Through the proposed algorithm, the features extracted from the image can be reduced to as less as 1 feature per block.
Pre-requisite to understand this project - Go through basic concepts of Hu's moments in https://en.wikipedia.org/wiki/Image_moment. Once go through the content and code uploaded in https://github.com/Tejas1415/Hu-s-Invariant-Moments-in-MATLAB.
Coded by Tejas K This was my 5th Research Paper. Publication details at the end.
Research Gap: Reducung the number of features obtained per block in an image to detect copy move forgery was always a hot topic for the researchers. Previously techniques involving Principle Component Analysis (PCA) etc were proposed on the sole purpose to reduce the number of features extracted per block. Here, we propose a novel algorithm involving Hu's invariant moments and Log polar Transforms to crub the number of features to 1 feature per block. This reduces almost 1,80,000 features for a 256 x 256 dimensional image. Imagine for 1086 x 1086p. The proposed algorithm thereby reduces the features count tremendously which inturn reduces the computtional complexity in both Memory and Time.