Hybrid Hashing Method for Similar Vehicle Image Search

Abstract

The novel hybrid method of a hash image calculation that can be applied in a search for similar vehicle images is proposed in this paper. The main novelty of the method described herein is the combination of two hashing types: the visual and semantic hash of the image. The method is based on SIFT and DCT algorithms. We use frontal vehicle images to test the method accuracy. The experimental results indicate that the proposed algorithm has the practical application of image search in the vehicle identification systems based on license plate recognition. We show that method is a novel in this area. The proposed method is also applicable for use in other problem domains.