Performance Comparison of Several Watermarking Methods Based on ICA and AQIM

In the paper, the original image is divided into four parts by two different methods, which are taken as the observation signals in Independent Component Analysis (ICA) model. Then the watermark is embedded into DCT domain of the feature with the largest variance using iterative Adaptive Quantization Index Modulation (AQIM) and AQIM algorithms respectively. By comparing the performance of four kinds of watermark embedding methods, we show that the method based on sub-sampled ICA and iterative AQIM is better than others mentioned.