Fast Algorithm for VVC Video Image Coding Based on Machine Learning

VVC (Versatile Video Coding) encoding has achieved outstanding results in terms of encoding performance, with a bitrate saving of around 50% compared to previous versions at the same image quality. However, it is complicated and time-consuming, which makes it difficult to be popularized. In this paper, we propose a convolutional neural network for the most time-consuming block segmentation stage, which directly predicts the optimal segmentation of the CU blocks of images in a video sequence during the coding process. After the training of tens of thousands of CUs, the final prediction accuracy can be up to 90% while maintaining the quality of the final video.