Computer Vision & Image Processing Project : Content Based Image Retrieval | Image Search Engine

The purpose of this project is to continue the process of learning how to manipulate and analyze images at a pixel level. In addition, this is the project where we will be doing matching, or pattern recognition.

The process can be generally described as the following four steps.

1. Compute the features Ft on the target image T
2. Compute the features {Fi} on all of the images in B
3. Compute the distance of T from all of the images in B using the distance metric D(Ft,Fi)
4. Sort the images in B according to their distance from T and return the best N matches

The entire process can be implemented as a command line program that takes in a target filename for T, a directory of images as the database B, the feature type, the matching method, and the number of images N to return. The program should print out the filenames of the top N matching images. If you're feeling fancy, you could create an OpenCV GUI that lets the user open an image and then displays the target and the top N matches.