Pulse Coupled Neural Network based Near-Duplicate Detection of Images (PCNN - NDD)

Abstract

Near Duplicate images are variants of original image with some transformations / manipulations / forgeries in it. The illegal copies of images are identified to protect copyright enforcement and reduce redundancy. The existing works in ND detection are less accurate in the identification of similar images as near duplicates. Pulse Coupled Neural Network (PCNN) is found to be a suitable processor for all the image processing techniques including feature extraction. In this paper, PCNN is applied in the detection of near duplicate (ND) images. The proposed work Pulse Coupled Neural Network based Near Duplicate Detection of Images (PCNN-NDD) is a two-step process – (1) feature extraction using PCNN and (2) fast image similarity measurement using correlation coefficient. Our system is capable of improving the accuracy effectively. The advantage of the proposed work lies in the proper setting of PCNN parameters to identify the similar images. The experimental results show that our PCNN-NDD system enhances the detection results and improves the accuracy when compared to other traditional systems.

Keywords

• computer vision
• copyright protection
• feature extraction
• image processing
• neural networks