Image copy-move tamper blind detection algorithm based on integrated feature vectors

Yanfen Gan and Junliu Zhong

Abstract

Copy-Move method is a simple effective tampering algorithm in digital image tamper, and regarding regional copy-move post-processing, existing detection algorithms are poor in robustness and high in time complexity. This dissertation provides a blind detection algorithm based on integrated eigenvector. This algorithm extracts from each image block the Tamura texture features and average gray-value information to make up integrated feature vectors and then sorts the feature vectors in a dictionary sorting method, and finally calculates the similarity among the image blocks by using confidence distance so as to detect and locate any tampered regions. The experimental result suggests that this algorithm can not only effectively detect and locate the tampered image regions but also effectively resist multiple post-processing of regional copy-move, including rotation, Gaussian noise addition, high/low pass filtering and JPEG compression, showing higher accuracy and lower time complexity than other algorithms.