Research On Application Of GPGPU In Digital Image Processing And HEVC Video Decoding

Visual perception is the main way of understanding this world for human, digital images and videos have become an important means of transmitting information because of their vivid, intuitive and efficient features. Under the impetus of modern science and technology, the increasing demand for high resolution images and videos has produced an extremely large amount of data. The solution to process massive images and videos efficiently becomes a really urgent requirement.Compared with the traditional CPU, Graphics Processing Unit(GPU) nowadays has a powerful parallel computing capabilities, effective programming methods, efficient memory access capability, GPU is more suitable for large-scale computing tasks, General Purpose Computing on Graphics Processing Unit(GPGPU) has been widely applied in many fields. Many researches have focused on how to utilize the GPGPU in heterogeneous computing platforms to achieve high performance in image processing and video decoding algorithm. In this paper, we first give an overview of the existing problems and the possible solutions in the field of image and video processing, then outlines the GPGPU development and its application in image and video processing followed by the future development trend. Later on the most popular CUDA GPU computing architecture is introduced, including hardware model, storage model, the principle of optimization, etc. An overview of the wavelet transform and the fast algorithm of wavelet for image wavelet decomposition process is also given. The key technology of the new generation of video coding standard HEVC is summarized. After a detailed description of the realization of parallelizing two-dimensional discrete wavelet transform on CUDA GPU computing platform, the application to the image ring artifact removal is introduced and the whole ring artifact removal algorithm has also been parallelized on GPU as possible. Finally the time-consuming motion compensation module in HEVC video decoding is accelerated with GPU. The experimental results prove that the image processing, video decoding algorithms can be speeded up abundantly by using CUDA GPU.