A CTU-Level Rate Control Algorithm for Screen Content Based on GLCM Characteristic Extraction
Zheng Liu, Chenlong He, Minge Jing, Zherong Liu, Jun Zhou, Yibo Fan · 2024
With the rapid development of interactive services such as cloud gaming and cloud desktop, the efficient compression of screen content has become increasingly important. These services offload application storage and computation to the cloud, reducing reliance on local hardware and emphasizing the need for high-quality video transmission over networks. This paper addresses the challenges in screen content coding (SCC) by proposing a novel rate control algorithm tailored for screen content, which is distinct from traditional video due to its unique characteristics like sharp text edges and repetitive graphic patterns. Our approach leverages the Grey Level Co-occurrence Matrix (GLCM) feature extraction to perform texture complexity analysis at the Coding Tree Unit (CTU) level, optimizing bit allocation based on the texture's angular second moment (ASM). We introduce a method to estimate the similarity of adjacent blocks and dynamically adjust bit allocation to account for this. The proposed method ensures low latency and stable bitrate, crucial for cloud gaming and desktop applications. Extensive experiments on the HM-16.10+SCM-7.0 and HCM-16.20+SCM-8.8 demonstrate that our approach significantly improves bitrate accuracy and compression efficiency while achieving superior subjective video quality. The algorithm's performance is benchmarked against state-of-the-art methods, showcasing its effectiveness in various screen content scenarios.