Improved Interpolation

Size: px
Start display at page:

Download "Improved Interpolation"

Transcription

1 Improved Interpolation Interpolation plays an important role for motion compensation with improved fractional pixel accuracy. The more precise interpolated we get, the smaller our prediction residual becomes, and the more compression we get. How do we get improved and practical interpolation? In Computer Graphics we have an analog problem of interpolating samples, but there it is samples given by a model or a designer, instead of the pixel grid of our image. Hence it makes sense to take a look at the approaches in Computer Graphics: Predetermined points (e.g. our samples) interpolated curve with various desired properties Parallel to Nyquist-Sampling Theorem for

2 reconstruction: -Interpolation of samples. -Tells us how to do the interpolation: With an ideal low pass filter, whose upper cutoff frequency is half the sampling rate (for an interpolation or upsampling factor of 2). The samples are our control points (the samples in above picture). Nyquist Reconstruction (or interpolation): Our samples correspond to our control points (with zeros in between). The result of their low pass filterings is our interpolation. Interpolation Samples Nyquist resonstruction assures that there are no frequency components above the Nyquist frequency, hence minimum frequency of change, maximally smooth interpolated curve. The sinc function

3 sinc(t)= sin(π t) π t is the impulse response of our ideal low pass filter as interpolation function: Low pass fltering: Convolution with this function Zero crossings in integers We assume that our samples are on integer sample points. The interpolation function of the other samples have a zero there, hence the interpolated curve has no influence of the other samples there, and the curve will go through the sample values. The original sampling points are maintained after the interpolation,

4 which is what we want. This can be seen in the following. In continuous space or time we can write our sequence of samples as sequence of Dirac pulse multiplied by the sample values, x(n)δ(t n). n=0 Filtering is a convolution of our samples with the Sinc function. Continuous time convolution is (e.g. with a(t) our sample sequence and b(t) our ideal low pass filter impulse response) Nyquist Samples/Kontrollpunkte Delta function - (Kronecker delta, integral over Kronecker delta is 1)

5 Result Parts will be summed Sinc function is infnite polynomial, infnitely continuously diferentiable

6 interpolated curve is infnitely diferentiable! Better than, for example Cubic splines (Computer Graphics), which are continuously diferentiable only until the nnd derivative. Problem: sinc function is infnitely long, not so practical. Alternative options for interpolation flter: convolution of our samples with the so-called "Box function" h 1 (t) as a basis function. t Value is retained until the next one comes. Resultat: No continuity, problem because we have mostly smooth intensity gradients in images. Another way to preserve the continuity: Linear interpolation:

7 Basis function is the Box function convolved with itself. h 2 (x) t = = t * h 2 (t)=h 1 (t) h 1 (t) Interpolation with this basis function t Q_1(u) Q_2(u) Sum Linear interpolation f(0) 0<u<1 f(1) Observe: After doing the convolution, we obtain an increasing and a decreasing part for each to be interpolated interval, as can be seen in above picture. Each interval between two sampling points contains two functions, the rising and the falling slope of our basis function. t

8 We can write this basis function as two linear functions or polynomials of 1 st order of a local variable u, which varies between 0 and 1 in this interval between two neighboring sampling points. For the rising part we get, Q 1 (u)=u for the falling part we get, Q 2 (u)=1 u Let the sample value at the lower end of the interval be f (0) and at the upper end be f (1). Then we can linearly interpolate the in-between value at point u as f (u)=f (0) Q 2 (u)+f (1) Q 1 (u) or f (u)=f (0) (1 u)+u f (1) Observe: This sum results from the convolution of the sampling points with the basis function. Example: We have n neighboring pixels, f (0)=10, f (1)=12. What is the sample right in the middle, using linear interpolation? Answer: We use our interpolation formula with u=0.5, f (u)=f (0) (1 u)+u f (1)=10 (1 u)+u 12= = =11

9 Observe: The convolution in space or time is a multiplication in the frequency domain. Hence a convolution of two box functions becomes a squared sinc function in the frequency domain! Note: basic function is 0 for the other samples Interpolated curve goes through our samples. C0 continuity, but no C1 continuity of the frst derivative. Next step: h 4 (t )=h 2 (t ) h 2 (t )=h 1 (t) h 1 (t ) h 1 (t ) h 1 (t) cubic B-Spline Observe: In the frequency domain, this B-Spline is a sinc function to the 4 th power! (sinc⁴) (The B is for Basis )

10 Example: We compute the discrete B-Spline basis function using convolution with Python, by discretizing each unit interval into 10 steps: import numpy as np import matplotlib.pyplot as plt import scipy.signal as sig h1=np.ones(10) h2=sig.convolve(h1,h1) h4=sig.convolve(h2,h2) plt.plot(h4) plt.show() Q1 Q2 Q3 Q4

11 Observe: We now have 4 unit intervals (between the sample points, in the plot at multiples of 10). Each unit interval will have its own polynomial, Q 1 (u),...,q 4 (u), as depicted in above plot. Observe that in this way we get a very compact support, which means our impulse response is quite short (length only from 0 to 4), and it can be represented by polynomials Q i (u) of third order. Both makes it quite efcient for an implementation, which is important for video coding purposes. The drawback is now that the B-Spline does not go through 0 at the other samples (at positions 1 and 3), such that an interpolation with a simple convolution will not go through the original sampling points/values, which is not good. To fx this, we can pre-process the original signal (image) frst, such that the original samples are again obtained after the convolution with the B- Splines, as we will see. We start with the 4 th derivative, because it is easy to compute, and then we integrate 4 times,

12 to obtain the polynomial functions Q_i itself. We can use this convolution for our interpolation, but this only produces interpolated values on a fxed grid between the sample points. If we want to have an interpolated value at an arbitrary position u between n sample points, we have to fnd the polynomials Q 1 (u),...,q 4 (u). How do we get them? By observing the derivatives! Derivation rule: because d dt (h 1(t) h 1 (t))= d dt h 1(t) h 1 (t) d dt a(u) b(t u)du= a(u) d dt b(t u)du =a(t) d dt b(t)

13 4th Derivation of the B-splines: d 4 dt 4 h 4(t )= d dt h 1(t) d dt h 1(t) d dt h 1(t) d dt h 1(t) d dt h 1(t )=δ(t) δ(t 1) d4 dt 4 h 4(t )=δ(t) 4δ(t 1)+6 δ(t 2) 4 δ(t 3)+δ(t 4) (using δ(x x 0 ) δ(x x 1 )=δ(x x 0 x 1 ) ) Computation of these coefcients in Python: import numpy as np import scipy.signal as sig #the 2 diracs of the derivative of the box function: dh=np.array([1,-1]) cdh=sig.convolve(dh,dh) print("cdh= ",cdh ) ('cdh= ', array([ 1, -2, 1])) #4 derivatives of box function convolved:

14 ccdh=sig.convolve(cdh,cdh) print("ccdh= ",ccdh) ('ccdh= ', array([ 1, -4, 6, -4, 1])) We obtain the third derivative of our spline by integrating the 4 th derivative (without integration constant), d 3 dt 3 = (δ(t) 4 δ(t 1)+6 δ(t 2) 4 δ(t 3)+δ(t 4))dt This is a piecewise constant step function

15 Now we want to obtain the polynomials Q 1 (u),...,q 4 (u) with the local variable u. We start with Q 1 (u). d 3 dt Q 1(u)=1 3 The second derivative of Q 1 (u) is the integral of the third derivative, d 2 du Q 1(u)= 1du=u 2 next more integration, d du Q 1(u)= u du= u2 2, Q 1 (u)= u 2 2 du= u3 6

16 For Q 2 (u) we also need conditions for continuity for the second and frst derivatives, and the function value, where Q 1 (u) and Q 2 (u) touch, at u=1 for the frst interval with Q 1 (u) and u=0 for the second interval with Q 2 (u). d 3 dt Q 2(u)= 3 3 d 2 dt Q 2(u)= 3u+c 2 2 To determine c 2 we need the values of the second derivatives identical at the upper end of the lower interval ( u=1 ) and the lower end of the upper interval ( u=0 ), for continuity of the second derivative, d 2 du Q 1(u=1)= d 2 2 dt Q 2(u=0) 2 hence we get 1= 3 0+c 2 c 2 =1 and d 2 dt 2 Q 2(u)= 3u+1 to continue integration, with integration constant c 1, d dt Q 2(u)= 3 2 u2 +u+c 1

17 Continuity of the frst derivative, d du Q 1(u=1)= d dt Q 2(u=0) we get 1 2 =c 1 hence d dt Q 2(u)= 3 2 u2 +u+ 1 2 continue integration, Q 2 (u)= 1 2 u3 + u2 2 + u 2 +c 0 We need the contunuity of the function values, Q 1 (u=1)=q 2 (u=0) we get 1 6 =c 0 hence we get Q 2 (u)= 1 2 u3 + u2 2 + u

18 For the upper two intervals: mirror the lower polynomials by replacing u by 1 u! We do that by using an open source symbolic mathematics package called wxmaxima,

19 Q 3 (u) = Q 2 (1-u) = Calculated with Maxima Q 4 (u) = Q 1 (1-u) = Together: Q 1 (u)=1/6 u 3 ; Q 2 (u)=1/6( 3 u 3 +3u 2 +3 u+1) Q 3 (u)=1/6(3u 3 6u 2 +4) ; Q 4 (u)=1/6( u 3 +3u 2 3u+1) We now can also write our spline function h 4 (t) as

20 Q 1 (t) Q h 4 (t )={ 2 (t 1) Q 3 (t 2) Q 4 (t 3) for 0 t 1 for 1 t 2 for 2 t 3 for 3 t 4 (1) Hence our convolution of our samples f(n) with the spline function h 4 (t) for interpolation becomes f (t)= n f (n)h 4(t +2 n) Since the spline function h 4 (n) has its maximum at t=2, we shift it accordingly so that it is centered at t=n. Now we can rewrite the continuous time variable t as the sum of the lower integer index n 0 plus the foat ofset u on the unit interval, t=n 0 +u, f (n 0 +u)= n f (n)h 4 (n 0 +u+2 n) Since the spline function is zero below argument 0 and above 4, we can limit the sum to this range, f (n 0 +u)= n=n 0 1 n 0 +2 f (n)h 4 (n 0 +u+2 n) Since we can write the spline function as in eq. (4) with the four polynomials, we can rewrite the sum as

21 f (n 0 +u)=f (n 0 1) Q 4 (u)+f (n 0 ) Q 3 (u)+f (n 0 +1) Q 2 (u)+f (n 0 +2) Q 1 (u) or 3 f (n 0 +u)= f (n0 1 n) Q 4 n (u) n=0 (n) We see that we get the weighted sum of the 4 spline polynomials, each weighted by the corresponding sample value. Example: Take the sample points f ( 1)=1, f (0)=2, f (1)=3, f (2)=4. We assume the other samples values to be zero. This is shown in the following picture, f(-1) f(0) f(1) f(2) Interval with u as position of interpolation

22 The blue peaks are the original sample values, the green and red lines are the weighted and shifted B-Spline functions, the long red line is the sum of them. We want to interpolate a value at position 0 u 1 in the unit interval between sample points f(0) and f(1). We see that the sample points at both ends of this interval are connected to the peaks of the spline functions (polynomials Q 3 (u) and Q 2 (u) ), so they have the most infuence on the interpolated value, as expected. The n neighboring samples behind them still have some infuence on the interpolated value, but only via the tails of the spline function, Q 1 (u) and Q 4 (u). Hence the considered to be interpolated interval has contributions from all 4 polynomials, which appear weighted by their sample values, and add up, as in this formula, which is eq. (n) with n 0 =0 3 f (u)= n=0 f (n 1) Q 4 n (u) Plot made in Python with: import numpy as np

23 import matplotlib.pyplot as plt import scipy.signal as sig xu=np.zeros(40) xu[0:40:10]=np.arange(1,5) h1=np.ones(10) h2=sig.convolve(h1,h1)/10 h4=sig.convolve(h2,h2)/10 plt.plot(sig.convolve(xu,h4),'r') plt.plot(np.concatenate((np.zeros(20),xu))) plt.plot(1*h4,'g') plt.plot(np.concatenate((np.zeros(10),2*h4) ),'r') plt.plot(np.concatenate((np.zeros(20),3*h4)),'g') plt.plot(np.concatenate((np.zeros(30),4*h4)),'r') plt.title('spline Interpolation by Superposition') plt.show()

24 We can see that our B-Spline has the value 1/6 at sample position 1 and 3 ( Q 1 (1)=Q 2 (0)=Q 3 (1)=Q 4 (0)=1/6 ) and it has the value of 4/6 at its center at sample position n ( Q 2 (1)=Q 3 (0)=4/6 ). Hence the centered sample will be multiplied (after convolution) with the factor of 4/6, and the n neighboring samples with 1/6. This is also illustrated in the following picture 4/6 1/6 (made with Octave)

25 Here you can also see: If we have a constant sequence x(n)=k, then the interpolation on the integer grid values n are: 1/6 x(n 1)+ 4/6 x(n)+1/ 6 x(n+1)=(4/ 6+1/6+1/6) k =k So in this case the interpolation does indeed go through the original sample values! This is diferent for other examples: The interpolation of the sequence 1,n,3,4,0 with splines as an example can be seen in this picture:

26 Obtained with: import numpy as np import matplotlib.pyplot as plt import scipy.signal as sig xu=np.zeros(40) xu[0:40:10]=np.arange(1,5) h1=np.ones(10) h2=sig.convolve(h1,h1)/10 h4=sig.convolve(h2,h2)/10 plt.plot(sig.convolve(xu,h4),'r') plt.plot(np.concatenate((np.zeros(20),xu))) plt.show() (We need a shift by n intervals, because the spline function becomes maximal at t=n) Here we have an interpolated grid of 1/10 th of a pixel. We can also see that the interpolation does not go through the original pixel values, even though it is a nice and smooth interpolation. How do we obtain a spline interpolation through the original pixels? To see that, we frst need to analyse what happens with the original pixels when we interpolate with the B-Spline. Assume we already have a pre-processing of our original image, which produces the pixel

27 values c(n) out of our image pixels x(n). The question is, how do we produce the c(n) such that the original pixel values x(n) are maintained after the spline interpolation. We would like to have the original, which means we would like to have the following: x n =1/6 c n 1 4/6 c n 1/6 c n 1 Here the factor 1/6 comes from the spline value at the neighboring positions 1 and 3, and the factor 4/6 comes from the spline value at its center position n. c comes from our pre-processed or compensated image (pre-processing with some fltering operation). So here we eust look at the integer pixel positions. Observe that this condition has to hold for all inter values of n! How do we obtain c(n)? We need to apply the z-transform as a tool, because it is made for such recursive equations. Delays or advances can be formulated as multiplications with z ¹ or with z. The above equation in the z-domain is: X z =1/6 z 1 C z 4/6 C z 1/6 z C z

28 This can be re-written as: X z =C z 1/6 z 1 4/6 1/6 z and we obtain following expression for C(z): X (z) C (z)= 1/6 z 1 (1) +4/6+1/6 z Now we can see that we obtain an IIR flter for the generation of C(z) or c(n)! We can re-write the denominator polynomial of (1) as the product of n simpler polynomials using its roots. X (z) C (z)= 1/6 z 1 +4/6+1/6 z = X (z) 1/6 z(z 2 +4 z 1 +1) We can compute its roots using Octave/Matlab: roots([1,4,1]) ans = Hence we get X ( z) 1/6 z (1 z 1 z 1 )(1 z 2 z 1 ) = X (z) 1/6 z 2 (1 z 1 z 1 )(z 1 2 z 1) with z 1 = and z 2 = =1/ z 1 Observe that pole z 2 would result in an

29 instability. Observe that 1/ z 2 =z 1. We can rewrite our desired compensation flter C(z) as C (z)= 6 z 1 X ( z) (1 z 1 z 1 )(1 z 1 z) Here we obtain one causal and one anti-causal pole at z 1 = Note that we have a non-causal part z in our equation for c(n). How do we implement it? In images we can solve this problem by separating a causal and an anti-causal filter, and then apply the anti-causal flter backwards over our image, as shown in following images. The frst, causal part, is P (z)= X ( z) (1 z 1 z 1 ) For that, we flter along the causal direction, from left to rigth or from up to down, and generate the intermediate signal p(n), with an IIR flter with pole at z 1 :

30 x(n) + p(n)=x(n)+z 1 p(n-1) z 1 z ¹ z 1 p(n-1) Its transfer function in the z-domain hence becomes: P (z)= X (z)+ z 1 z 1 P(z) P (z)(1 z 1 z 1 )= X (z) P (z) X (z) = 1 1 z 1 z 1 Which is indeed our desired transfer function. We see that it has indeed a pole at position z=z 1. This now represents one half of our compensation flter. Then we apply the second part, flter the intermediate signal p(n) in the remaining anticausal direction, from right to left or from down to up, C (z)= 6 z 1 P( z) (1 z 1 z) We can implement it wit the following structure, c(n)= 6 z 1 p(n)+z 1 c(n+1)

31 p(n) + -6 z_1 c(n) z 1 z z 1 p(n+1) Its transfer function becomes (using again the z- transform): C (z)= 6 z 1 P (z)+ z 1 z C (z) C (z) (1 z 1 z)= 6 z 1 P (z) C ( z) P(z) = 6 z 1 1 z 1 z Which is indeed our desired transfer function. So the total structure is like in the following diagram,

32 x (n) p(n) + + c(n) -6z_1 z 1 z 1 Causal part z 1 z Anti-causal part This is exactly what the HHI proposal is: p(n)=x(n)+ z 1 p(n 1) c(n)=z 1 c(n+1) 6 z 1 p(n) where z_1 = -0.n6795, and where the resulting c(n) are used for the spline interpolation (p(n) is the result of the causal flter part).

33 The resulting system is here: Spline interpolation c(n) computation The block IIR flter generates the css, and the FIR flter is the spline interpolator. In-Loop processing is anti-blocking fltering and similar. (From: 28th Picture Coding Symposium, PCS2010, December 8-10, 2010, Nagoya, Japan, FRACTIONAL-SAMPLE MOTION COMPENSATION USING GENERALIZED INTERPOLATION Haricharan Lakshman, Benjamin Bross, Heiko Schwarz, and Thomas Wiegand) Pyhton Example import numpy as np import matplotlib.pyplot as plt import scipy.signal as sig

34 from scipy.signal import lflter x = np.array([0,0,1,2,3,4,0,0]) #upsampled x: xu=np.zeros(800) xu[::100]=x z1= a = np.array([1., -z1]) b = np.array([1.]) p=lflter(b, a, x,axis=0) c=-6*z1*np.fiplr([lflter([1],[1,-z1], (p[::-1]))])[0]; h1=np.ones((100)); h2=sig.convolve(h1,h1)/100; h4=sig.convolve(h2,h2)/100; print(len(h4)) cupsampled=np.zeros(100*len(c)) cupsampled[::100]=c xi=lflter(h4,1,cupsampled); plt.plot(xi,color='blue'); # defne xu plt.plot(np.hstack((np.zeros((200)),xu)),color='red')

35 plt.show() Explanation: As the signal we again take a ramp function, as in the previous example, and ist upsampled version, x = np.array([0,0,1,2,3,4,0,0]) #upsampled x: xu=np.zeros(800) xu[::100]=x we compute the signal from causal fltering as z1= a = np.array([1., -z1]) b = np.array([1.]) p=lflter(b, a, x,axis=0) We compute the anti-causal fltering by using fiplr at the input and output of our flter operation, to reverse the flter direction, c=-6*z1*np.fiplr([lflter([1],[1,-z1], (p[::-1]))])[0]; We compute our B-Spline basis function with 100 sampling points for each unit interval, h1=np.ones((100)); h2=sig.convolve(h1,h1)/100; h4=sig.convolve(h2,h2)/100; print(len(h4)) ans = 397 The length of only 397 instead of 400 samples

36 shows that this is only a more or less unprecise approximation of the B-Spline basis function, and for the reason of increasing accuracy we went to 100 samples per unit interval, instead of eust 10 in the previous example. Now we can compute the interpolated curve from the upsampled c, cupsampled=np.zeros(100*len(c)) cupsampled[::100]=c xi=lflter(h4,1,cupsampled); plt.plot(xi,color='blue'); plt.plot(np.hstack((np.zeros((200)),xu)),color='red') plt.show()

37 We can see that the interpolated curve (blue) indeed goes through the original samples (red) at full 100ss, up to the accuracy of our approximated spline function!

Lecture 10, Multirate Signal Processing Transforms as Filter Banks. Equivalent Analysis Filters of a DFT

Lecture 10, Multirate Signal Processing Transforms as Filter Banks. Equivalent Analysis Filters of a DFT Lecture 10, Multirate Signal Processing Transforms as Filter Banks Equivalent Analysis Filters of a DFT From the definitions in lecture 2 we know that a DFT of a block of signal x is defined as X (k)=

More information

Use: Analysis of systems, simple convolution, shorthand for e jw, stability. Motivation easier to write. Or X(z) = Z {x(n)}

Use: Analysis of systems, simple convolution, shorthand for e jw, stability. Motivation easier to write. Or X(z) = Z {x(n)} 1 VI. Z Transform Ch 24 Use: Analysis of systems, simple convolution, shorthand for e jw, stability. A. Definition: X(z) = x(n) z z - transforms Motivation easier to write Or Note if X(z) = Z {x(n)} z

More information

Laplace Transforms Chapter 3

Laplace Transforms Chapter 3 Laplace Transforms Important analytical method for solving linear ordinary differential equations. - Application to nonlinear ODEs? Must linearize first. Laplace transforms play a key role in important

More information

Digital Filters Ying Sun

Digital Filters Ying Sun Digital Filters Ying Sun Digital filters Finite impulse response (FIR filter: h[n] has a finite numbers of terms. Infinite impulse response (IIR filter: h[n] has infinite numbers of terms. Causal filter:

More information

Digital Signal Processing Lecture 4

Digital Signal Processing Lecture 4 Remote Sensing Laboratory Dept. of Information Engineering and Computer Science University of Trento Via Sommarive, 14, I-38123 Povo, Trento, Italy Digital Signal Processing Lecture 4 Begüm Demir E-mail:

More information

Z - Transform. It offers the techniques for digital filter design and frequency analysis of digital signals.

Z - Transform. It offers the techniques for digital filter design and frequency analysis of digital signals. Z - Transform The z-transform is a very important tool in describing and analyzing digital systems. It offers the techniques for digital filter design and frequency analysis of digital signals. Definition

More information

Lecture 2 Discrete-Time LTI Systems: Introduction

Lecture 2 Discrete-Time LTI Systems: Introduction Lecture 2 Discrete-Time LTI Systems: Introduction Outline 2.1 Classification of Systems.............................. 1 2.1.1 Memoryless................................. 1 2.1.2 Causal....................................

More information

VU Signal and Image Processing

VU Signal and Image Processing 052600 VU Signal and Image Processing Torsten Möller + Hrvoje Bogunović + Raphael Sahann torsten.moeller@univie.ac.at hrvoje.bogunovic@meduniwien.ac.at raphael.sahann@univie.ac.at vda.cs.univie.ac.at/teaching/sip/18s/

More information

III. Time Domain Analysis of systems

III. Time Domain Analysis of systems 1 III. Time Domain Analysis of systems Here, we adapt properties of continuous time systems to discrete time systems Section 2.2-2.5, pp 17-39 System Notation y(n) = T[ x(n) ] A. Types of Systems Memoryless

More information

Optimized Compact-support Interpolation Kernels

Optimized Compact-support Interpolation Kernels Optimized Compact-support Interpolation Kernels Ramtin Madani, Student Member, IEEE, Ali Ayremlou, Student Member, IEEE, Arash Amini, Farrokh Marvasti, Senior Member, IEEE, Abstract In this paper, we investigate

More information

Fourier Transform Chapter 10 Sampling and Series

Fourier Transform Chapter 10 Sampling and Series Fourier Transform Chapter 0 Sampling and Series Sampling Theorem Sampling Theorem states that, under a certain condition, it is in fact possible to recover with full accuracy the values intervening between

More information

LTI Systems (Continuous & Discrete) - Basics

LTI Systems (Continuous & Discrete) - Basics LTI Systems (Continuous & Discrete) - Basics 1. A system with an input x(t) and output y(t) is described by the relation: y(t) = t. x(t). This system is (a) linear and time-invariant (b) linear and time-varying

More information

Notes 22 largely plagiarized by %khc

Notes 22 largely plagiarized by %khc Notes 22 largely plagiarized by %khc LTv. ZT Using the conformal map z e st, we can transfer our knowledge of the ROC of the bilateral laplace transform to the ROC of the bilateral z transform. Laplace

More information

Convolution and Linear Systems

Convolution and Linear Systems CS 450: Introduction to Digital Signal and Image Processing Bryan Morse BYU Computer Science Introduction Analyzing Systems Goal: analyze a device that turns one signal into another. Notation: f (t) g(t)

More information

Bridge between continuous time and discrete time signals

Bridge between continuous time and discrete time signals 6 Sampling Bridge between continuous time and discrete time signals Sampling theorem complete representation of a continuous time signal by its samples Samplingandreconstruction implementcontinuous timesystems

More information

DIGITAL SIGNAL PROCESSING UNIT 1 SIGNALS AND SYSTEMS 1. What is a continuous and discrete time signal? Continuous time signal: A signal x(t) is said to be continuous if it is defined for all time t. Continuous

More information

Wavelets and Multiresolution Processing

Wavelets and Multiresolution Processing Wavelets and Multiresolution Processing Wavelets Fourier transform has it basis functions in sinusoids Wavelets based on small waves of varying frequency and limited duration In addition to frequency,

More information

1. SINGULARITY FUNCTIONS

1. SINGULARITY FUNCTIONS 1. SINGULARITY FUNCTIONS 1.0 INTRODUCTION Singularity functions are discontinuous functions or their derivatives are discontinuous. A singularity is a point at which a function does not possess a derivative.

More information

# FIR. [ ] = b k. # [ ]x[ n " k] [ ] = h k. x[ n] = Ae j" e j# ˆ n Complex exponential input. [ ]Ae j" e j ˆ. ˆ )Ae j# e j ˆ. y n. y n.

# FIR. [ ] = b k. # [ ]x[ n  k] [ ] = h k. x[ n] = Ae j e j# ˆ n Complex exponential input. [ ]Ae j e j ˆ. ˆ )Ae j# e j ˆ. y n. y n. [ ] = h k M [ ] = b k x[ n " k] FIR k= M [ ]x[ n " k] convolution k= x[ n] = Ae j" e j ˆ n Complex exponential input [ ] = h k M % k= [ ]Ae j" e j ˆ % M = ' h[ k]e " j ˆ & k= k = H (" ˆ )Ae j e j ˆ ( )

More information

Laplace Transforms. Chapter 3. Pierre Simon Laplace Born: 23 March 1749 in Beaumont-en-Auge, Normandy, France Died: 5 March 1827 in Paris, France

Laplace Transforms. Chapter 3. Pierre Simon Laplace Born: 23 March 1749 in Beaumont-en-Auge, Normandy, France Died: 5 March 1827 in Paris, France Pierre Simon Laplace Born: 23 March 1749 in Beaumont-en-Auge, Normandy, France Died: 5 March 1827 in Paris, France Laplace Transforms Dr. M. A. A. Shoukat Choudhury 1 Laplace Transforms Important analytical

More information

2. CONVOLUTION. Convolution sum. Response of d.t. LTI systems at a certain input signal

2. CONVOLUTION. Convolution sum. Response of d.t. LTI systems at a certain input signal 2. CONVOLUTION Convolution sum. Response of d.t. LTI systems at a certain input signal Any signal multiplied by the unit impulse = the unit impulse weighted by the value of the signal in 0: xn [ ] δ [

More information

PLC Papers. Created For:

PLC Papers. Created For: PLC Papers Created For: Algebra and proof 2 Grade 8 Objective: Use algebra to construct proofs Question 1 a) If n is a positive integer explain why the expression 2n + 1 is always an odd number. b) Use

More information

UNIT 1. SIGNALS AND SYSTEM

UNIT 1. SIGNALS AND SYSTEM Page no: 1 UNIT 1. SIGNALS AND SYSTEM INTRODUCTION A SIGNAL is defined as any physical quantity that changes with time, distance, speed, position, pressure, temperature or some other quantity. A SIGNAL

More information

ESE 531: Digital Signal Processing

ESE 531: Digital Signal Processing ESE 531: Digital Signal Processing Lec 8: February 12th, 2019 Sampling and Reconstruction Lecture Outline! Review " Ideal sampling " Frequency response of sampled signal " Reconstruction " Anti-aliasing

More information

DEPARTMENT OF ELECTRICAL AND ELECTRONIC ENGINEERING EXAMINATIONS 2010

DEPARTMENT OF ELECTRICAL AND ELECTRONIC ENGINEERING EXAMINATIONS 2010 [E2.5] IMPERIAL COLLEGE LONDON DEPARTMENT OF ELECTRICAL AND ELECTRONIC ENGINEERING EXAMINATIONS 2010 EEE/ISE PART II MEng. BEng and ACGI SIGNALS AND LINEAR SYSTEMS Time allowed: 2:00 hours There are FOUR

More information

6.003: Signals and Systems. Sampling and Quantization

6.003: Signals and Systems. Sampling and Quantization 6.003: Signals and Systems Sampling and Quantization December 1, 2009 Last Time: Sampling and Reconstruction Uniform sampling (sampling interval T ): x[n] = x(nt ) t n Impulse reconstruction: x p (t) =

More information

Sampling in 1D ( ) Continuous time signal f(t) Discrete time signal. f(t) comb

Sampling in 1D ( ) Continuous time signal f(t) Discrete time signal. f(t) comb Sampling in 2D 1 Sampling in 1D Continuous time signal f(t) Discrete time signal t ( ) f [ k] = f( kt ) = f( t) δ t kt s k s f(t) comb k 2 Nyquist theorem (1D) At least 2 sample/period are needed to represent

More information

Massachusetts Institute of Technology Department of Electrical Engineering and Computer Science. Fall Solutions for Problem Set 2

Massachusetts Institute of Technology Department of Electrical Engineering and Computer Science. Fall Solutions for Problem Set 2 Massachusetts Institute of Technology Department of Electrical Engineering and Computer Science Issued: Tuesday, September 5. 6.: Discrete-Time Signal Processing Fall 5 Solutions for Problem Set Problem.

More information

! Introduction. ! Discrete Time Signals & Systems. ! Z-Transform. ! Inverse Z-Transform. ! Sampling of Continuous Time Signals

! Introduction. ! Discrete Time Signals & Systems. ! Z-Transform. ! Inverse Z-Transform. ! Sampling of Continuous Time Signals ESE 531: Digital Signal Processing Lec 25: April 24, 2018 Review Course Content! Introduction! Discrete Time Signals & Systems! Discrete Time Fourier Transform! Z-Transform! Inverse Z-Transform! Sampling

More information

Digital Signal Processing Lecture 5

Digital Signal Processing Lecture 5 Remote Sensing Laboratory Dept. of Information Engineering and Computer Science University of Trento Via Sommarive, 14, I-38123 Povo, Trento, Italy Digital Signal Processing Lecture 5 Begüm Demir E-mail:

More information

Analog LTI system Digital LTI system

Analog LTI system Digital LTI system Sampling Decimation Seismometer Amplifier AAA filter DAA filter Analog LTI system Digital LTI system Filtering (Digital Systems) input output filter xn [ ] X ~ [ k] Convolution of Sequences hn [ ] yn [

More information

Lecture 11 FIR Filters

Lecture 11 FIR Filters Lecture 11 FIR Filters Fundamentals of Digital Signal Processing Spring, 2012 Wei-Ta Chu 2012/4/12 1 The Unit Impulse Sequence Any sequence can be represented in this way. The equation is true if k ranges

More information

UNIT - III PART A. 2. Mention any two techniques for digitizing the transfer function of an analog filter?

UNIT - III PART A. 2. Mention any two techniques for digitizing the transfer function of an analog filter? UNIT - III PART A. Mention the important features of the IIR filters? i) The physically realizable IIR filters does not have linear phase. ii) The IIR filter specification includes the desired characteristics

More information

6 The Fourier transform

6 The Fourier transform 6 The Fourier transform In this presentation we assume that the reader is already familiar with the Fourier transform. This means that we will not make a complete overview of its properties and applications.

More information

Time Series Analysis: 4. Linear filters. P. F. Góra

Time Series Analysis: 4. Linear filters. P. F. Góra Time Series Analysis: 4. Linear filters P. F. Góra http://th-www.if.uj.edu.pl/zfs/gora/ 2012 Linear filters in the Fourier domain Filtering: Multiplying the transform by a transfer function. g n DFT G

More information

The objective of this LabVIEW Mini Project was to understand the following concepts:

The objective of this LabVIEW Mini Project was to understand the following concepts: 1. Objective The objective of this LabVIEW Mini Project was to understand the following concepts: The convolution of two functions Creating LABVIEW Virtual Instruments see the visual representation of

More information

Module 4. Multi-Resolution Analysis. Version 2 ECE IIT, Kharagpur

Module 4. Multi-Resolution Analysis. Version 2 ECE IIT, Kharagpur Module 4 Multi-Resolution Analysis Lesson Multi-resolution Analysis: Discrete avelet Transforms Instructional Objectives At the end of this lesson, the students should be able to:. Define Discrete avelet

More information

Grades will be determined by the correctness of your answers (explanations are not required).

Grades will be determined by the correctness of your answers (explanations are not required). 6.00 (Fall 20) Final Examination December 9, 20 Name: Kerberos Username: Please circle your section number: Section Time 2 am pm 4 2 pm Grades will be determined by the correctness of your answers (explanations

More information

Stability of Feedback Control Systems: Absolute and Relative

Stability of Feedback Control Systems: Absolute and Relative Stability of Feedback Control Systems: Absolute and Relative Dr. Kevin Craig Greenheck Chair in Engineering Design & Professor of Mechanical Engineering Marquette University Stability: Absolute and Relative

More information

Linear Convolution Using FFT

Linear Convolution Using FFT Linear Convolution Using FFT Another useful property is that we can perform circular convolution and see how many points remain the same as those of linear convolution. When P < L and an L-point circular

More information

21.4. Engineering Applications of z-transforms. Introduction. Prerequisites. Learning Outcomes

21.4. Engineering Applications of z-transforms. Introduction. Prerequisites. Learning Outcomes Engineering Applications of z-transforms 21.4 Introduction In this Section we shall apply the basic theory of z-transforms to help us to obtain the response or output sequence for a discrete system. This

More information

QUESTION BANK SIGNALS AND SYSTEMS (4 th SEM ECE)

QUESTION BANK SIGNALS AND SYSTEMS (4 th SEM ECE) QUESTION BANK SIGNALS AND SYSTEMS (4 th SEM ECE) 1. For the signal shown in Fig. 1, find x(2t + 3). i. Fig. 1 2. What is the classification of the systems? 3. What are the Dirichlet s conditions of Fourier

More information

UNIVERSITY OF OSLO. Please make sure that your copy of the problem set is complete before you attempt to answer anything.

UNIVERSITY OF OSLO. Please make sure that your copy of the problem set is complete before you attempt to answer anything. UNIVERSITY OF OSLO Faculty of mathematics and natural sciences Examination in INF3470/4470 Digital signal processing Day of examination: December 9th, 011 Examination hours: 14.30 18.30 This problem set

More information

Z-Transform. x (n) Sampler

Z-Transform. x (n) Sampler Chapter Two A- Discrete Time Signals: The discrete time signal x(n) is obtained by taking samples of the analog signal xa (t) every Ts seconds as shown in Figure below. Analog signal Discrete time signal

More information

Lecture 19 IIR Filters

Lecture 19 IIR Filters Lecture 19 IIR Filters Fundamentals of Digital Signal Processing Spring, 2012 Wei-Ta Chu 2012/5/10 1 General IIR Difference Equation IIR system: infinite-impulse response system The most general class

More information

Digital Signal Processing:

Digital Signal Processing: Digital Signal Processing: Mathematical and algorithmic manipulation of discretized and quantized or naturally digital signals in order to extract the most relevant and pertinent information that is carried

More information

Image Filtering. Slides, adapted from. Steve Seitz and Rick Szeliski, U.Washington

Image Filtering. Slides, adapted from. Steve Seitz and Rick Szeliski, U.Washington Image Filtering Slides, adapted from Steve Seitz and Rick Szeliski, U.Washington The power of blur All is Vanity by Charles Allen Gillbert (1873-1929) Harmon LD & JuleszB (1973) The recognition of faces.

More information

CITY UNIVERSITY LONDON. MSc in Information Engineering DIGITAL SIGNAL PROCESSING EPM746

CITY UNIVERSITY LONDON. MSc in Information Engineering DIGITAL SIGNAL PROCESSING EPM746 No: CITY UNIVERSITY LONDON MSc in Information Engineering DIGITAL SIGNAL PROCESSING EPM746 Date: 19 May 2004 Time: 09:00-11:00 Attempt Three out of FIVE questions, at least One question from PART B PART

More information

Laplacian Filters. Sobel Filters. Laplacian Filters. Laplacian Filters. Laplacian Filters. Laplacian Filters

Laplacian Filters. Sobel Filters. Laplacian Filters. Laplacian Filters. Laplacian Filters. Laplacian Filters Sobel Filters Note that smoothing the image before applying a Sobel filter typically gives better results. Even thresholding the Sobel filtered image cannot usually create precise, i.e., -pixel wide, edges.

More information

z-transforms Definition of the z-transform Chapter

z-transforms Definition of the z-transform Chapter z-transforms Chapter 7 In the study of discrete-time signal and systems, we have thus far considered the time-domain and the frequency domain. The z- domain gives us a third representation. All three domains

More information

Image Acquisition and Sampling Theory

Image Acquisition and Sampling Theory Image Acquisition and Sampling Theory Electromagnetic Spectrum The wavelength required to see an object must be the same size of smaller than the object 2 Image Sensors 3 Sensor Strips 4 Digital Image

More information

MEDE2500 Tutorial Nov-7

MEDE2500 Tutorial Nov-7 (updated 2016-Nov-4,7:40pm) MEDE2500 (2016-2017) Tutorial 3 MEDE2500 Tutorial 3 2016-Nov-7 Content 1. The Dirac Delta Function, singularity functions, even and odd functions 2. The sampling process and

More information

2.004 Dynamics and Control II Spring 2008

2.004 Dynamics and Control II Spring 2008 MIT OpenCourseWare http://ocw.mit.edu 2.004 Dynamics and Control II Spring 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. 8 I * * Massachusetts

More information

Algebra I Number and Quantity The Real Number System (N-RN)

Algebra I Number and Quantity The Real Number System (N-RN) Number and Quantity The Real Number System (N-RN) Use properties of rational and irrational numbers N-RN.3 Explain why the sum or product of two rational numbers is rational; that the sum of a rational

More information

Filter Banks with Variable System Delay. Georgia Institute of Technology. Atlanta, GA Abstract

Filter Banks with Variable System Delay. Georgia Institute of Technology. Atlanta, GA Abstract A General Formulation for Modulated Perfect Reconstruction Filter Banks with Variable System Delay Gerald Schuller and Mark J T Smith Digital Signal Processing Laboratory School of Electrical Engineering

More information

Up/down-sampling & interpolation Centre for Doctoral Training in Healthcare Innovation

Up/down-sampling & interpolation Centre for Doctoral Training in Healthcare Innovation Up/down-sampling & interpolation Centre for Doctoral Training in Healthcare Innovation Dr. Gari D. Clifford, University Lecturer & Director, Centre for Doctoral Training in Healthcare Innovation, Institute

More information

EECE 301 Signals & Systems Prof. Mark Fowler

EECE 301 Signals & Systems Prof. Mark Fowler EECE 3 Signals & Systems Prof. ark Fowler Note Set #28 D-T Systems: DT Filters Ideal & Practical /4 Ideal D-T Filters Just as in the CT case we can specify filters. We looked at the ideal filter for the

More information

ESS Finite Impulse Response Filters and the Z-transform

ESS Finite Impulse Response Filters and the Z-transform 9. Finite Impulse Response Filters and the Z-transform We are going to have two lectures on filters you can find much more material in Bob Crosson s notes. In the first lecture we will focus on some of

More information

The Discrete-time Fourier Transform

The Discrete-time Fourier Transform The Discrete-time Fourier Transform Rui Wang, Assistant professor Dept. of Information and Communication Tongji University it Email: ruiwang@tongji.edu.cn Outline Representation of Aperiodic signals: The

More information

Time Series Analysis: 4. Digital Linear Filters. P. F. Góra

Time Series Analysis: 4. Digital Linear Filters. P. F. Góra Time Series Analysis: 4. Digital Linear Filters P. F. Góra http://th-www.if.uj.edu.pl/zfs/gora/ 2018 Linear filters Filtering in Fourier domain is very easy: multiply the DFT of the input by a transfer

More information

MAXIMIZING SIGNAL-TO-NOISE RATIO (SNR) IN 3-D LARGE BANDGAP SEMICONDUCTOR PIXELATED DETECTORS IN OPTIMUM AND NON-OPTIMAL FILTERING CONDITIONS

MAXIMIZING SIGNAL-TO-NOISE RATIO (SNR) IN 3-D LARGE BANDGAP SEMICONDUCTOR PIXELATED DETECTORS IN OPTIMUM AND NON-OPTIMAL FILTERING CONDITIONS 9 International Nuclear Atlantic Conference - INAC 9 Rio de Janeiro,RJ, Brazil, September7 to October, 9 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR - ABEN ISBN: 978-85-994-3-8 MAXIMIZING SIGNAL-TO-NOISE

More information

Linear Operators and Fourier Transform

Linear Operators and Fourier Transform Linear Operators and Fourier Transform DD2423 Image Analysis and Computer Vision Mårten Björkman Computational Vision and Active Perception School of Computer Science and Communication November 13, 2013

More information

Lecture 3 - Design of Digital Filters

Lecture 3 - Design of Digital Filters Lecture 3 - Design of Digital Filters 3.1 Simple filters In the previous lecture we considered the polynomial fit as a case example of designing a smoothing filter. The approximation to an ideal LPF can

More information

Determining Constant Optical Flow

Determining Constant Optical Flow Determining Constant Optical Flow Berthold K.P. Horn Copyright 003 The original optical flow algorithm [1] dealt with a flow field that could vary from place to place in the image, as would typically occur

More information

Polynomial Interpolation

Polynomial Interpolation Polynomial Interpolation (Com S 477/577 Notes) Yan-Bin Jia Sep 1, 017 1 Interpolation Problem In practice, often we can measure a physical process or quantity (e.g., temperature) at a number of points

More information

3.1 Interpolation and the Lagrange Polynomial

3.1 Interpolation and the Lagrange Polynomial MATH 4073 Chapter 3 Interpolation and Polynomial Approximation Fall 2003 1 Consider a sample x x 0 x 1 x n y y 0 y 1 y n. Can we get a function out of discrete data above that gives a reasonable estimate

More information

EE 261 The Fourier Transform and its Applications Fall 2007 Problem Set Eight Solutions

EE 261 The Fourier Transform and its Applications Fall 2007 Problem Set Eight Solutions EE 6 he Fourier ransform and its Applications Fall 7 Problem Set Eight Solutions. points) A rue Story: Professor Osgood and a graduate student were working on a discrete form of the sampling theorem. his

More information

Discrete-Time Systems

Discrete-Time Systems FIR Filters With this chapter we turn to systems as opposed to signals. The systems discussed in this chapter are finite impulse response (FIR) digital filters. The term digital filter arises because these

More information

CMSC427 Parametric curves: Hermite, Catmull-Rom, Bezier

CMSC427 Parametric curves: Hermite, Catmull-Rom, Bezier CMSC427 Parametric curves: Hermite, Catmull-Rom, Bezier Modeling Creating 3D objects How to construct complicated surfaces? Goal Specify objects with few control points Resulting object should be visually

More information

ECE 301 Division 1 Final Exam Solutions, 12/12/2011, 3:20-5:20pm in PHYS 114.

ECE 301 Division 1 Final Exam Solutions, 12/12/2011, 3:20-5:20pm in PHYS 114. ECE 301 Division 1 Final Exam Solutions, 12/12/2011, 3:20-5:20pm in PHYS 114. The exam for both sections of ECE 301 is conducted in the same room, but the problems are completely different. Your ID will

More information

EEL3135: Homework #4

EEL3135: Homework #4 EEL335: Homework #4 Problem : For each of the systems below, determine whether or not the system is () linear, () time-invariant, and (3) causal: (a) (b) (c) xn [ ] cos( 04πn) (d) xn [ ] xn [ ] xn [ 5]

More information

Low-Pass Filter _012 Triple Butterworth Filter to avoid numerical instability for high order filter. In [5]: import numpy as np import matplotlib.pyplot as plt %matplotlib notebook fsz = (7,5) # figure

More information

Module 4 MULTI- RESOLUTION ANALYSIS. Version 2 ECE IIT, Kharagpur

Module 4 MULTI- RESOLUTION ANALYSIS. Version 2 ECE IIT, Kharagpur Module MULTI- RESOLUTION ANALYSIS Version ECE IIT, Kharagpur Lesson Multi-resolution Analysis: Theory of Subband Coding Version ECE IIT, Kharagpur Instructional Objectives At the end of this lesson, the

More information

Optimized Spline Interpolation

Optimized Spline Interpolation 1 Optimized Spline Interpolation Ramtin Madani, Student Member, IEEE, Ali Ayremlou, Student Member, IEEE, Arash Amini, Farrokh Marvasti, Senior Member, IEEE, arxiv:11050011v1 [csmm] 29 Apr 2011 Abstract

More information

Conditioning and Stability

Conditioning and Stability Lab 17 Conditioning and Stability Lab Objective: Explore the condition of problems and the stability of algorithms. The condition number of a function measures how sensitive that function is to changes

More information

Review: Continuous Fourier Transform

Review: Continuous Fourier Transform Review: Continuous Fourier Transform Review: convolution x t h t = x τ h(t τ)dτ Convolution in time domain Derivation Convolution Property Interchange the order of integrals Let Convolution Property By

More information

Introduction to Digital Signal Processing

Introduction to Digital Signal Processing Introduction to Digital Signal Processing 1.1 What is DSP? DSP is a technique of performing the mathematical operations on the signals in digital domain. As real time signals are analog in nature we need

More information

Tennessee s State Mathematics Standards - Algebra I

Tennessee s State Mathematics Standards - Algebra I Domain Cluster Standards Scope and Clarifications Number and Quantity Quantities The Real (N Q) Number System (N-RN) Use properties of rational and irrational numbers Reason quantitatively and use units

More information

POLYNOMIAL-BASED INTERPOLATION FILTERS PART I: FILTER SYNTHESIS*

POLYNOMIAL-BASED INTERPOLATION FILTERS PART I: FILTER SYNTHESIS* CIRCUITS SYSTEMS SIGNAL PROCESSING c Birkhäuser Boston (27) VOL. 26, NO. 2, 27, PP. 115 146 DOI: 1.17/s34-5-74-8 POLYNOMIAL-BASED INTERPOLATION FILTERS PART I: FILTER SYNTHESIS* Jussi Vesma 1 and Tapio

More information

EE Experiment 11 The Laplace Transform and Control System Characteristics

EE Experiment 11 The Laplace Transform and Control System Characteristics EE216:11 1 EE 216 - Experiment 11 The Laplace Transform and Control System Characteristics Objectives: To illustrate computer usage in determining inverse Laplace transforms. Also to determine useful signal

More information

Complex Numbers. Visualize complex functions to estimate their zeros and poles.

Complex Numbers. Visualize complex functions to estimate their zeros and poles. Lab 1 Complex Numbers Lab Objective: Visualize complex functions to estimate their zeros and poles. Polar Representation of Complex Numbers Any complex number z = x + iy can be written in polar coordinates

More information

2.161 Signal Processing: Continuous and Discrete Fall 2008

2.161 Signal Processing: Continuous and Discrete Fall 2008 MIT OpenCourseWare http://ocw.mit.edu 2.161 Signal Processing: Continuous and Discrete Fall 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. Massachusetts

More information

Stability of Recursive Gaussian Filtering for Piecewise Linear Bilateral Filtering

Stability of Recursive Gaussian Filtering for Piecewise Linear Bilateral Filtering Stability of Recursive Gaussian Filtering for Piecewise Linear Bilateral Filtering Koichiro Watanabe, Yoshihiro Maeda, and Norishige Fukushima Nagoya Institute of Technology, Nagoya, Japan fukushima@nitech.ac.jp

More information

Digital Signal Processing, Homework 1, Spring 2013, Prof. C.D. Chung

Digital Signal Processing, Homework 1, Spring 2013, Prof. C.D. Chung Digital Signal Processing, Homework, Spring 203, Prof. C.D. Chung. (0.5%) Page 99, Problem 2.2 (a) The impulse response h [n] of an LTI system is known to be zero, except in the interval N 0 n N. The input

More information

Chap 4. Sampling of Continuous-Time Signals

Chap 4. Sampling of Continuous-Time Signals Digital Signal Processing Chap 4. Sampling of Continuous-Time Signals Chang-Su Kim Digital Processing of Continuous-Time Signals Digital processing of a CT signal involves three basic steps 1. Conversion

More information

Introduction to Computer Vision. 2D Linear Systems

Introduction to Computer Vision. 2D Linear Systems Introduction to Computer Vision D Linear Systems Review: Linear Systems We define a system as a unit that converts an input function into an output function Independent variable System operator or Transfer

More information

Mathematics. Number and Quantity The Real Number System

Mathematics. Number and Quantity The Real Number System Number and Quantity The Real Number System Extend the properties of exponents to rational exponents. 1. Explain how the definition of the meaning of rational exponents follows from extending the properties

More information

Discrete Time Signals and Systems Time-frequency Analysis. Gloria Menegaz

Discrete Time Signals and Systems Time-frequency Analysis. Gloria Menegaz Discrete Time Signals and Systems Time-frequency Analysis Gloria Menegaz Time-frequency Analysis Fourier transform (1D and 2D) Reference textbook: Discrete time signal processing, A.W. Oppenheim and R.W.

More information

NONLINEAR DIFFUSION PDES

NONLINEAR DIFFUSION PDES NONLINEAR DIFFUSION PDES Erkut Erdem Hacettepe University March 5 th, 0 CONTENTS Perona-Malik Type Nonlinear Diffusion Edge Enhancing Diffusion 5 References 7 PERONA-MALIK TYPE NONLINEAR DIFFUSION The

More information

Observations Homework Checkpoint quizzes Chapter assessments (Possibly Projects) Blocks of Algebra

Observations Homework Checkpoint quizzes Chapter assessments (Possibly Projects) Blocks of Algebra September The Building Blocks of Algebra Rates, Patterns and Problem Solving Variables and Expressions The Commutative and Associative Properties The Distributive Property Equivalent Expressions Seeing

More information

Question Paper Code : AEC11T02

Question Paper Code : AEC11T02 Hall Ticket No Question Paper Code : AEC11T02 VARDHAMAN COLLEGE OF ENGINEERING (AUTONOMOUS) Affiliated to JNTUH, Hyderabad Four Year B. Tech III Semester Tutorial Question Bank 2013-14 (Regulations: VCE-R11)

More information

Let H(z) = P(z)/Q(z) be the system function of a rational form. Let us represent both P(z) and Q(z) as polynomials of z (not z -1 )

Let H(z) = P(z)/Q(z) be the system function of a rational form. Let us represent both P(z) and Q(z) as polynomials of z (not z -1 ) Review: Poles and Zeros of Fractional Form Let H() = P()/Q() be the system function of a rational form. Let us represent both P() and Q() as polynomials of (not - ) Then Poles: the roots of Q()=0 Zeros:

More information

Tutorial Sheet #2 discrete vs. continuous functions, periodicity, sampling

Tutorial Sheet #2 discrete vs. continuous functions, periodicity, sampling 2.39 utorial Sheet #2 discrete vs. continuous functions, periodicity, sampling We will encounter two classes of signals in this class, continuous-signals and discrete-signals. he distinct mathematical

More information

MITOCW MITRES_6-007S11lec09_300k.mp4

MITOCW MITRES_6-007S11lec09_300k.mp4 MITOCW MITRES_6-007S11lec09_300k.mp4 The following content is provided under a Creative Commons license. Your support will help MIT OpenCourseWare continue to offer high quality educational resources for

More information

ESE 531: Digital Signal Processing

ESE 531: Digital Signal Processing ESE 531: Digital Signal Processing Lec 9: February 13th, 2018 Downsampling/Upsampling and Practical Interpolation Lecture Outline! CT processing of DT signals! Downsampling! Upsampling 2 Continuous-Time

More information

! Downsampling/Upsampling. ! Practical Interpolation. ! Non-integer Resampling. ! Multi-Rate Processing. " Interchanging Operations

! Downsampling/Upsampling. ! Practical Interpolation. ! Non-integer Resampling. ! Multi-Rate Processing.  Interchanging Operations Lecture Outline ESE 531: Digital Signal Processing Lec 10: February 14th, 2017 Practical and Non-integer Sampling, Multirate Sampling! Downsampling/! Practical Interpolation! Non-integer Resampling! Multi-Rate

More information

Stanford Exploration Project, Report 97, July 8, 1998, pages

Stanford Exploration Project, Report 97, July 8, 1998, pages Stanford Exploration Project, Report 97, July 8, 1998, pages 337 342 336 Stanford Exploration Project, Report 97, July 8, 1998, pages 337 342 Short Note Factorization of cross spectra Jon Claerbout 1 keywords:

More information

Filter structures ELEC-E5410

Filter structures ELEC-E5410 Filter structures ELEC-E5410 Contents FIR filter basics Ideal impulse responses Polyphase decomposition Fractional delay by polyphase structure Nyquist filters Half-band filters Gibbs phenomenon Discrete-time

More information

Lecture 16: Multiresolution Image Analysis

Lecture 16: Multiresolution Image Analysis Lecture 16: Multiresolution Image Analysis Harvey Rhody Chester F. Carlson Center for Imaging Science Rochester Institute of Technology rhody@cis.rit.edu November 9, 2004 Abstract Multiresolution analysis

More information

Complex Numbers. A complex number z = x + iy can be written in polar coordinates as re i where

Complex Numbers. A complex number z = x + iy can be written in polar coordinates as re i where Lab 20 Complex Numbers Lab Objective: Create visualizations of complex functions. Visually estimate their zeros and poles, and gain intuition about their behavior in the complex plane. Representations

More information