Copyright protection using KELM-PSO based multi-spectral image watermarking in DCT domain with local texture information based selection

Abstract

Since many years, researchers have been developing techniques for copyright protection. This paper presents a novel approach for copyright protection of multi-spectral images using particle swarm optimization (PSO) and kernel extreme learning machine (KELM) based watermarking technique. It begins with the selection of non-overlapping blocks of host image based on local texture information. Next, discrete cosine transform (DCT) is performed on these selected blocks, after which coefficients scanned in zig-zag manner are used to form the dataset for training and testing of KELM. KELM is applied as a regression model and predicts the output for given input vector. The watermark bits are then embedded into the output vector. PSO is a nature- inspired meta- heuristic technique used for optimization of scaling factors that control the strength of copyright logo bits being inserted into the original image. The promising results obtained by using this technique have been compared with other state-of-art approaches.