Assignment 2: Stitching photo mosaics, The goal of the assignment is to automatically stitch images acquired by a panning camera into a mosaic as illustrated in figures 1 and 2 below.


The goal of the assignment is to automatically stitch images acquired by a panning camera into a mosaic as illustrated in figures 1 and 2 below.
Fig.1: Three images acquired by a panning camera.


img2.png
Fig.2: Images stitched to a mosaic.

Algorithm outline:

1. Choose one image as the reference frame.
2. Estimate homography between each of the remaining images and the reference image. To estimate homography between two images use the following procedure:
a. Detect local features in each image.
b. Extract feature descriptor for each feature point.
c. Match feature descriptors between two images. 
d. Robustly estimate homography using RANSAC.
3.      Wrap each image into the reference frame and composite warped images into a single mosaic.



Solution:
Complete Matlab Code Stitching photo mosaics:









mainfile code:

 % Read images
image1_color = double(imread('a.jpg'))/255;
image2_color = double(imread('b.jpg'))/255;
image3_color = double(imread('c.jpg'))/255;

% Conversaion of Color image to GrayScale
image1_gray = rgb2gray(image1_color);
image2_gray = rgb2gray(image2_color);
image3_gray = rgb2gray(image3_color);

% % show images
% figure(1); clf;
% subplot(1,3,1); imagesc(imargb); axis image; axis off; title('Image a');
% subplot(1,3,2); imagesc(imbrgb); axis image; axis off; title('Image b');
% subplot(1,3,3); imagesc(imcrgb); axis image; axis off; title('Image c');


% Harris Corner points Detector
Rthreshold = 500; % Threshold for Harris corners points
[Harris_x_image1,Harris_y_image1,strengtha] = harris(image1_gray,Rthreshold);
[Harris_x_image2,Harris_y_image2,strengthb] = harris(image2_gray,Rthreshold);
[Harris_x_image3,Harris_y_image3,strengthc] = harris(image3_gray,Rthreshold);

% Descriptors Extraction
[Descriptor_image1,Descriptor_x_image1,Descriptor_y_image1] = ext_desc(image1_gray,Harris_x_image1,Harris_y_image1);
[Descriptor_image2,Descriptor_x_image2,Descriptor_y_image2] = ext_desc(image2_gray,Harris_x_image2,Harris_y_image2);
[Descriptor_image3,Descriptor_x_image3,Descriptor_y_image3] = ext_desc(image3_gray,Harris_x_image3,Harris_y_image3);


% Computation of approximate matches between image1 and image2 by matching local features
Distance2        = dist2(Descriptor_image1',Descriptor_image2');
[Weee,Exol1]     = sort(Distance2,2);  
ratior        = Weee(:,1)./Weee(:,2);
Cope     = find(ratior<.8);  
Exol1         = Exol1(Cope);    
X_of_image1       = Descriptor_x_image1(Cope);
Y_of_image1       = Descriptor_y_image1(Cope);
X_of_image2       = Descriptor_x_image2(Exol1);
Y_of_image2       = Descriptor_y_image2(Exol1);


% Computation of approximate matches between image2 and image3 by matching local features
Distance3        = dist2(Descriptor_image3',Descriptor_image2');
[Wee1,Exol1]     = sort(Distance3,2);
ratior1        = Wee1(:,1)./Wee1(:,2);
Cope1     = find(ratior1<.8);
Exol1         = Exol1(Cope1);

X_of_image11       = Descriptor_x_image3(Cope1);
Y_of_image11       = Descriptor_y_image3(Cope1);
X_of_image21       = Descriptor_x_image2(Exol1);
Y_of_image21       = Descriptor_y_image2(Exol1);


% Threshold of inlier image1 and image2
TR              = 0.1;
[MaD, inliers] = ransacfithomography([X_of_image1; Y_of_image1], [X_of_image2; Y_of_image2], TR);
% Threshold of inlier image2 and image3
TR              = 0.1;
[MaD1, inliers1] = ransacfithomography([X_of_image11; Y_of_image11], [X_of_image21; Y_of_image21], TR);


% Combining the image
emptybox=[-400 1200 -200 700];


% Image2 is reference frame
figure(6); clf;
i2 = vgg_warp_H(image2_color, eye(3), 'linear', emptybox);
imshow(i2); axis image;

% Combining the image 1 to the reference image 2
figure(7); clf;
i1 = vgg_warp_H(image1_color, MaD, 'linear', emptybox);
imshow(i1); axis image;

% Combining the image 3 to the reference image 2
figure(8); clf;
i3 = vgg_warp_H(image3_color, MaD1, 'linear', emptybox);
imshow(i3); axis image;


figure(9); clf;
imagesc(double(max(i2,i1))); % (Homography) combining the image1 with image2 in one picture
figure(10); clf;
imagesc(double(max(i2,i3))); % (Homography) combining the image3 with image2 in one picture

% combine the 3 images in one picture
figure(11); clf;
imagesc(max(i3,double(max(i2,i1))));

axis image; axis off;


Download Complete Matlab Code Stitching photo mosaics from my google drive. 

If you face any problem while downloading code, send me email.

JawwadJLF@gmail.com

Reacties

Een reactie posten

Populaire posts van deze blog

SeedR

I am using seedr.cc website for many years. it's just amazing. I have tried many websites that were giving low speed with less space. but when I moved on seedr.cc I got 5+ Gb space even with very good speed. I strongly recommend all the people to use this. Simply just copy the link from the torrent website and paste it in seedr. You'll see your files there in some second. A very good thing that i like about it. that it instantly caught your file in their system. you don't need to wait for a long time. even you can get low seed or leches file therein seedr.cc. it's just amazing thing. Just use it. When you started to use it. you will never go anywhere. I am also a big fan of this website. they are working amazingly. even in future, they are introducing a feature, that you will be able to add selective files in seedr.cc . Thank you for reading. I have written what was best to my knowledge. 
Meer lezen