Block-correlation-based intra prediction for VVC

The new generation video coding standard Versatile Video Coding (VVC) has been officially released. Many novel technologies were utilized to improve the coding performance. In this paper, we propose an efficient intra prediction algorithm based on block correlation to improve the performance of VVC. First, since the single prediction mode sometimes does not predict very well and the high correlation of neighboring blocks is not fully used, we propose a multi-modes fusion method to merge the modes of adjacent blocks. Second, to better predict complex areas, we design an adaptive template matching method. Different weighting methods are utilized for different size blocks because they show different texture features. In addition, we combine Derived Mode (DM) with Cross-Component Linear Model (CCLM) in an adaptive way to form a new chroma mode which compensates for the shortcomings of single linear prediction. Experimental results indicated the superior performance of our algorithm. Compared with the VVC anchor (VTM 9.1), our proposed algorithm saved the bitrate of 0.78%, 0.87%, and 0.99% on average for Y, Cb, and Cr. Furthermore, the using probabilities of our designed modes were higher than that of many other traditional modes in VVC.