Dr. Scott Umbaugh, SIUE 20051 Discrete Transforms.

Slides:



Advertisements
Similar presentations
Computer Vision Lecture 7: The Fourier Transform
Advertisements

Lecture 7: Basis Functions & Fourier Series
Digital Image Processing
Image Processing Lecture 4
Chapter 3 Image Enhancement in the Spatial Domain.
Fourier Transform (Chapter 4)
Frequency Domain The frequency domain
Fourier Transform – Chapter 13. Image space Cameras (regardless of wave lengths) create images in the spatial domain Pixels represent features (intensity,
Chapter Four Image Enhancement in the Frequency Domain.
Chap 4 Image Enhancement in the Frequency Domain.
Mapping: Scaling Rotation Translation Warp
School of Computing Science Simon Fraser University
HCI 530 : Seminar (HCI) Damian Schofield. HCI 530: Seminar (HCI) Transforms –Two Dimensional –Three Dimensional The Graphics Pipeline.
Chapter 8: The Discrete Fourier Transform
Digital Image Processing
Computer Graphics Recitation 6. 2 Motivation – Image compression What linear combination of 8x8 basis signals produces an 8x8 block in the image?
Digital Image Processing Chapter 4: Image Enhancement in the Frequency Domain.
Image Enhancement in the Frequency Domain Part I Image Enhancement in the Frequency Domain Part I Dr. Samir H. Abdul-Jauwad Electrical Engineering Department.
Some Properties of the 2-D Fourier Transform Translation Distributivity and Scaling Rotation Periodicity and Conjugate Symmetry Separability Convolution.
Image Enhancement.
Transforms: Basis to Basis Normal Basis Hadamard Basis Basis functions Method to find coefficients (“Transform”) Inverse Transform.
DREAM PLAN IDEA IMPLEMENTATION Introduction to Image Processing Dr. Kourosh Kiani
Digital Image Processing, 2nd ed. © 2002 R. C. Gonzalez & R. E. Woods Chapter 4 Image Enhancement in the Frequency Domain Chapter.
CH#3 Fourier Series and Transform
1 Chapter 8 The Discrete Fourier Transform 2 Introduction  In Chapters 2 and 3 we discussed the representation of sequences and LTI systems in terms.
Motivation Music as a combination of sounds at different frequencies
The Story of Wavelets.
Digital Image Processing Chapter # 4 Image Enhancement in Frequency Domain Digital Image Processing Chapter # 4 Image Enhancement in Frequency Domain.
Transforms. 5*sin (2  4t) Amplitude = 5 Frequency = 4 Hz seconds A sine wave.
Chapter 4: Image Enhancement in the Frequency Domain Chapter 4: Image Enhancement in the Frequency Domain.
Image Processing © 2002 R. C. Gonzalez & R. E. Woods Lecture 4 Image Enhancement in the Frequency Domain Lecture 4 Image Enhancement.
Image Enhancement in the Frequency Domain Spring 2006, Jen-Chang Liu.
Digital Signal Processing – Chapter 10
Module 2 SPECTRAL ANALYSIS OF COMMUNICATION SIGNAL.
1 Chapter 5 Image Transforms. 2 Image Processing for Pattern Recognition Feature Extraction Acquisition Preprocessing Classification Post Processing Scaling.
Chapter 7: The Fourier Transform 7.1 Introduction
Fourier Series. Introduction Decompose a periodic input signal into primitive periodic components. A periodic sequence T2T3T t f(t)f(t)
Part I: Image Transforms DIGITAL IMAGE PROCESSING.
1 © 2010 Cengage Learning Engineering. All Rights Reserved. 1 Introduction to Digital Image Processing with MATLAB ® Asia Edition McAndrew ‧ Wang ‧ Tseng.
Digital Image Processing Chapter 4 Image Enhancement in the Frequency Domain Part I.
Digital Image Processing, 2nd ed. © 2002 R. C. Gonzalez & R. E. Woods Background Any function that periodically repeats itself.
Lecture 7: Sampling Review of 2D Fourier Theory We view f(x,y) as a linear combination of complex exponentials that represent plane waves. F(u,v) describes.
October 29, 2013Computer Vision Lecture 13: Fourier Transform II 1 The Fourier Transform In the previous lecture, we discussed the Hough transform. There.
Digital Image Processing Chapter 4: Image Enhancement in the Frequency Domain 22 June 2005 Digital Image Processing Chapter 4: Image Enhancement in the.
CH#3 Fourier Series and Transform
7- 1 Chapter 7: Fourier Analysis Fourier analysis = Series + Transform ◎ Fourier Series -- A periodic (T) function f(x) can be written as the sum of sines.
Vibrationdata 1 Unit 6a The Fourier Transform. Vibrationdata 2 Courtesy of Professor Alan M. Nathan, University of Illinois at Urbana-Champaign.
1 CMPB 345: IMAGE PROCESSING DISCRETE TRANSFORM 2.
Fourier Transform.
CS 376b Introduction to Computer Vision 03 / 17 / 2008 Instructor: Michael Eckmann.
Computer Graphics & Image Processing Chapter # 4 Image Enhancement in Frequency Domain 2/26/20161.
2D Fourier Transform.
Chapter 13 Discrete Image Transforms
Frequency domain analysis and Fourier Transform
BYST Xform-1 DIP - WS2002: Fourier Transform Digital Image Processing Bundit Thipakorn, Ph.D. Computer Engineering Department Fourier Transform and Image.
Chapter 2. Signals and Linear Systems
Fourier transform.
The Fourier Transform.
CH#3 Fourier Series and Transform 1 st semester King Saud University College of Applied studies and Community Service 1301CT By: Nour Alhariqi.
Fourier Transform (Chapter 4) CS474/674 – Prof. Bebis.
Digital Image Processing Lecture 8: Fourier Transform Prof. Charlene Tsai.
Image Enhancement in the
Fourier Transform.
UNIT II Analysis of Continuous Time signal
ENG4BF3 Medical Image Processing
4. Image Enhancement in Frequency Domain
Chapter 8 The Discrete Fourier Transform
Digital Image Procesing Unitary Transforms Discrete Fourier Trasform (DFT) in Image Processing DR TANIA STATHAKI READER (ASSOCIATE PROFFESOR) IN SIGNAL.
Chapter 8 The Discrete Fourier Transform
Digital Image Procesing Unitary Transforms Discrete Fourier Trasform (DFT) in Image Processing DR TANIA STATHAKI READER (ASSOCIATE PROFFESOR) IN SIGNAL.
Presentation transcript:

Dr. Scott Umbaugh, SIUE Discrete Transforms

2 Introduction and Overview A transform is essentially a mathematical mapping process A transform is essentially a mathematical mapping process Used in image analysis and processing to provide information regarding the rate at which the gray levels change within an image – the spatial frequency content of an image Used in image analysis and processing to provide information regarding the rate at which the gray levels change within an image – the spatial frequency content of an image The principal component transform decorrelates multiband image data The principal component transform decorrelates multiband image data

Dr. Scott Umbaugh, SIUE The wavelet and the haar transforms retain both spatial and frequency information The wavelet and the haar transforms retain both spatial and frequency information A transform maps image data into a different mathematical space via a transformation equation A transform maps image data into a different mathematical space via a transformation equation Most of the discrete transforms map the image data from the spatial domain to the frequency domain (also called the spectral domain), where all the pixels in the input (spatial domain) contribute to each value in the output (frequency domain) Most of the discrete transforms map the image data from the spatial domain to the frequency domain (also called the spectral domain), where all the pixels in the input (spatial domain) contribute to each value in the output (frequency domain)

Dr. Scott Umbaugh, SIUE Discrete Transforms

Dr. Scott Umbaugh, SIUE These transforms are used as tools in many areas of engineering and science, including digital imaging commonly in their discrete (sampled) forms These transforms are used as tools in many areas of engineering and science, including digital imaging commonly in their discrete (sampled) forms The discrete form is created by sampling the continuous form of the functions on which these transforms are based, that is, the basis functions The discrete form is created by sampling the continuous form of the functions on which these transforms are based, that is, the basis functions Basis vectors are sampled versions of basis functions for one-dimensional (1-D) case Basis vectors are sampled versions of basis functions for one-dimensional (1-D) case

Dr. Scott Umbaugh, SIUE Basis images or basis matrices are Basis images or basis matrices are two-dimensional (2-D) versions of basis vectors The process of transforming the image data into another domain, or mathematical space, amounts to projecting the image onto the basis images The process of transforming the image data into another domain, or mathematical space, amounts to projecting the image onto the basis images The mathematical term for this projection process is an inner product The mathematical term for this projection process is an inner product Frequency transforms can be performed on the entire image or smaller blocks Frequency transforms can be performed on the entire image or smaller blocks

Dr. Scott Umbaugh, SIUE 20057

8 Spatial frequency and sequency relates to how brightness levels change relative to spatial coordinates Spatial frequency and sequency relates to how brightness levels change relative to spatial coordinates Frequency is the term for sinusoidal transforms, sequency for rectangular wave transforms Frequency is the term for sinusoidal transforms, sequency for rectangular wave transforms Rapidly changing brightness values correspond to high frequency (or sequency) terms, slowly changing brightness values correspond to low frequency (or sequency) terms Rapidly changing brightness values correspond to high frequency (or sequency) terms, slowly changing brightness values correspond to low frequency (or sequency) terms

Dr. Scott Umbaugh, SIUE 20059

10 The lowest spatial frequency, called the zero frequency term ( DC term), corresponds to an image with a constant value The lowest spatial frequency, called the zero frequency term ( DC term), corresponds to an image with a constant value The general form of the transformation equation, assuming an N x N image, is given by: The general form of the transformation equation, assuming an N x N image, is given by:

Dr. Scott Umbaugh, SIUE where u and v are the frequency domain variables, k is a constant that is transform dependent, T (u,v) are the transform coefficients, and B (r, c; u,v) correspond to the basis images The notation B (r, c; u,v) defines a set of basis images, corresponding to each different value for u and v, and the size of each is r by c The notation B (r, c; u,v) defines a set of basis images, corresponding to each different value for u and v, and the size of each is r by c

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE The transform coefficients, T (u,v), are the projections of I (r,c) onto each B (u,v) The transform coefficients, T (u,v), are the projections of I (r,c) onto each B (u,v) These coefficients tell us how similar the image is to the basis image; the more alike they are, the bigger the coefficient These coefficients tell us how similar the image is to the basis image; the more alike they are, the bigger the coefficient This transformation process amounts to decomposing the image into a weighted sum of the basis images, where the coefficients T (u,v) are the weights This transformation process amounts to decomposing the image into a weighted sum of the basis images, where the coefficients T (u,v) are the weights

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE Example Let I (r,c) = and let B (u,v;r,c) =

Dr. Scott Umbaugh, SIUE Then T (u,v)=

Dr. Scott Umbaugh, SIUE To obtain the image from the transform coefficients we apply the inverse transform equation: To obtain the image from the transform coefficients we apply the inverse transform equation:

Dr. Scott Umbaugh, SIUE Example

Dr. Scott Umbaugh, SIUE Is this correct? No, since I (r,c) = Comparing our results we see that we must multiply our answer by ¼ Comparing our results we see that we must multiply our answer by ¼ It also tells us that the transform pair, It also tells us that the transform pair, B(u,v;r,c) and B -1 (u,v;r,c) are not properly defined, we need to be able to recover our original image to have a proper transform pair

Dr. Scott Umbaugh, SIUE We can solve this by letting k’ = ¼, or by letting k = k’ = ½ We can solve this by letting k’ = ¼, or by letting k = k’ = ½ ½ will normalize the magnitude of the basis images to 1 ½ will normalize the magnitude of the basis images to 1 The magnitude of the basis vector is: The magnitude of the basis vector is: Therefore, to normalize the magnitude to 1, we need to divide by 2, or multiply by ½ Therefore, to normalize the magnitude to 1, we need to divide by 2, or multiply by ½

Dr. Scott Umbaugh, SIUE Basis images should be orthogonal and orthonormal Basis images should be orthogonal and orthonormal Orthogonal basis images  Have vector inner products equal to zero  Have nothing in common  Are uncorrelated  Remove redundant information Orthonormal basis images Are orthogonal and have magnitudes of one Are orthogonal and have magnitudes of one

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE Fourier Transform Fourier transform decomposes a complex signal into a weighted sum of a zero frequency term (the DC term which is related to the average value), and sinusoidal terms, the basis functions, where each sinusoid is a harmonic of the fundamental Fourier transform decomposes a complex signal into a weighted sum of a zero frequency term (the DC term which is related to the average value), and sinusoidal terms, the basis functions, where each sinusoid is a harmonic of the fundamental The fundamental is the basic or lowest frequency The fundamental is the basic or lowest frequency

Dr. Scott Umbaugh, SIUE Harmonics are frequency multiples of the fundamental (the fundamental is also called the first harmonic) Harmonics are frequency multiples of the fundamental (the fundamental is also called the first harmonic) Original signal can be recreated by adding the fundamental and all the harmonics, with each term weighted by its corresponding transform coefficient Original signal can be recreated by adding the fundamental and all the harmonics, with each term weighted by its corresponding transform coefficient

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE Figure (contd) Decomposing a Square Wave with a Fourier Transform CVIPtools screen capture of a square and successively adding more harmonics Across the top are the reconstructed squares with 8, 16 and then 32 harmonics Across the bottom are the corresponding Fourier transform magnitude images

Dr. Scott Umbaugh, SIUE One-dimensional continuous transform can be defined as follows: One-dimensional continuous transform can be defined as follows: The basis functions, e -j2πvc, are complex exponentials and are sinusoidal in nature The basis functions, e -j2πvc, are complex exponentials and are sinusoidal in nature Continuous Fourier transform theory assumes that the functions start at -∞ and go to +∞, so they are continuous and everywhere Continuous Fourier transform theory assumes that the functions start at -∞ and go to +∞, so they are continuous and everywhere

Dr. Scott Umbaugh, SIUE Fourier Transform Example a) The one-dimensional rectangle function b) the Fourier transform of the 1-D rectangle function

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE Fourier Transform Example (contd) c) Two-dimensional rectangle function as an image d) Magnitude of Fourier spectrum of the 2-D rectangle

Dr. Scott Umbaugh, SIUE The previous example illustrates: The previous example illustrates: 1. Continuous and infinite nature of the basis functions in the underlying theory 2. When we have a function that ends abruptly in one domain, such as the function F(c), it leads to a continuous series of decaying ripples in the other domain 3. The width of the rectangle in one domain is inversely proportional to the spacing of the ripples in the other domain

Dr. Scott Umbaugh, SIUE The One-dimensional Discrete Fourier Transform The One-dimensional Discrete Fourier Transform The equation for the one-dimensional discrete Fourier transform (DFT) is: The equation for the one-dimensional discrete Fourier transform (DFT) is:

Dr. Scott Umbaugh, SIUE The inverse DFT is given by: The inverse DFT is given by: where the F -1 notation represents the inverse transform The basis functions are sinusoidal in nature, as can be seen by Euler's identity: The basis functions are sinusoidal in nature, as can be seen by Euler's identity:

Dr. Scott Umbaugh, SIUE Putting this equation into the DFT equation by substituting θ = -2πvc/N, the one- dimensional DFT equation can be written as: Putting this equation into the DFT equation by substituting θ = -2πvc/N, the one- dimensional DFT equation can be written as: The F (v) terms can be broken down into a magnitude and phase component The F (v) terms can be broken down into a magnitude and phase component

Dr. Scott Umbaugh, SIUE  Magnitude:  Phase:  The magnitude of a sinusoid is simply its peak value, and the phase determines where the origin is, or where the sinusoid starts

Dr. Scott Umbaugh, SIUE Magnitude and phase of sinusoidal waves

Dr. Scott Umbaugh, SIUE Complex Numbers A complex number shown as a vector and expressed in rectangular form, in terms of the real, R, and imaginary components,I A complex number expressed in exponential form in terms of magnitude, M, and angle, θ Note: θ is measured from the real axis counterclockwise

Dr. Scott Umbaugh, SIUE The angle is measured from the real axis counterclockwise, so: A memory aid for evaluating e jθ

Dr. Scott Umbaugh, SIUE Example Given I(c) = [3,2,2,1], corresponding to the brightness values of one row of a digital image. Find F (v) in both rectangular form, and in exponential form

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE The Two-dimensional Discrete Fourier Transform The Two-dimensional Discrete Fourier Transform We can decompose an image into a weighted sum of 2-D sinusoidal terms We can decompose an image into a weighted sum of 2-D sinusoidal terms Equation for the 2-D discrete Fourier transform is: Equation for the 2-D discrete Fourier transform is:

Dr. Scott Umbaugh, SIUE Physical Interpretation of a Two-Dimensional Sinusoid

Dr. Scott Umbaugh, SIUE Fourier transform equation is: Fourier transform equation is: The magnitude is given by: The magnitude is given by: The phase is given by: The phase is given by:

Dr. Scott Umbaugh, SIUE Fourier Transform Phase Information a) Original imageb) Phase only image c) Contrast enhanced version of image (b) to show detail Note: Phase data contains information about where objects are in the image

Dr. Scott Umbaugh, SIUE The inverse 2-D DFT is given by: The inverse 2-D DFT is given by: where F -1 notation represents the inverse transform  This equation illustrates that the function, I(r,c), is represented by a weighted sum of the basis functions, and that the transform coefficients, F(u,v), are the weights

Dr. Scott Umbaugh, SIUE  With the inverse Fourier transform, the sign on the basis functions’ exponent is changed from -1 to +1  However, this only corresponds to the phase and not the frequency and magnitude of the basis functions

Dr. Scott Umbaugh, SIUE Separability Separability  One of important properties of the Fourier transform  Means that the two-dimensional basis image can be decomposed into two product terms where each term depends only on the rows or columns  2-D DFT can be found by successive application of two 1-D DFTs

Dr. Scott Umbaugh, SIUE  Consists of following steps: 1. Separate the basis image term (also called the transform kernel) into a product, as follows: 2. We can then write Fourier transform equation in the following form:

Dr. Scott Umbaugh, SIUE  The advantage of the separability property is that F(u, v) or I(r,c) can be obtained in two steps by successive applications of the one ‑ dimensional Fourier transform or its inverse  The equation can be expressed as:

Dr. Scott Umbaugh, SIUE where For each value of r, the expression inside the brackets is a one ‑ dimensional transform with frequency values v = 0,1,2,3,... N ‑ 1 For each value of r, the expression inside the brackets is a one ‑ dimensional transform with frequency values v = 0,1,2,3,... N ‑ 1 Hence the two ‑ dimensional function F(r, v) is obtained by taking a transform along each row of I(r,c) and multiplying the result by N Hence the two ‑ dimensional function F(r, v) is obtained by taking a transform along each row of I(r,c) and multiplying the result by N The desired result, F(u, v) is obtained by taking a transform along each column of F(r, v) The desired result, F(u, v) is obtained by taking a transform along each column of F(r, v)

Dr. Scott Umbaugh, SIUE Fourier Transform Properties Fourier Transform Properties A Fourier transform pair A Fourier transform pair  Refers to an equation in a one domain, either spatial or spectral, and its corresponding equation in the other domain  Implies that if we know what is done in one domain, we know what will occur in the other domain

Dr. Scott Umbaugh, SIUE Linearity Linearity  The Fourier transform is a linear operator and is shown by the following equations: where a and b are constants where a and b are constants

Dr. Scott Umbaugh, SIUE Convolution Convolution  Convolution in one domain is the equivalent of multiplication in the other domain  This is what allows us to perform filtering in the spatial domain with convolution masks  We use * to denote the convolution operation, and F[ ] for the forward Fourier transform and F -1 [ ] for the inverse Fourier transform

Dr. Scott Umbaugh, SIUE  The equations that define convolution property are:  It may be computationally less intensive to apply filters in the spatial domain of the image rather than the frequency domain of the image, especially if parallel hardware is available

Dr. Scott Umbaugh, SIUE Translation Translation  The translation property of the Fourier transform is given by the following equations:  These equations tell us that if the image is moved, the resulting Fourier spectrum undergoes a phase shift, but magnitude of the spectrum remains the same

Dr. Scott Umbaugh, SIUE Translation Property a) Original imageMagnitude of the Fourier spectrum of (a) Phase of the Fourier spectrum of (a) d) Original image shifted by 128 rows and 128 columns Magnitude of the Fourier spectrum of (d) Phase of the Fourier spectrum of (d)

Dr. Scott Umbaugh, SIUE Translation Property (contd) g) Original imageMagnitude of the Fourier spectrum of (g) Phase of the Fourier spectrum of (g) These images illustrate that when an image is translated, the phase changes, even though magnitude remains the same

Dr. Scott Umbaugh, SIUE Modulation Modulation  The modulation property, also called the frequency translation property, is given by:  These equations tell us that if the image is multiplied by a complex exponential (sinusoid), its corresponding spectrum is shifted

Dr. Scott Umbaugh, SIUE Modulation Property Results in Frequency Shift a) Original image b) Magnitude of the Fourier spectrum of (a) c) (a) multiplied by a vertical cosine wave at a relative frequency of 16 d) Magnitude of the Fourier spectrum of (c)

Dr. Scott Umbaugh, SIUE Rotation Rotation  The rotation property can be easily illustrated by using polar coordinates:  The Fourier transform pair I (r,c) and F (u,v) become I (x,θ) and F (w,Ø) respectively

Dr. Scott Umbaugh, SIUE  Fourier transform pair to illustrate the rotation property is as follows:  This property tell us that if an image is rotated by an angle θ 0, then F(u, v) is rotated by the same angle, and vice versa

Dr. Scott Umbaugh, SIUE Rotation Property a) Original image b) Fourier spectrum image of original image c) Original image rotated by 90 degrees d) Fourier spectrum image of rotated image Rotation results in Corresponding Rotations with Image and Spectrum

Dr. Scott Umbaugh, SIUE Periodicity Periodicity  The DFT is periodic with period N, for an N x N image. This means:. This property defines the implied symmetry in the Fourier spectrum that results from certain theoretical considerations, which have not been rigorously developed here

Dr. Scott Umbaugh, SIUE

Dr. Scott Umbaugh, SIUE Displaying the Fourier Spectrum Displaying the Fourier Spectrum The Fourier spectrum consists of complex floating point numbers, stored in CVIPtools with a data format of complex – where the image structure contains a matrix structure with one matrix for the real part and one for the imaginary part The Fourier spectrum consists of complex floating point numbers, stored in CVIPtools with a data format of complex – where the image structure contains a matrix structure with one matrix for the real part and one for the imaginary part In CVIPtools we shift the origin to the center of the image by applying the properties of periodicity and modulation with u 0 = v 0 = N/2 In CVIPtools we shift the origin to the center of the image by applying the properties of periodicity and modulation with u 0 = v 0 = N/2

Dr. Scott Umbaugh, SIUE Application of periodicity and modulation properties with u 0 = v 0 = N/2 : Application of periodicity and modulation properties with u 0 = v 0 = N/2 : In other words, we can shift the spectrum by N/2 by multiplying the original image by (-1) (r+c) which will shift the origin to the center of the image In other words, we can shift the spectrum by N/2 by multiplying the original image by (-1) (r+c) which will shift the origin to the center of the image

Dr. Scott Umbaugh, SIUE It is done in CVIPtools for the following reasons: It is done in CVIPtools for the following reasons:  Easier to understand the spectral information with the origin in the center and frequency increasing from the center out towards the edges  Makes it easier to visualize the filters  Looks better

Dr. Scott Umbaugh, SIUE The actual dynamic range of the Fourier spectrum is much greater than the 256 gray levels (8-bits) available with most image display devices The actual dynamic range of the Fourier spectrum is much greater than the 256 gray levels (8-bits) available with most image display devices Therefore, we have to remap it to 256 levels, enabling us to only see the largest values, which are typically the low frequency terms around the origin and/or terms along the u and v axis Therefore, we have to remap it to 256 levels, enabling us to only see the largest values, which are typically the low frequency terms around the origin and/or terms along the u and v axis Thus we tend to miss much of the visual information due to the limited dynamic range of display device and the human visual system’s response Thus we tend to miss much of the visual information due to the limited dynamic range of display device and the human visual system’s response

Dr. Scott Umbaugh, SIUE Direct Mapping of Fourier Magnitude Data a) Original imageb) Fourier magnitude directly remapped to without any enhancement c) Contrast enhanced versions of (b)d) Contrast enhanced versions of (b)

Dr. Scott Umbaugh, SIUE Direct Mapping of Fourier Magnitude Data (contd) f) Contrast enhanced versions of (b)e) Contrast enhanced versions of (b) Note that in (f), where we can see the most, the image is visually reduced to being either black or white, most of the dynamic range is lost

Dr. Scott Umbaugh, SIUE To take advantage of the human visual system’s response to brightness we can greatly enhance the visual information available by displaying the following log transform of the spectrum: To take advantage of the human visual system’s response to brightness we can greatly enhance the visual information available by displaying the following log transform of the spectrum: The log function compresses the data, and the scaling factor k remaps the data to the range The log function compresses the data, and the scaling factor k remaps the data to the range

Dr. Scott Umbaugh, SIUE Displaying DFT Spectrum with Various Remap Methods DIRECT REMAPCONTRAST ENHANCED LOG REMAP Cam.pgm An Ellipse

Dr. Scott Umbaugh, SIUE Displaying DFT Spectrum with Various Remap Methods (contd) DIRECT REMAPCONTRAST ENHANCED LOG REMAP House.pgm A Rectangle

Dr. Scott Umbaugh, SIUE Images of Simple Geometric Shapes and Their Fourier Spectral Images LOG REMAP Image of a Square Image of an Ellipse

Dr. Scott Umbaugh, SIUE Images of Simple Geometric Shapes and Their Fourier Spectral Images (contd) LOG REMAP Image of a Circle Image of a Small Circle

Dr. Scott Umbaugh, SIUE Images of Simple Geometric Shapes and Their Fourier Spectral Images (contd) LOG REMAP Image of a Vertical sine wave

Dr. Scott Umbaugh, SIUE Phase  Displayed primarily to illustrate phase changes  Has a range of 0 to 360 degrees, or 0 to 2π radians  Floating point data, so it has a larger dynamic range than the 256 levels typically available for display

Dr. Scott Umbaugh, SIUE Displaying Phase Shows Phase Change a) Original image Phase of the Fourier spectrum of (a) b) Original image shifted by 128 rows and 128 columns Phase of the Fourier spectrum of (b)