing American type derivatives Lecture 21: Numercal methods for prcng Amercan type dervatves Xaoguang Wang STAT 598W Aprl 10th, 2014 (STAT 598W) Lecture 21 1 / 26 Outlne 1 Fnte Dfference Method Explct Method Penalty Method (STAT 598W)

More information

Stochastic Analysis of Image Acquisition, Interpolation and Scale-space Smoothing

Stochastic Analysis of Image Acquisition, Interpolation and Scale-space Smoothing Stochastc Analyss of Image Acquston, Interpolaton and Scale-space Smoothng Kalle Åström, Anders Heyden Dept of Mathematcs, Lund Unversty Box 8, S- Lund, Sweden emal: kalle@maths.lth.se, heyden@maths.lth.se

More information

VQ widely used in coding speech, image, and video

VQ widely used in coding speech, image, and video at Scalar quantzers are specal cases of vector quantzers (VQ): they are constraned to look at one sample at a tme (memoryless) VQ does not have such constrant better RD perfomance expected Source codng

More information

MAE140 - Linear Circuits - Winter 16 Final, March 16, 2016

MAE140 - Linear Circuits - Winter 16 Final, March 16, 2016 ME140 - Lnear rcuts - Wnter 16 Fnal, March 16, 2016 Instructons () The exam s open book. You may use your class notes and textbook. You may use a hand calculator wth no communcaton capabltes. () You have

More information

COMPUTATIONALLY EFFICIENT WAVELET AFFINE INVARIANT FUNCTIONS FOR SHAPE RECOGNITION. Erdem Bala, Dept. of Electrical and Computer Engineering,

COMPUTATIONALLY EFFICIENT WAVELET AFFINE INVARIANT FUNCTIONS FOR SHAPE RECOGNITION. Erdem Bala, Dept. of Electrical and Computer Engineering, COMPUTATIONALLY EFFICIENT WAVELET AFFINE INVARIANT FUNCTIONS FOR SHAPE RECOGNITION Erdem Bala, Dept. of Electrcal and Computer Engneerng, Unversty of Delaware, 40 Evans Hall, Newar, DE, 976 A. Ens Cetn,

More information

Math1110 (Spring 2009) Prelim 3 - Solutions

Math1110 (Spring 2009) Prelim 3 - Solutions Math 1110 (Sprng 2009) Solutons to Prelm 3 (04/21/2009) 1 Queston 1. (16 ponts) Short answer. Math1110 (Sprng 2009) Prelm 3 - Solutons x a 1 (a) (4 ponts) Please evaluate lm, where a and b are postve numbers.

More information

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system Transfer Functons Convenent representaton of a lnear, dynamc model. A transfer functon (TF) relates one nput and one output: x t X s y t system Y s The followng termnology s used: x y nput output forcng

More information

8.6 The Complex Number System

8.6 The Complex Number System 8.6 The Complex Number System Earler n the chapter, we mentoned that we cannot have a negatve under a square root, snce the square of any postve or negatve number s always postve. In ths secton we want

More information

Line Drawing and Clipping Week 1, Lecture 2

Line Drawing and Clipping Week 1, Lecture 2 CS 43 Computer Graphcs I Lne Drawng and Clppng Week, Lecture 2 Davd Breen, Wllam Regl and Maxm Peysakhov Geometrc and Intellgent Computng Laboratory Department of Computer Scence Drexel Unversty http://gcl.mcs.drexel.edu

More information

The Fourier Transform

The Fourier Transform e Processng ourer Transform D The ourer Transform Effcent Data epresentaton Dscrete ourer Transform - D Contnuous ourer Transform - D Eamples + + + Jean Baptste Joseph ourer Effcent Data epresentaton Data

More information

Workshop: Approximating energies and wave functions Quantum aspects of physical chemistry

Workshop: Approximating energies and wave functions Quantum aspects of physical chemistry Workshop: Approxmatng energes and wave functons Quantum aspects of physcal chemstry http://quantum.bu.edu/pltl/6/6.pdf Last updated Thursday, November 7, 25 7:9:5-5: Copyrght 25 Dan Dll (dan@bu.edu) Department

More information

Mixed Noise Suppression in Color Images by Signal-Dependent LMS L-Filters

Mixed Noise Suppression in Color Images by Signal-Dependent LMS L-Filters 46 R. HUDEC MIXED OISE SUPPRESSIO I COLOR IMAGES BY SIGAL-DEPEDET LMS L-FILTERS Mxed ose Suppresson n Color Images by Sgnal-Dependent LMS L-Flters Róbert HUDEC Dept. of Telecommuncatons Unversty of Žlna

More information

Lecture 12: Classification

Lecture 12: Classification Lecture : Classfcaton g Dscrmnant functons g The optmal Bayes classfer g Quadratc classfers g Eucldean and Mahalanobs metrcs g K Nearest Neghbor Classfers Intellgent Sensor Systems Rcardo Guterrez-Osuna

More information

Some Comments on Accelerating Convergence of Iterative Sequences Using Direct Inversion of the Iterative Subspace (DIIS)

Some Comments on Accelerating Convergence of Iterative Sequences Using Direct Inversion of the Iterative Subspace (DIIS) Some Comments on Acceleratng Convergence of Iteratve Sequences Usng Drect Inverson of the Iteratve Subspace (DIIS) C. Davd Sherrll School of Chemstry and Bochemstry Georga Insttute of Technology May 1998

More information

White Noise Reduction of Audio Signal using Wavelets Transform with Modified Universal Threshold

White Noise Reduction of Audio Signal using Wavelets Transform with Modified Universal Threshold Whte Nose Reducton of Audo Sgnal usng Wavelets Transform wth Modfed Unversal Threshold MATKO SARIC, LUKI BILICIC, HRVOJE DUJMIC Unversty of Splt R.Boskovca b.b, HR 1000 Splt CROATIA Abstract: - Ths paper

More information

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

More information

1 Derivation of Point-to-Plane Minimization

1 Derivation of Point-to-Plane Minimization 1 Dervaton of Pont-to-Plane Mnmzaton Consder the Chen-Medon (pont-to-plane) framework for ICP. Assume we have a collecton of ponts (p, q ) wth normals n. We want to determne the optmal rotaton and translaton

More information

CHAPTER 4 SPEECH ENHANCEMENT USING MULTI-BAND WIENER FILTER. In real environmental conditions the speech signal may be

CHAPTER 4 SPEECH ENHANCEMENT USING MULTI-BAND WIENER FILTER. In real environmental conditions the speech signal may be 55 CHAPTER 4 SPEECH ENHANCEMENT USING MULTI-BAND WIENER FILTER 4.1 Introducton In real envronmental condtons the speech sgnal may be supermposed by the envronmental nterference. In general, the spectrum

More information

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017 U.C. Berkeley CS94: Beyond Worst-Case Analyss Handout 4s Luca Trevsan September 5, 07 Summary of Lecture 4 In whch we ntroduce semdefnte programmng and apply t to Max Cut. Semdefnte Programmng Recall that

More information

Invariant deformation parameters from GPS permanent networks using stochastic interpolation

Invariant deformation parameters from GPS permanent networks using stochastic interpolation Invarant deformaton parameters from GPS permanent networks usng stochastc nterpolaton Ludovco Bag, Poltecnco d Mlano, DIIAR Athanasos Dermans, Arstotle Unversty of Thessalonk Outlne Startng hypotheses

More information

Tutorial on Image Reconstruction Based on Weighted Sum (WS) Filter Approach: From Single Image to Multi-Frame Image

Tutorial on Image Reconstruction Based on Weighted Sum (WS) Filter Approach: From Single Image to Multi-Frame Image AU J.. 3(): 75-86 (Oct. 009) utoral on Image Reconstructon Based on Weghted Sum (WS) Flter Approach: From Sngle Image to ult-frame Image Vorapoj Patanavjt Department of Computer and Network Engneerng,

More information

Time-Varying Systems and Computations Lecture 6

Time-Varying Systems and Computations Lecture 6 Tme-Varyng Systems and Computatons Lecture 6 Klaus Depold 14. Januar 2014 The Kalman Flter The Kalman estmaton flter attempts to estmate the actual state of an unknown dscrete dynamcal system, gven nosy

More information

ECE 472/572 - Digital Image Processing. Roadmap. Questions. Lecture 6 Geometric and Radiometric Transformation 09/27/11

ECE 472/572 - Digital Image Processing. Roadmap. Questions. Lecture 6 Geometric and Radiometric Transformation 09/27/11 ECE 472/572 - Dgtal Image Processng Lecture 6 Geometrc and Radometrc Transformaton 09/27/ Roadmap Introducton Image format vector vs. btmap IP vs. CV vs. CG HLIP vs. LLIP Image acquston Percepton Structure

More information

De-noising Method Based on Kernel Adaptive Filtering for Telemetry Vibration Signal of the Vehicle Test Kejun ZENG

De-noising Method Based on Kernel Adaptive Filtering for Telemetry Vibration Signal of the Vehicle Test Kejun ZENG 6th Internatonal Conference on Mechatroncs, Materals, Botechnology and Envronment (ICMMBE 6) De-nosng Method Based on Kernel Adaptve Flterng for elemetry Vbraton Sgnal of the Vehcle est Kejun ZEG PLA 955

More information

Lecture 14: Forces and Stresses

Lecture 14: Forces and Stresses The Nuts and Bolts of Frst-Prncples Smulaton Lecture 14: Forces and Stresses Durham, 6th-13th December 2001 CASTEP Developers Group wth support from the ESF ψ k Network Overvew of Lecture Why bother? Theoretcal

More information

Bezier curves. Michael S. Floater. August 25, These notes provide an introduction to Bezier curves. i=0

Bezier curves. Michael S. Floater. August 25, These notes provide an introduction to Bezier curves. i=0 Bezer curves Mchael S. Floater August 25, 211 These notes provde an ntroducton to Bezer curves. 1 Bernsten polynomals Recall that a real polynomal of a real varable x R, wth degree n, s a functon of the

More information

Moments of Inertia. and reminds us of the analogous equation for linear momentum p= mv, which is of the form. The kinetic energy of the body is.

Moments of Inertia. and reminds us of the analogous equation for linear momentum p= mv, which is of the form. The kinetic energy of the body is. Moments of Inerta Suppose a body s movng on a crcular path wth constant speed Let s consder two quanttes: the body s angular momentum L about the center of the crcle, and ts knetc energy T How are these

More information

1 Matrix representations of canonical matrices

1 Matrix representations of canonical matrices 1 Matrx representatons of canoncal matrces 2-d rotaton around the orgn: ( ) cos θ sn θ R 0 = sn θ cos θ 3-d rotaton around the x-axs: R x = 1 0 0 0 cos θ sn θ 0 sn θ cos θ 3-d rotaton around the y-axs:

More information

Chapter 4. Velocity analysis

Chapter 4. Velocity analysis 1 Chapter 4 Velocty analyss Introducton The objectve of velocty analyss s to determne the sesmc veloctes of layers n the subsurface. Sesmc veloctes are used n many processng and nterpretaton stages such

More information

Which Separator? Spring 1

Which Separator? Spring 1 Whch Separator? 6.034 - Sprng 1 Whch Separator? Mamze the margn to closest ponts 6.034 - Sprng Whch Separator? Mamze the margn to closest ponts 6.034 - Sprng 3 Margn of a pont " # y (w $ + b) proportonal

More information

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur Analyss of Varance and Desgn of Exerments-I MODULE III LECTURE - 2 EXPERIMENTAL DESIGN MODELS Dr. Shalabh Deartment of Mathematcs and Statstcs Indan Insttute of Technology Kanur 2 We consder the models

More information

Originated from experimental optimization where measurements are very noisy Approximation can be actually more accurate than

Originated from experimental optimization where measurements are very noisy Approximation can be actually more accurate than Surrogate (approxmatons) Orgnated from expermental optmzaton where measurements are ver nos Approxmaton can be actuall more accurate than data! Great nterest now n applng these technques to computer smulatons

More information

INF 5860 Machine learning for image classification. Lecture 3 : Image classification and regression part II Anne Solberg January 31, 2018

INF 5860 Machine learning for image classification. Lecture 3 : Image classification and regression part II Anne Solberg January 31, 2018 INF 5860 Machne learnng for mage classfcaton Lecture 3 : Image classfcaton and regresson part II Anne Solberg January 3, 08 Today s topcs Multclass logstc regresson and softma Regularzaton Image classfcaton

More information

2 Finite difference basics

2 Finite difference basics Numersche Methoden 1, WS 11/12 B.J.P. Kaus 2 Fnte dfference bascs Consder the one- The bascs of the fnte dfference method are best understood wth an example. dmensonal transent heat conducton equaton T

More information

Digital Modems. Lecture 2

Digital Modems. Lecture 2 Dgtal Modems Lecture Revew We have shown that both Bayes and eyman/pearson crtera are based on the Lkelhood Rato Test (LRT) Λ ( r ) < > η Λ r s called observaton transformaton or suffcent statstc The crtera

More information

CS 3710: Visual Recognition Classification and Detection. Adriana Kovashka Department of Computer Science January 13, 2015

CS 3710: Visual Recognition Classification and Detection. Adriana Kovashka Department of Computer Science January 13, 2015 CS 3710: Vsual Recognton Classfcaton and Detecton Adrana Kovashka Department of Computer Scence January 13, 2015 Plan for Today Vsual recognton bascs part 2: Classfcaton and detecton Adrana s research

More information

Digital Audio Signal Processing DASP. Lecture-2: Noise Reduction-I. Single-Channel Noise Reduction. Marc Moonen

Digital Audio Signal Processing DASP. Lecture-2: Noise Reduction-I. Single-Channel Noise Reduction. Marc Moonen Dgtal Audo Sgnal Processng DASP Lecture-: Nose Reducton-I Sngle-Channel Nose Reducton Marc Moonen Dept. E.E./ESA-SADIUS, KU Leuven marc.moonen@kuleuven.be homes.esat.kuleuven.be/~moonen/ Sngle-Channel

More information

Cathy Walker March 5, 2010

Cathy Walker March 5, 2010 Cathy Walker March 5, 010 Part : Problem Set 1. What s the level of measurement for the followng varables? a) SAT scores b) Number of tests or quzzes n statstcal course c) Acres of land devoted to corn

More information

10-701/ Machine Learning, Fall 2005 Homework 3

10-701/ Machine Learning, Fall 2005 Homework 3 10-701/15-781 Machne Learnng, Fall 2005 Homework 3 Out: 10/20/05 Due: begnnng of the class 11/01/05 Instructons Contact questons-10701@autonlaborg for queston Problem 1 Regresson and Cross-valdaton [40

More information