Quick search algorithms based on ethnic facial image database

Abstract

The current popular image features index structure can be divided into tree-based structures, hash-based structures and machine learning based structures. In face recognition, selecting the appropriate image feature indexing structure to achieve large-scale face image matching has aways been a problem. In this paper, we present a global image features indexing method based on complete binary tree, using the ethnic facial image database, by contrast with the local sensitive hash(LSH), and principal component analysis (PCA) is adopted to extract facial image features for convienent. Experimental results show that the proposed method is superior to the local sensitive hashing in velocity.