Study of an efficient transmission technique for motion-image media based on an optimal-frame-rate estimation

基于最优帧率估计的运动图像媒体高效传输技术研究

• 批准号: 15500059
• 负责人: YOSHIDA Toshiyuki
• 金额: $ 1.86万
• 依托单位: University of Fukui
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 2003
• 资助国家: 日本
• 起止时间: 2003 至 2004
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-15500059/
• 关键词: Motion image coding; Frame rate; Optimization; Rate control; Subjective quality; Transcoding; Frame control; Variable frame rate

Since the bit rate of a motion image is a product of the average amount of bits per second(BPF) and the number of frames per second(FPS), these two values are inversely proportional to each other under a bit rate constraint. Therefore, a proper selection of the FPS and BPF is necessary in a efficient transmission of a target motion image with a limited bandwidth. The head investigator of this research has considered this problem and has proposed a fundamental technique for estimating the optimal FPS for a target image sequence which maximizes the subjective quality of the target sequences under a fixed transmission bit rate. The objective of this research is to develop a detailed algorithm for applying the technique to an actual motion image, and to implement it as an actual motion image encoder (transcoder).We have first developed an algorithm for applying the optimal-frame-rate estimation technique to actual long motion images which consist of numbers of "scenes" having time-varying motion and frame activities. We have also considered a rate control algorithm for a constant-bit-rate encoding to develop a detailed procedure for controlling the rate for time-varying FPSs.Finally, we have implemented the developed algorithms in an MPEG4 encoder to realize a MPEG4-based transcoder, which convert an incoming MPEG4 bit stream to one with a given lower bit rate. Comparisons with a fixed FPS approach have confirmed the effectiveness of the proposed approach.

由于运动图像的比特率是平均每秒比特量（BPF）和每秒帧数（FPS）的乘积，因此这两个值在比特率约束下彼此成反比。因此，FPS和BPF的适当选择在具有有限带宽的目标运动图像的有效传输中是必要的。本研究的首席研究员已经考虑了这个问题，并提出了一个基本的技术，用于估计最佳FPS的目标图像序列，最大限度地提高主观质量的目标序列下的一个固定的传输比特率。本研究的目的是开发一个详细的算法，将该技术应用到一个实际的运动图像，并将其实现为一个实际的运动图像编码器（transcoder）。我们首先开发了一个算法，应用最佳帧速率估计技术，实际长的运动图像，其中包括数量的“场景”具有随时间变化的运动和帧活动。我们还考虑了一个恒定的比特率编码的速率控制算法，开发了一个详细的过程，为控制速率随时间变化的FPSs.Finally，我们已经实现了开发的算法在一个MPEG4编码器，实现基于MPEG4的转码器，它转换一个传入的MPEG4比特流到一个给定的较低的比特率。与固定FPS方法的比较证实了所提出的方法的有效性。