LOSSLESS VIDEO CODING BASED ON PROBABILITY MODEL OPTIMIZATION UTILIZING EXAMPLE SEARCH AND ADAPTIVE PREDICTION

We previously proposed a novel lossless coding method that utilizes example search and adaptive prediction within a framework of probability model optimization for still images. Additionally, we also proposed a lossless video coding method where the example search is performed on not only the current but also the previous frames to exploit intra- and inter-frame correlations. In this paper, we integrate these two methods for efficient lossless video coding. Moreover, we extend the adaptive prediction to exploit both spatial and temporal correlations simultaneously. In other words, reference pels used for the prediction are taken from both the current and the motion-compensated previous frames, and their weights, i.e. prediction coefficients, are trained pel-by-pel in a weighted least square manner. The experimental results show that the proposed method achieves better coding performance than the VVC-based lossless video coding scheme.