490 likes | 637 Vues
Introduction to Computer Vision. Lecture 10 Dr. Roger S. Gaborski. Straight Line Detection. Man made object often contain straight lines Straight lines can be detected by first filtering the image with line kernels and then linking together edges that are on a straight line
E N D
Introduction to Computer Vision Lecture 10 Dr. Roger S. Gaborski
Straight Line Detection • Man made object often contain straight lines • Straight lines can be detected by first filtering the image with line kernels and then linking together edges that are on a straight line • The Hough Transform is another method for detecting straight lines- works with EDGE DATA • Hough transforms can also detect other shapes, such as, circles Roger S. Gaborski
Consider a set of points • We would like to find a subset of the points that lie on a straight line Roger S. Gaborski
Parameter Space • Consider a particular point (xi,yi) in the x,y plane • An infinite number of lines can pass through this point • All satisfy slope-intercept equation: yi= a xi + b for some a,b • Solve for b: b = -a xi + yi • Parameter space a-b plane yields single line for (xi,yi) Roger S. Gaborski
Parameter Space • Consider a second point (xj,yj) in the x,y plane • Also has a line associated with it in parameter space • This line intersects the line associated with (xi,yi) at ( a’,b’ ) • a’ is the slope and b’ is the intercept of the line containing both (xi,yi) and (xj,yj) in the x-y space • All points on this line have lines in the parameter space that intersect at ( a’, b’ ) Roger S. Gaborski
Chapter 10 Image Segmentation Roger S. Gaborski
Parameter Space • (a, b) space usually referred to as (, ) space • Each line plots to a point in (, ) parameter space, but must satisfy: • = x1 cos + y1 sin where x1 y1 are constants Roger S. Gaborski
Parameter Space • Locus of such lines in x,y space is a sinusoid in parameter space • Any point in x,y plane corresponds to a sinusoid curve in , space Roger S. Gaborski
Chapter 10 Image Segmentation Roger S. Gaborski
( , ) Space • Define and • For each point from from edge detector • Plug in values for x and y: = x cos + ysin • For each value of in parameter space, solve for • For each pair (from previous step) record in quantized space (this is a hit) • After evaluation of all edge points, number of hits in each block corresponds to the number of pixels on the line as defined by the values Roger S. Gaborski
Hough Transform – General Idea (one point) min max min Sample of two lines (x1y1) max Two of an infinite number of lines through point (x1,y1) Space Roger S. Gaborski
Hough Transform – General Idea (three points) min max min Line through three Points (count=3) (x1y1) max Two of an infinite number of lines through point (x1,y1) Space Roger S. Gaborski
MATLAB Hough >> [H,theta,rho]=hough(f); >> figure, imshow(theta,rho,H, [ ], 'notruesize') >> axis on, axis normal; >> xlabel('\theta'),ylabel('\rho'), >> colorbar >> colormap(jet) Roger S. Gaborski
rho is 80 units Roger S. Gaborski
Peak Detection • Need to find peaks in Hough Transform • Due to quantization, pixels not in perfectly straight line, peaks lie in more than one cell • Approach: • Find Hough cell containing maximum number of entries and record location • Set immediate neighbors to zero • Look for next maximum peak, • etc Roger S. Gaborski
Find Peaks >> [H,theta,rho]=hough(f); >> figure, imshow(theta,rho,H, [ ], 'notruesize') >> axis on, axis normal; >> xlabel('\theta'),ylabel('\rho'), >> colorbar >> colormap(jet) >>[r,c]=houghpeaks(H,5); >> hold on >> plot(theta(c), rho(r), 'linestyle', 'none','marker' , ... 's', 'color', 'w') Roger S. Gaborski
Peak Roger S. Gaborski
Find Lines >> lines = houghlines(f, theta, rho, r, c) lines = point1: [100 79.5571] point2: [1 79.5571] length: 99 theta: -90 rho: -78.5571 >> figure, imshow(f), hold on >> for k = 1:length(lines) %lines is a structure xy=[lines(k).point1;lines(k).point2]; plot(xy(:,2), xy(:,1), 'LineWidth', 4, 'Color', [.6 .6 .6]); end Roger S. Gaborski
Detected Line Input image with original line Line detected Roger S. Gaborski
f Roger S. Gaborski
Hough Do not know how long line is Roger S. Gaborski
Input image f Output image Roger S. Gaborski
Crop Out Sign Region Gray Scale Image Edge Image (Canny) Roger S. Gaborski
Moose Sign – 6 Peaks Roger S. Gaborski
Hough Lines Detected ( Six) Roger S. Gaborski
Hough Line Detection Cropped image – 6 lines Roger S. Gaborski
Hough Line Detection 10 Lines Roger S. Gaborski
Chapter 10 Image Segmentation Roger S. Gaborski
Color Image Processing • RGB Color Image • M x N x 3 array of color pixels • The three planes correspond to red, green and blue values. Each pixel location is actually a triplet, r, g and b • Each plane drives the respective red, green and blue inputs of a color monitor • Data classes: • Double [0,1] • Uint8 [ 0 255] Roger S. Gaborski
Color Image imread('RedCar'); Roger S. Gaborski
Chapter 6 Color Image Processing Roger S. Gaborski
Chapter 6 Color Image Processing hsv_image = rgb2hsv(rgb_image); rgb_image = hsv2rgb(hsv_image); Roger S. Gaborski
Chapter 6 Color Image Processing Roger S. Gaborski
Extract individual color planes: • fRed = rgb_image(:,:,1); • fGreen = rgb_image(:,:,2); • fBlue = rgb_image(:,:,3); • Combine individual frames • Rgb_image = cat(3, fRed, fGreen, fBlue); Roger S. Gaborski
Image Segmentation in RGB • Partition image into regions • Regions have a ‘common feature’ • Color • Texture • Gray level • Color-based segmentation • Obtain a sample of color pixels that match region we wish to segment • Estimate the mean of the three colors • Store color values in column vector m • Classify each pixel in image as having a color specified by m Roger S. Gaborski
Classify Pixels- Euclidean Distance • We need a measure of similarity to classify pixels – how ‘close’ is a given pixel to m? • Let z denote arbitrary pixel in the image • z is similar to m if the distance between them is less than some threshold, T • Euclidean distance: D(z,m) = z-m Where x is the norm of x Roger S. Gaborski
Euclidean Metric • D(m,z) = [ (z-m)T (z-m) ]1/2 = [(zR-mR)2 + (zG-mG)2 + (zB-mB)2]1/2 D(m,z) T is a solid sphere of radius T. Points within, or on the surface are defined as being part of the segmented region Roger S. Gaborski
Mahalanobis Metric • Generalization D(z,m) = [ (z-m)T C-1 (z-m) ]1/2 where C is the covariance of the sample data • D(z,m) T is a solid 3-D elliptical body • The principal axes are orientation in direction of maximum data spread • If C = I, the Mahalanobis reduces to the Euclidean distance Roger S. Gaborski
Covariance • Variance – measure of the deviation from the mean for points in one dimension e.g. heights • Covariance as a measure of how much each of the dimensions vary from the mean with respect to each other. • Covariance is measured between 2 dimensions to see if there is a relationship between the 2 dimensions e.g. number of hours studied & grades obtained. • The covariance between one dimension and itself is the variance Roger S. Gaborski
Covariance variance (X) = Σi=1n(Xi – X) (Xi – X) (n -1) covariance (X,Y) = Σi=1n(Xi – X) (Yi – Y) (n -1) • So, if you had a 3-dimensional data set (x,y,z), then you could measure thecovariance between the x and y dimensions, the y and z dimensions, and the x and z dimensions. Measuring the covariance between x and x , or y and y , or z and z wouldgive you the variance of the x , y and z dimensions respectively. Roger S. Gaborski
Covariance • What is the interpretation of covariance calculations? e.g.: 2 dimensional data set x: number of hours studied for a subject y: grades obtained in that subject covariance value is say: 104.53 what does this value mean? Roger S. Gaborski
Covariance • Exact value is not as important as it’s sign. • A positive value of covariance indicates both dimensions increase or decrease together e.g. as the number of hours studied increases, the marks in that subject increase. • A negative value indicates while one increases the other decreases, or vice-versa e.g. active social life at RIT vs performance in CS dept. • If covariance is zero: the two dimensions are independent of each other e.g. heights of students vs the marks obtained in a subject Roger S. Gaborski
Covariance Matrix • Representing Covariance between dimensions as a matrix e.g. for 3 dimensions: cov(x,x) cov(x,y) cov(x,z) C = cov(y,x) cov(y,y) cov(y,z) cov(z,x) cov(z,y) cov(z,z) • Diagonal is the variances of x, y and z • cov(x,y) = cov(y,x) hence matrix is symmetrical about the diagonal • N-dimensional data will result in nxn covariance matrix Roger S. Gaborski
Covariance • MATLAB: cov • RG cov = GR cov, etc. • Diagonal are variances • Matrix is symmetric Roger S. Gaborski