Pixelwise adaptive prediction for lossless video compression

In this paper, we will propose a novel pixelwise adaptive prediction method for lossless video compression. The proposed method adequately takes use of the temporal and spatial redundancy in the video sequences. Firstly, the pixel to be encoded is predicted in temporal domain and spatial domain respectively, and then the final prediction values are chosen by an adaptive mode selection. In addition, we introduce a forgetting factor α in the temporal prediction and the adaptive mode selection to improve the prediction accuracy. Experimental results show that our proposed method gains the minimum zero-entropy compared to other prediction methods.