An embedding algorithm for small payload using convolutional codes

A matrix embedding (ME) code was developed as a commonly used steganography technique by taking advantage of linear block codes to improve the embedding efficiency. However, in most cases, such as the linear block codes with sufficiently large dimension, the main common disadvantage is that the maximum likelihood decoding cannot be performed because of the high complexity required in ME codes with sufficiently large length. A novel family of embedding codes, generated by the systematic convolutional codes, is presented in this paper. Through a puncturing technique, a convolutional code at a high coding rate can be easily converted into an embedding code with low embedding rate. The proposed method employs the Viterbi decoding algorithm for embedding to identify the coset leader of convolutional codes for small payloads. Experimental results show that the embedding efficiency of the proposed scheme using convolutional codes with puncturing techniques is substantially superior to that of the scheme using linear block codes.