Efficiency enhancement of K-multi constraints paths in hybrid video coder using information based quantization method

Abstract

Optimal coding of video data depends upon the content of the frame. In a video frame, static and motion area exist. It is necessary to handle the static and motion parts separately for the optimal video coding. Doing that task firstly it is necessary to separate out the motion activity and the static portion of the frame separately with the help of some frame partitioning technique. In this paper quadtree (QT) based frame partitioning strategy is used. In the case of quad partitioning strategy on the basis of the edge threshold value motion part is represented by the small size of the block and a static portion of the frame represented by the large size of the block. In this paper on the basis of this block-based information level, an information-based variable quantization method (IV-QM) is developed. In this method, the value of the quantization parameter is fixed on the basis of information level in the block. In our suggested method, the rate-distortion algorithm was determined and tested for a variety of video sequences both fast and slow motion. Therefore the output is the testing of these various sequences for the newly proposed technique was done because of the variable R–D behavior for a variety of test sequence. The purposed technique is represented in a way that during the improvement of the rate and distortion, the estimate does not violate the multi-constraints parameters. Results are validated with the two video sequences, Mother-daughter and Foreman having different motion activity and varying frame rate 10 and 5 fps. In case of slow motion Mother-daughter video frame at frame rate 5 fps total bit budget reduced 5615–3185 bits and complexity reduced 102.57–99.23 s with acceptable degradation in PSNR 30.14 dB to 29.25 and at frame rate 10 fps bit budget reduced 4071–1949 bits and complexity reduced 56.58–55.65 s with acceptable degradation in PSNR 30.48 dB to 29.58 using proposed algorithm. Same algorithm is verified at fast motion foreman video sequence at 5 fps and found that total bit budget reduced 10,778–9152 bits and complexity reduced 76.10–75.15 s with acceptable degradation in PSNR 30.15 dB to 29.19 and at frame rate 10 fps bit budget reduced 10,666 bits to 8741 bits and complexity reduced 67.34–54.65 s with acceptable degradation in PSNR 30.17 dB to 29.02 using proposed algorithm. When the performance parameters are compared, it is concluded that bits requirement and complexity is reduced while moving from the fixed quantization parameters to variable quantization parameter and some acceptable degradation in PSNR is comes into the picture.