Distributed video coding(DVC)is a new paradigm for video compression

based on Wyner-Ziv’s lossy source coding theory. DVC features low-complexity video encoding and is robust against transmission errors. Common DVC loses information during the side information(SI)interpolation at the decoder. In this paper

we extract features from the adjacent frames which are used to interpolate the SI so as to predict the next bit of the reconstructed frame. The predicted bit is used to modify the probability density function(PDF)of reconstruction. The experimental results show that our method achieves higher PSNR and better subjective effects than other methods even when fewer bit planes are transmitted.