Here two images of leena are used as input. The images have one common portion which needs to be used only once. The identification of the common portion comes under image registration while identification of optimal point(column) is used as the stitching point. In the code the first image is read and saved as a new image. A small patch (a matrix of rows * 20) is used as a mask. This mask is placed on the first image. The corresponding difference of the pixels is found out, squared and added. The values are stored in a blank matrix. Once the mask covers all the pixels of the first image, the process is stopped. Now the minimum value from the 1* (columns - 20) matrix is found out. The corresponding optimal point (column value) is used as the reference for stitching. Now the second image is appended to the original image from the optimal point. This gives the resulting stitched image.
Paresh Kamble (2020). Basic image stitching code using sum of squares (https://www.mathworks.com/matlabcentral/fileexchange/40587-basic-image-stitching-code-using-sum-of-squares), MATLAB Central File Exchange. Retrieved .