Research On HD Video Frame Rate Up-conversion Algorithm And Deblocking Algorithm

With the rapid development of multimedia technology, the digital TV system and display equipment have higher requirements for the quality of video signals, which makes video post-processing technology become more and more important. The frame rate up-conversion(FRUC) technique and deblocking technique are becoming indispensable for the video post-processing technologies. The FRUC technique is used to promote video frame rate, therefore it can resolve the problem of low fluency and clarity. Block-based discrete cosine transform(BDCT) technology has become the standard of image and video compression, but the decoded image occurs blocking artifacts when at a low bit rate. The deblocking technique used to remove blocking artifacts in decoded images and improve the quality of images.This thesis is focused on FRUC algorithm and deblocking algorithm. Firstly, this thesis investigates the research status at home and abroad for FRUC algorithm and deblocking algorithm. Then this thesis proposes an improved FRUC algorithm and deblocking algorithm. The main work of this thesis includes:(1) In this thesis, a frame rate up-conversion algorithm based on accurate motion vector field is proposed. The quality of interpolated frame is directly determined by the accuracy of the motion vector field. In order to get a more accurate motion vector field,this algorithm mainly contains two aspects of work. Firstly, the dual motion estimation method proposed by predecessors has been improved. This algorithm introduces high frequency information of the image into the matching criterion and the overlapped blocks are used during motion estimation to improve the ability of acquiring accurate motion vector field. Secondly, the adaptive block splitting method is applied to the boundary blocks to further improve the accuracy of the vector field. In this thesis, the traditional overlapped block motion compensation method is improved by introducing the block edge matching degree of to be compensated block vector. This method can effectively eliminate blocking artifacts in the interpolated frame and avoids blurring the interpolated frame. Experimental results verify that the proposed algorithm can effectively reduce the distortion in the interpolated frame and improve the subjective and objective quality of the interpolated frame.(2) A post-processing deblocking algorithm based on DCT domain is proposed in this thesis. Firstly, the model of blocking artifacts is built and human visual system's masking property for blocking artifacts is used to determine the visibility of the blocking artifacts. And then an accurate block partitioning method based on Canny operator edge detection is used to divide the image blocks into smooth blocks and edge blocks. For smooth blocks the coefficient related to blocking artifacts is modified and the step block is replaced by linear block to eliminate the blocking artifacts. Then the smooth blocks are further processed by the post-filtering method in the DCT domain to eliminate the possible block artifacts. For edge blocks,bilateral filter in spatial domain is used to eliminate the block artifacts and protect the image's edge information. Experimental results verify that the deblocking algorithm in this thesis can get better deblocking effect for images with different bit rates and different contents.