The Robust Algorithm of 3D Medical Image Retrieval Based on Perceptual Hashing

Abstract

In order to deal with the difficulties of the heavy workloads and ambiguity of text labeling in the traditional image retrieval and the problem that the studies on 3D medical images retrieval were lacking, this paper proposes an robust algorithm based on perceptual hashing for 3D medical images retrieval. At first, the hash value of 3D medical image to be retrieved is extracted through perceptual hashing. And then the hash value database is established. Next, the NC(Normalized Cross Correlation Coefficient, NC) between the hash value of the image to be retrieved and each one in the hash value database is computed automatically. Finally the corresponding image with the highest NC value is retrieved and the 3D medical image retrieval is realized. The results show that this algorithm can distinguish different 3D images remarkably and has ideal robustness against Guassian noise, JPEG compression, Median filtering attacks, which is obviously better than other algorithms based on DCT, DFT. In addition, this algorithm has rapid retrieval capability and good practicability. Key words-perceptual hashing; 3D medical images; image retrieval; hash value; NC