Fuzzy Rule - Based Mpeg Video

In this paper we investigated a fuzzy logic-based video rate control technique which aims to regulate the data rate of compressed video at a constant transmission rate without objectionable quality degradation. It is considered that conventional fuzzy rule-based control (FRC) does not eeectively control the two variables (video data rate and video quality) which are mutually contradictory. The primary reason for this is that it is not easy to eeectively project the control variables on to the fuzzy rules due to the contradiction on a rate-distortion theoretic basis. However, it is clear that the video quality should be considered as an equally important variable as well as the video rate. We developed a FRC scheme which also takes the video quality into account by using feed-forward scaling factors whose inputs are scene change features. The performance of this scheme was compared with the MPEG-2 Test Model (TM) 5 and the FRC scheme without the quality consideration, in terms of the buuer occupancy, the number of coded bits per frame and peak signal-to-noise ratio (PSNR). 1 INTRODUCTION Fuzzy logic and fuzzy set theory has been extensively used, particularly, for industrial and commercial control applications 1, 2] since its concept 3] was rst published. Fuzzy logic control is known to be eeective in conveying the meaning of linguistic variables to target systems 4]. Although an analogy can be found in other industrial applications, video rate control does not just fall into a matter of controlling the level of liquid reservoir which has been treated as a typical FRC application. The occupancy of video buuer has direct innuence on video quality while the liquid level is irrelevant to the output uid quality. In order to improve the FRC performance for video rate control, we introduced an adaptation induced by scaling factors in a way which they change depending on the scene change features. First, we examined the conventional FRC where the number of fuzzy variables is one, and its performance was evaluated in various settings of the fuzzy control parameters. As an adaptive algorithm, a FRC scheme with feed-forward scaling factors was designed and its performance was compared to the conventional FRC. This paper is organised as follows: Sec. 2 describes the basic FRC connguration for video rate control. Sec. 3 and Sec. 4, respectively, discuss the design process of fuzzy rules and matters of fuzzy control parameters. In Sec. 5 we present …