Optimized image watermarking with artificial neural networks and histogram shape

Abstract

In today’s era, Digital watermarking is a well-known application in direction of proving the authenticity of digital content. In this concern, a new optimized image watermarking method has been suggested by incorporating the concepts of artificial neural networks and histogram together. In this paper, the histogram shape concept has been kept into practice to implant the watermark information inside the host image with a view to maintain imperceptibility and resistance. In this suggested method, optimization of extraction process has been holded as a substantial problem that helps in enhancing the resistance in front of distinct attacks. So, the problem of strengthening resistance in attacked environment is solved by implementing the artificial neural networks in the proposed method. The optimization of extraction process is accomplished through two distinct neural networks called backpropagation neural networks and autoencoder neural networks. As an outcome, the experimental outcome of suggested image watermarking method is assessed on a set of three images in form of PSNR and NC. The resistance of the suggested method has been checked under different attacks alike rotation, histogram equalization, gaussian noise, cropping, JPEG compression, poisson noise, average filter, speckle noise, median filter and salt and pepper noise validates that proposed watermarking has accomplished its objective. The experimental results in the form of PSNR and NC verify that proposed scheme has accomplished its objective.