Low-complexity Compression And Low-latency Transmission For Interactive Screen Sharing

With the fast development of the digital devices, such as PC, smart phone and smart TV, the requirement to combine multiple devices to achieve excellent user experience is increasing. Interactive screen sharing is a technology to fulfill such requirements. Some key technologies of interactive screen sharing are studied in this paper. Extremely low end-to-end latency is required in interactive screen sharing systems. Therefore, the screen compression algorithms with very low complexity, as well as the low-latency screen transmission methods are required. These two aspects are studied in this paper. In addition, the compound images spread over the World Wide Web can be considered as a special type of the computer screen. However there are some different requirements of them from the screen. Thus the compression method of such images is studied in this paper. What’s more, based on the work about interactive screen sharing, a multi-device cooperative browsing system is proposed to improve the web experience in living room.First, interactive screen sharing systems require low-complexity screen compres-sion method. This paper presents a block-based screen compression method. The iden-tical regions between two successive frames are detected for inter-frame coding, and the rest of the frame is classified as pictorial blocks and textual blocks for intra-frame coding. The block classification method utilizes the block-level statistical features. We propose a low-complexity and efficient textual block compression method, including a quantization method and an entropy-coding method. Experimental results show that this codec can encode a720P image within30ms. It achieves comparable or higher (by at most13～15dB) coding efficiency than JPEG2000and X264intra coding on typical screen images, while it achieves better visual quality. Besides, it only consumes about32%bits of X264when compressing the sequence of reading a document.Second, the compound images spread over the World Wide Web can be considered as a special type of screen content. This paper proposes a browser-friendly compound image compression method. A simple and efficient classification method is proposed to classify the blocks as pictorial blocks and textual blocks. The compound image is then s-plit into two layers, a pictorial layer and a textual layer, based on the block-classification result. The textual layer is preprocessed with proposed quantization scheme, followed by PNG as entropy coder. A joint quality control method is proposed to balance the quantization errors of the pictorial layer and the textual layer. The evaluation results show that the codec outperforms JPEG2000by at most16dB in coding efficiency, and its visual quality is better than JPEG, JPEG2000and PNG at the same time.Third, this paper studies the low-latency screen transmission method to achieve in-teractive screen sharing. A low-latency transmission framework is proposed first. Then the latency caused by different transmission error control methods in screen sharing s-cenario is analyzed. An improved ARQ method is proposed to decrease the latency. The experimental results show that the latency of the proposed method is lower than the popular screen sharing systems, such as RDP, by40%-70%.Finally, this paper proposes a multi-device cooperative browsing system based on our work about interactive screen sharing. A proxy-based thin-client browser frame-work is presented, which enables full-feature web browsing in the devices of TV and mobile browser. A touch controller based on smart phone for TV browser is then p-resented, and a scalable screen compression scheme is proposed to provide multiple screen resolutions and frame rates in single bit stream. To leverage the strength of multiple devices, a browsing session migration scheme is proposed, and the browsing states can be fully preserved. The evaluation results show that the page-loading laten-cy of the thin-client-based mobile browser is about1/4of IE Mobile, and it consumes less bandwidth and provides better visual quality against Skyfire. Also the overhead of compression efficiency and complexity from scalable coding is ignorable. Besides, the browsing session can be migrated among different devices seamlessly within0.8s, while the web browsing experience is not broken down during migration.