Study On Video Compression Algorithm

With the development of information technology, video applications are increasingly widespread in various fields of life, people demand for higher video resolution and definition. With the emergency of new generation video coding standard which named High Efficiency Video Coding (HEVC), video compression performance and video definition have been advanced. How to further improve the video coding performance and video quality of decoded side has been a focus of research in the field of video compression. In this paper, from the respective of frame rate conversion technique, we carry a series of researches on the compression of the bi-directional predicted frame in video sequence.Firstly, in order to realize rate conversion, this paper introduces critical-point filter (CPF) based image matching algorithm into image interpolation. CPF based image matching is effective in expressing varies of motion, such as translation, rotation, scaling and so on. At the same time, CPF based image matching algorithm can automatically match the images and pixel-level motion field can be obtained through it. Compared with block based image matching algorithm, CPF based image matching can achieve better performance of interpolated frame. To improve performance of interpolated frame, a refine model based on bi-directional image matching is proposed to deal with the occlusion and uncoverd problems.Secondly, this paper integrated frame rate conversion technique into bi-directional predicted frame compression. Directed frame skip coding algorithm is firstly been tested based on HM12.0framework. Experiment results show that, directed frame skip algorithm can ensure the video quality while saving bit-rate for slow motion video sequences. To solve the failure of directed frame skip algorithm for fast motion video sequences, a frame skip coding algorithm with compensation is introduced. This algorithm is effective in high bit-rate video coding. For low bit-rate video coding, block artifacts may occur in local areas when using frame skip coding with compensation algorithm. A dSSIM based rate distortion optimization algorithm is proposed to address this problem, and experimental results present that this algorithm can solve the block artifact to some content.Since different frame skip algorithms are suitable for video sequences with different bit-rate and different characteristics, this paper proposes an adaptive frame skip coding algorithm based on statistical characteristics. According to the particularity of the input video sequence, the algorithm can adaptively choose the most appropriate skipping encoding algorithm. Experimental results show that, for all the tested video sequences, the adaptive algorithm can ensure the subjective quality of the interpolation frame while reducing the bit-rate of the video sequence.