Scribd
Upload
17 views

Code

This document loads an image, converts it to grayscale, calculates image gradients, and applies four different corner detection algorithms - Harris, Triggs, Szeliski, and Shi-Tomasi - to detect corners in the image. For each algorithm, it calculates a response matrix, applies non-maximum suppression to find local maxima, thresholds to extract corner points, and plots the detected corners on the original image for comparison.

Uploaded by

Sahar Liaqat
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
17 views

Code

This document loads an image, converts it to grayscale, calculates image gradients, and applies four different corner detection algorithms - Harris, Triggs, Szeliski, and Shi-Tomasi - to detect corners in the image. For each algorithm, it calculates a response matrix, applies non-maximum suppression to find local maxima, thresholds to extract corner points, and plots the detected corners on the original image for comparison.

Uploaded by

Sahar Liaqat
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

I_rgb=imread('Image_2.

jpg');
I=rgb2gray(I_rgb);

% Using Prewitt approximation for image gradients Ix, Iy


dx=[-1 0 1;-1 0 1; -1 0 1];
dy=[1 1 1;0 0 0; -1 -1 -1];

% Image Gradients of each pixel


Ix=conv2(double(I),dx,'same');
Iy=conv2(double(I),dy,'same');

% 6x6 Gaussian Filter for Harris


sigma=1;
G=fspecial('gaussian', max(1,fix(6*sigma)),sigma);

% Structure Tensor elements


Ix2=conv2(Ix.*Ix,G,'same');
Ixy=conv2(Ix.*Iy,G,'same');
Iy2=conv2(Iy.*Iy,G,'same');

% Threshold for decision


threshold=0.36;

% Trace of Structure Tensor Matrix


trace=Ix2+Iy2;

% Determinant of Structure Tensor Matrix


det=((Ix2.*Iy2)-(Ixy.^2));

% Eigen values of Structure Tensor Matrix


Lambda0=abs(0.5*((trace)+sqrt((trace.^2)-(4*det))));
Lambda1=abs(0.5*((trace)-sqrt((trace.^2)-(4*det))));

% Harris corner detector


R1=det-(0.04*(trace.^2));
minr = min(min(R1));
maxr = max(max(R1));
R1 = (R1 - minr) / (maxr - minr);%scaling to 0-1 range
maxima1 = ordfilt2(R1, 25, ones(5));%finding maxima
mask1 = (R1 == maxima1) & (R1 > threshold);%applying threshold
maxima1 = mask1.*R1;%extracting the intensities of corner points
[r1,c1]=find(maxima1);%finding the indices of corner points
subplot(221)
imshow(I)
hold on
plot(c1,r1,'r+')
title('Harris')

% Triggs corner detector


R2=Lambda0-(0.05*Lambda1);
minr = min(min(R2));
maxr = max(max(R2));
R2 = (R2 - minr) / (maxr - minr);%scaling to 0-1 range
maxima2 = ordfilt2(R2, 25, ones(5));%finding maxima
mask2 = (R2 == maxima2) & (R2 > threshold);%applying threshold
maxima2 = mask2.*R2;%extracting the intensities of corner points
[r2,c2]=find(maxima2);%finding the indices of corner points
subplot(222)
imshow(I)
hold on
plot(c2,r2,'r+')
title('Triggs')

%Szeliski corner detector


R3 = det ./ trace;
minr = min(min(R3));
maxr = max(max(R3));
R3 = (R3 - minr) / (maxr - minr);%scaling to 0-1 range
maxima3 = ordfilt2(R3, 25, ones(5));%finding maxima
mask3 = (R3 == maxima3) & (R3 > threshold);%applying threshold
maxima3 = mask3.*R3;%extracting the intensities of corner points
[r3,c3]=find(maxima3);%finding the indices of corner points
subplot(223)
imshow(I)
hold on
plot(c3,r3,'r+')
title('Szleiski')

%Shi-Tomasi corner detector


R4=min(Lambda0,Lambda1);
minr = min(min(R4));
maxr = max(max(R4));
R4 = (R4 - minr) / (maxr - minr);%scaling to 0-1 range
maxima4 = ordfilt2(R4, 25, ones(5));%finding maxima
mask4 = (R4 == maxima4) & (R4 > threshold);%applying threshold
maxima4 = mask4.*R4;%extracting the intensities of corner points
[r4,c4]=find(maxima4);%finding the indices of corner points
subplot(224)
imshow(I)
hold on
plot(c4,r4,'r+')
title('Shi-Tomasi')

You might also like