Frame-rate Interpolation Technique And Its Application In Low Bit Rate Channel

Nowadays, video transmission technology has being an important part of the network applications. Due to the large amount of data, video should be compressed to meet the transport condition. Although the video compression technology is developing rapidly and it can provide a higher compression ratio, but in some application of limited bandwidth, there is still a need for new technology to reduce the video bit-rate. Currently there are two methods to reduce video compressed bit-rate, one is reducing the input video time resolution and another is reducing spatial resolution. Lowering spatial resolution will reduce the output video quality. Lowering the frame rate can produce good video quality, but will make the output video skipping phenomenon which can be resolved by applying a frame-rate conversion technology in decoder. This paper proposes a new frame-rate conversion technology, and researches the applying in the low bit-rate channels.The paper first introduces the basic theory of the frame-rate conversion technology and the existing video coding motion estimation algorithm. And proposes a bidirectional motion estimation and adaptive overlap block motion compensation(AOBMC) frame rate conversion method which can avoid the black holes and block overlapping problems, the latter problems exist in traditional block-matching motion estimation(OBMC). The paper presents a zone diving method to solve the problem of the motion estimation’ unreliability which divides the picture into complex region and simple region and these region use the methods of VS-BMC and AOBMC respectively. The paper uses K-Means clustering analysis method to divide the area. VS-BMC algorithm uses an adaptive block size estimation strategy, which can effectively produce more accurate motion estimation of the complex motion area. Based on the conventional algorithm of OBMC, AOBMC can correct the coefficient according to the correlation between adjacent of the blocks. In addition, based on the existing camera changing detection algorithms, the paper presents a new detection algorithm which combines pixels comparison and Sobel edge detection. This method is effective and has low computational complexity.Finally, through the analysis of the video encoding rate control technology, the paper proposes an adaptive frame drop policy on the model of quadratic rate-distortion. The experiments proves under the low-rate channel applications, the method this paper improved could get better video quality than the way that increasing directly in the quantization step. So the proposed method can used in low bit rate channel which requires high video quality.