Project 5: IMAGE WARPING and MOSAICING

Zhi Chen   cs194-26-afa

Overview

In some situations, we may not get the best pictures we want because of the limit of camera or the distances between camera and objects. In this project, we will find a way to warp pictures and blend pictures to create a mosaic image in order to solve part of this problem.

Shoot the Pictures

Before processing our mosaic algorithm, I first took pictures near or in my house as the sample pictures.

Recover Homographies

Then I find the corresponding points for all images and build the homographic relationship by following below image so that I can find the affine transformation between corresponding images.

Warp the Images and Rectify Them

Using the homographic matrix I find above, I can now transform the images into the shape that I want. Below are the rectified images I made to certify my algorithm.

Blend the images into a mosaic

By finding the corresponding points after wraping, I can know where should I blend two pictures.

Reflection

The homographic matrix is really important in this project, the matrix can not only be used to transform a image forward to get the new image, but also used to transform a target point backward with its inverse one. Only by transforming backward, we can know all values of pixels in the final wrapped image.