A duplicate image deduplication approach via Haar wavelet technology

Traditional deduplication technologies can only eliminate exactly the same images and are unavailable for duplicate images. In order to solve this problem, we propose a duplicate image deduplication approach based on Haar wavelet. The proposed approach employs Haar wavelet decomposition to extract feature vectors of images, and exploits the Manhattan distance of feature vectors to judge the similarity of images. If two images are similar, we extract part data from feature vectors of corresponding images to create collections, and judge whether to deduplication by the comparison between the number of same elements of different collections and the threshold. The experimental results show that the proposed approach can achieve higher deduplication ratio and deduplication accuracy by setting suitable thresholds.