Convolutional Neural Network Decoder for Enhanced Hadamard Error Correcting Code and Its Application in Video Watermarking

Abstract

Error Correcting Codes play an essential role in the digital communication. Especially a new digital technology like video watermarking demands sufficient error correcting capabilities, because of very high compression ratio (about 1:200). Normally the watermarks can barely survive such massive attacks, despite very sophisticated embedding strategies. In this paper, the authors introduce a new approach for Error Correcting Code based on 2D Hadamard Code and convolutional Neuronal Network (CNN). The main idea is that the 2D-Hadamard code words can be represented as 2D basis images. The errors cause a noise in these basis images. The decoding procedure of this 2D-Codewords is realized by a CNN, which was before trained with these basis images. With this approach, it is possible to overcome the theoretical limit of error correcting capability of (d − 1)/2 bits, where d is a minimum Hamming distance. To prove the efficiency and practicability of this new 2D Hadamard Code, the method was applied to a video Watermarking Coding Scheme. The Video Watermarking Embedding procedure decomposes the initial video through Multi-Level Interframe Wavelet Transform. The low pass filtered part of the video stream is used for embedding the watermarks, which are protected respectively by CNN based 2D Hadamard Code.