Department of Electrical Engineering and Computer Science Fall Semester 2011 Lecture Notes: Introduction and Overview This work is licensed under the Creative Commons Attribution-Noncommercial 2.5 License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc/2.5/ or send a letter to Creative Commons, 543 Howard Street, 5th Floor, San Francisco, California, 94105, USA. . 1 % { . { . . . . . Introduction and Overview This presentation is an overview of some of the ideas and techniques to be covered during the course. 25 August 2011 1999-2011 by Richard Alan Peters II 2 1. Image formation 2. Point processing and equalization 3. Color correction 4. The Fourier transform 5. Convolution 6. Image sampling, warping, and stitching 7. Spatial filtering 8. Noise reduction 9. Mathematical morphology 10. High dynamic range imaging 11. Image compression Topics 25 August 2011 1999-2011 by Richard Alan Peters II 3 Wallace and Gromit Wallace Gromit likes cheese reads Electronics for Dogs http://www.aardman.com/wallaceandgromit/index.shtml Wallace and Gromit will be subjects of some of the imagery in this introduction. 25 August 2011 1999-2011 by Richard Alan Peters II 4 Image Formation 25 August 2011 1999-2011 by Richard Alan Peters II 5 Image Formation 25 August 2011 1999-2011 by Richard Alan Peters II 6 Image Formation projection through lens image of object 25 August 2011 1999-2011 by Richard Alan Peters II 7 Image Formation projection onto discrete sensor array. digital camera 25 August 2011 1999-2011 by Richard Alan Peters II 8 Image Formation sensors register average color. sampled image 25 August 2011 1999-2011 by Richard Alan Peters II 9 Image Formation continuous colors, discrete locations. discrete real- valued image 25 August 2011 1999-2011 by Richard Alan Peters II 10 Digital Image Formation: Quantization continuous color input d i s c r e t e
c o l o r
o u t p u t
continuous colors mapped to a finite, discrete set of colors. 25 August 2011 1999-2011 by Richard Alan Peters II 11 Sampling and Quantization pixel grid sampled real image quantized sampled & quantized 25 August 2011 1999-2011 by Richard Alan Peters II 12 Digital Image a grid of squares, each of which contains a single color each square is called a pixel (for picture element) Color images have 3 values per pixel; monochrome images have 1 value per pixel. 25 August 2011 1999-2011 by Richard Alan Peters II 13 Color Images Are constructed from three intensity maps. Each intensity map is pro- jected through a color filter (e.g., red, green, or blue, or cyan, magenta, or yellow) to create a monochrome image. The intensity maps are overlaid to create a color image. Each pixel in a color image is a three element vector. 25 August 2011 1999-2011 by Richard Alan Peters II 14 Color Images On a CRT 25 August 2011 1999-2011 by Richard Alan Peters II 15 Point Processing original + gamma - gamma + brightness - brightness original + contrast - contrast histogram EQ histogram mod 25 August 2011 1999-2011 by Richard Alan Peters II 16 Color Processing requires some knowledge of how we see colors 25 August 2011 1999-2011 by Richard Alan Peters II 17 Eyes Light Sensors #(blue) << #(red) < #(green) cone density near fovea 25 August 2011 1999-2011 by Richard Alan Peters II 18 Color Sensing / Color Perception These are approximations of the responses to the visible spectrum of the red, green, and blue receptors of a typical human eye. 25 August 2011 1999-2011 by Richard Alan Peters II 19 These are approximations of the responses to the visible spectrum of the red, green, and blue receptors of a typical human eye. The simultaneous red + blue response causes us to perceive a continuous range of hues on a circle. No hue is greater than or less than any other hue. Color Sensing / Color Perception 25 August 2011 1999-2011 by Richard Alan Peters II 20 l u m i n a n c e
h u e
s a t u r a t i o n
photo receptors brain The eye has 3 types of photoreceptors: sensitive to red, green, or blue light. The brain transforms RGB into separate brightness and color channels (e.g., LHS). Color Sensing / Color Perception 25 August 2011 1999-2011 by Richard Alan Peters II 21 Color Perception all bands luminance chrominance red green blue 16 pixelization of: luminance and chrominance (hue+saturation) are perceived with different resolutions, as are red, green and blue. 25 August 2011 1999-2011 by Richard Alan Peters II 22 Color Perception all bands luminance chrominance red green blue 16 pixelization of: 25 August 2011 1999-2011 by Richard Alan Peters II 23 Color Balance and Saturation Uniform changes in color components result in change of tint. E.g., if all G pixel values are multiplied by > 1 then the image takes a green cast. 25 August 2011 1999-2011 by Richard Alan Peters II 24 Color Transformations 222 222 218 222 222 185 17 122 114 236 227 106 240 171 103 240 171 160 17 121 171 240 230 166 17 17 122 121 114 171 222 222 222 222 185 218 240 240 171 171 103 160 236 240 227 230 106 166 Image aging: a transformation, , that mapped: 25 August 2011 1999-2011 by Richard Alan Peters II 25 The 2D Fourier Transform of a Digital Image 2 1 1 0 0 , , , ur vc i R C R C u v I r c u v e I 1 1 2 1 0 0 ( , ) ur vc R C i R C RC r c u,v I r c e I Let I(r,c) be a single-band (intensity) digital image with R rows and C columns. Then, I(r,c) has Fourier representation where are the R x C Fourier coefficients. these complex exponentials are 2D sinusoids. 25 August 2011 1999-2011 by Richard Alan Peters II 26 2D Sinusoids: orientation ... are plane waves with grayscale amplitudes, periods in terms of lengths, ... 2 , cos cos sin 1 2 C R A c r I r c A = phase shift r c 25 August 2011 1999-2011 by Richard Alan Peters II 27 2D Sinusoids: ... specific orientations, and phase shifts. r c r c 25 August 2011 1999-2011 by Richard Alan Peters II 28 The Value of a Fourier Coefficient is a complex number with a real part and an imaginary part. If you represent that number as a magnitude, A, and a phase, , ..these represent the amplitude and offset of the sinusoid with frequency and direction . 25 August 2011 1999-2011 by Richard Alan Peters II 29 The Sinusoid from the Fourier Coeff. at (u,v) 25 August 2011 1999-2011 by Richard Alan Peters II 30 Here is the same coefficient plotted as magnitude, A, and a phase, , and displayed in the space domain as a sinusoid. I |F{I}| [F{I}] The Fourier Transform of an Image magnitude phase 25 August 2011 1999-2011 by Richard Alan Peters II 31 Continuous Fourier Transform The continuous Fourier transform assumes a continuous image exists in a finite region of an infinite plane. 2 ( ) I , , i uc vr r c u v e dudv I 2 ( ) , I , i uc vr u v r c e dcdr I The BoingBoing Bloggers 25 August 2011 1999-2011 by Richard Alan Peters II 32 Discrete Fourier Transform The discrete Fourier transform assumes a digital image exists on a closed surface, a torus. 1 1 2 0 0 I ( ) uc vr R C i C R v u r,c u,v e I 1 1 2 0 0 , I , cu rv R C i C R r c u v r c e I The BoingBoing Bloggers 25 August 2011 1999-2011 by Richard Alan Peters II 33 Convolution 16 , 16 c r 0 , 0 c r 16 , 16 c r 16 , 16 c r 16 , 16 c r Sum times 1/5 Sums of shifted and weighted copies of images or Fourier transforms. 25 August 2011 1999-2011 by Richard Alan Peters II 34 Convolution Property of the Fourier Transform The Fourier Transform of a product equals the convolution of the Fourier Transforms. Similarly, the Fourier Transform of a convolution is the product of the Fourier Transforms . by computed be can n convolutio spatial a Then, tion multiplica pointwise represents n convolutio represents . } { Moreover, . } { Then, ). , ( and ) , ( Transforms Fourier have ) , ( and ) , ( functions Let 1 G F g f G F g f G F g f v u G v u F c r g c r f - F F F 25 August 2011 1999-2011 by Richard Alan Peters II 35 Sampling, Aliasing, & Frequency Convolution aliasing (the jaggies) no aliasing (smooth lines) 25 August 2011 1999-2011 by Richard Alan Peters II 36 Sampling, Aliasing, & Frequency Convolution (a) (b) (c) (d) (a) aliased (b) power spectrum (c) unaliased (d) power spectrum 25 August 2011 1999-2011 by Richard Alan Peters II 37 8 16 Resampling nearest neighbor nearest neighbor bicubic interpolation bicubic interpolation (resizing) 25 August 2011 1999-2011 by Richard Alan Peters II 38 Rotation a n d m o t i o n b l u r
25 August 2011 1999-2011 by Richard Alan Peters II 39 Image Warping 25 August 2011 1999-2011 by Richard Alan Peters II 40 Panorama via Overlay Originals Merged *
* not so good. B r u n o
P o s t l e
h t t p : / / h u g i n . s o u r c e f o r g e . n e t / t u t o r i a l s / t w o - p h o t o s / e n . s h t m l
25 August 2011 1999-2011 by Richard Alan Peters II 41 Panorama via Stitching Originals Merged *
* much better. B r u n o
P o s t l e
h t t p : / / h u g i n . s o u r c e f o r g e . n e t / t u t o r i a l s / t w o - p h o t o s / e n . s h t m l
25 August 2011 1999-2011 by Richard Alan Peters II 42 Gaussian LPF in FD Original Image Power Spectrum Image size: 512x512 SD filter sigma = 8 Frequency Domain (FD) Filtering 25 August 2011 1999-2011 by Richard Alan Peters II 43 Original Image Filtered Image Filtered Power Spectrum Image size: 512x512 SD filter sigma = 8 FD Filtering: Lowpass 25 August 2011 1999-2011 by Richard Alan Peters II 44 Original Image Filtered Image Filtered Power Spectrum Image size: 512x512 FD notch sigma = 8 FD Filtering: Highpass 25 August 2011 1999-2011 by Richard Alan Peters II 45 Original Image Filtered Image Filtered Power Spectrum Image size: 512x512 FD notch sigma = 8 FD Filtering: Highpass signed image with 0 at middle gray 25 August 2011 1999-2011 by Richard Alan Peters II 46 original blurred sharpened Spatial Filtering 25 August 2011 1999-2011 by Richard Alan Peters II 47 Spatial Filtering bandpass filter unsharp masking original 25 August 2011 1999-2011 by Richard Alan Peters II 48 Spatial Filtering bandpass filter unsharp masking original signed image with 0 at middle gray 25 August 2011 1999-2011 by Richard Alan Peters II 49 Motion Blur vertical regional zoom rotational original 25 August 2011 1999-2011 by Richard Alan Peters II 50 color noise blurred image color-only blur Noise Reduction 25 August 2011 1999-2011 by Richard Alan Peters II 51 5x5 Wiener filter color noise blurred image Noise Reduction 25 August 2011 1999-2011 by Richard Alan Peters II 52 Noise Reduction original periodic noise frequency tuned filter 25 August 2011 1999-2011 by Richard Alan Peters II 53 Shot Noise or Salt & Pepper Noise + shot noise - shot noise s&p noise 25 August 2011 1999-2011 by Richard Alan Peters II 54 Nonlinear Filters: the Median s&p noise original median filter 25 August 2011 1999-2011 by Richard Alan Peters II 55 Nonlinear Filters: Min and Maxmin + shot noise min filter maxmin filter 25 August 2011 1999-2011 by Richard Alan Peters II 56 Nonlinear Filters: Max and Minmax - shot noise max filter minmax 25 August 2011 1999-2011 by Richard Alan Peters II 57 Nonlinear Processing: Binary Morphology L shaped SE O marks origin Foreground: white pixels Background: black pixels Cross-hatched pixels are indeterminate. 25 August 2011 1999-2011 by Richard Alan Peters II 58 Used after opening to grow back pieces of the original image that are connected to the opening. Permits the removal of small regions that are disjoint from larger objects without distorting the small features of the large objects. original opened reconstructed Nonlinear Processing: Binary Reconstruction 25 August 2011 1999-2011 by Richard Alan Peters II 59 L shaped SE O marks origin Foreground: white pixels Background: black pixels Cross-hatched pixels are indeterminate. Nonlinear Processing: Grayscale Morphology 25 August 2011 1999-2011 by Richard Alan Peters II 60 Grayscale Morphology: Opening opening: erosion then dilation opened & original 25 August 2011 1999-2011 by Richard Alan Peters II 61 Grayscale Morphology: Opening erosion & opening erosion & opening & original 25 August 2011 1999-2011 by Richard Alan Peters II 62 reconstructed opening original Nonlinear Processing: Grayscale Reconstruction 25 August 2011 1999-2011 by Richard Alan Peters II 63 Forensic Analysis of Photographs Photographs by Robert Fenton of a battlefield in the Crimean war taken on 23 April 1855. From Morris, Errol, Which Came First, the Chicken or the Egg?, Parts 1-3, New York Times, Zoom Editorial Section, 25 Sept. 2007 (pt.1), 7 Oct. 2007 (pt.2), 30 Oct. 2007 (pt.3). Which came first? 25 August 2011 1999-2011 by Richard Alan Peters II 64 Photographs by Robert Fenton of a battlefield in the Crimean war taken on 23 April 1855. From Morris, Errol, Which Came First, the Chicken or the Egg?, Parts 1-3, New York Times, Zoom Editorial Section, 25 Sept. 2007 (pt.1), 7 Oct. 2007 (pt.2), 30 Oct. 2007 (pt.3). Which came first? Forensic Analysis of Photographs 25 August 2011 1999-2011 by Richard Alan Peters II 65 Photographs by Robert Fenton of a battlefield in the Crimean war taken on 23 April 1855. From Morris, Errol, Which Came First, the Chicken or the Egg?, Parts 1-3, New York Times, Zoom Editorial Section, 25 Sept. 2007 (pt.1), 7 Oct. 2007 (pt.2), 30 Oct. 2007 (pt.3). Forensic Analysis of Photographs Which came first? 25 August 2011 1999-2011 by Richard Alan Peters II 66 High Dynamic Range (HDR) Imaging under exposed B a r t l o m i e j
O k o n e k
h t t p : / / w w w . e a s y h d r . c o m / e x a m p l e s . p h p
25 August 2011 1999-2011 by Richard Alan Peters II 67 High Dynamic Range (HDR) Imaging default exposure B a r t l o m i e j
O k o n e k
h t t p : / / w w w . e a s y h d r . c o m / e x a m p l e s . p h p
25 August 2011 1999-2011 by Richard Alan Peters II 68 High Dynamic Range (HDR) Imaging over exposed B a r t l o m i e j
O k o n e k
h t t p : / / w w w . e a s y h d r . c o m / e x a m p l e s . p h p
25 August 2011 1999-2011 by Richard Alan Peters II 69 High Dynamic Range (HDR) Imaging combined B a r t l o m i e j
O k o n e k
h t t p : / / w w w . e a s y h d r . c o m / e x a m p l e s . p h p
25 August 2011 1999-2011 by Richard Alan Peters II 70 Image Compression Yoyogi Park, Tokyo, October 1999. Photo by Alan Peters. Original image is 5244w x 4716h @ 1200 ppi: 127MBytes 25 August 2011 1999-2011 by Richard Alan Peters II 71 Image Compression: JPEG J P E G
q u a l i t y
l e v e l
F i l e
s i z e
i n
b y t e s
25 August 2011 1999-2011 by Richard Alan Peters II 72 J P E G
q u a l i t y
l e v e l
F i l e
s i z e
i n
b y t e s
Image Compression: JPEG 25 August 2011 1999-2011 by Richard Alan Peters II 73 Image Compositing Combine parts from separate images to form a new image. Its difficult to do well. Requires relative positions, orientations, and scales to be correct. Lighting of objects must be consistent within the separate images. Brightness, contrast, color balance, and saturation must match. Noise color, amplitude, and patterns must be seamless. 25 August 2011 1999-2011 by Richard Alan Peters II 74 Prof. Peters in his home office. Needs a better shirt. Image Compositing Example 25 August 2011 1999-2011 by Richard Alan Peters II 75 This shirt demands a monogram. Image Compositing Example 25 August 2011 1999-2011 by Richard Alan Peters II 76 He needs some more color. Image Compositing Example 25 August 2011 1999-2011 by Richard Alan Peters II 77 Nice. Now for the way hed wear his hair if he had any. Image Compositing Example 25 August 2011 1999-2011 by Richard Alan Peters II 78 He cant stay in the office like this. Image Compositing Example 25 August 2011 1999-2011 by Richard Alan Peters II 79 Wheres a hepcat Daddy-O like this belong? Image Compositing Example 25 August 2011 1999-2011 by Richard Alan Peters II 80 In the studio! Collar this jive, Jackson. Like crazy, Man ! Image Compositing Example 25 August 2011 1999-2011 by Richard Alan Peters II 81