Compressive sensing based image watermarking using gradient descent algorithm

Abstract

The procedure for image watermarking using only available image samples, is presented in the paper. Namely, watermark is embedded into a certain number of available image samples, where the samples positions are precisely determined in advance. Therefore, samples positions serve as a watermarking key. Block based Compessive Sensing reconstruction is performed in the paper, by using gradient algorithm. Procedure is tested on several natural images. Satisfactory watermark detection, as well as good reconstructed image quality, is obtained.