SIFT Based Feature Matching Algorithm for Cartoon Plagiarism Detection

Abstract

The rise of digital technology has injected new vitality into the development of the animation industry. However, the problem of copyright infringement of cartoon images has also become a major obstacle to its development. The theoretical defects of the current law, the concealment of infringement forms, and the low cost of infringement are the main reasons for this dilemma. With the rapid development of Internet information and digital image processing technology, the use, acquisition, transmission and exchange of image information has become more and more convenient. Large-scale digital images will appear on the Internet and in human life all the time. This topic intends to study the extraction process and matching process based on SIFT feature descriptors, and analyze the advantages and problems of the algorithm at the same time, and finally to propose an improvement method for the lack of color information in the SIFT algorithm. Applying the image grayscale algorithm to the first step of the SIFT algorithm, the image is first converted from the RGB color space to the HSV color space, then is calculated the chromaticity difference between adjacent pixels, and finally is performed the chromaticity difference iterative optimization to obtain the final grayscale image.