Distortion estimation for reference frame modification methods

Due to the transmission of encoded video over error prone channels, using error resilient techniques at the encoder has become an essential issue. These techniques try to decrease the impact of transmission errors by using different approaches such as inserting Intra MacroBlocks (MBs), changing the prediction structure, or considering the channel state in selecting the best MB modes. In this work, we make use of the channel aware mode decision scheme used in the Loss Aware Rate Distortion Optimization (LARDO) method while simultaneously using the prediction structure of the Improved Generalized Source Channel Prediction (IGSCP) technique. In order to combine these two schemes, we estimate the end-to-end distortion for the IGSCP prediction structure in the H.264/AVC encoder. Simulation results, using the JSVM software, demonstrate the effectiveness of our technique for different sequences.