Detector performance analysis of watermark-based error concealment in image communications

In this work, the detector performance of a watermark-based error concealment (WEC) system proposed in C.B. Adsumilli is mathematically analyzed. The spread spectrum based WEC algorithm embeds a low resolution version of an image in itself during encoding. At the receiver, this embedded watermark is extracted and used as a reference to conceal the transmission losses in the reconstructed image. For the application of error concealment, the detector modelling varies from the conventional watermark detectors due to the fact that we use a full frame DCT to embed the watermark data and vary the strength of the watermark in accordance with the embedding coefficients to reduce the standard deviation of the resulting distribution. We further approximate the full frame DCT coefficients to follow a generalized Gaussian distribution (GGD) like in the case of modelling block-based DCT coefficients and consequently, we provide a comparison between the performance of this system and conventional block-based embedding strategies for various packet loss probabilities. Both analytical and simulation results show that our system of full frame DCT embedding is more effective for error concealment in detecting the embedded bits with lesser probability of error.