Research On The Algorithms Of H.264-based Wireless Video Transmission And Improvement Of Video’s Quality

With the advances in the wireless communication technologies and the increase of the datatransmission rate, the demand for wireless video communication has a rapid growth. Wireless videocommunication has its inadequacies, such as the instability of the wireless channel and the mobilityof the user. So that the video communication quality always fluctuate, it is difficult to maintain thestability of the video communication quality.In view of the above problems, this thesis is to study how to improve the image quality inwireless video communication. The aim is to provide a stable and satisfactory video communicationquality as far as possible. This thesis mainly presents two solutions:(1) We propose an adaptive resource allocation method over OFDMA system for H.264SVCtransmission. Firstly we propose an improved call admission control algorithm. The maximumnumber of users the system can support can be obtained by this algorithm. Then this algorithmchooses the highest possible video resolution for each user based on the channel conditions of theuser and the power limit. This algorithm can improve the sending image quality in the existingresource conditions.(2) At the receiving end of the wireless communication system, we propose EBIBP (enhancedbilateral iterative back projection). The basic idea of this super-resolution reconstruction method isto filter the reconstruction error by enhanced bilateral filter in each iteration of the iterative backprojection. The enhanced bilateral filtering can guide the back-projection process by the directionand intensity of image’s edge. The simulation results show that this algorithm has better results. Wehave also proposed EIBP(edge-directed iterative back projection). This super-resolutionreconstruction method improves the interpolation method of reconstruction error in each iteration ofthe iterative back projection. Traditional iterative the backprojection process uses the bicubicinterpolation, but the bicubic interpolation process does not consider the direction of the edge ofinformation which leads to the serrated edges of the reconstructed image. We use the edge-directedbicubic interpolation instead of bicubic interpolation. Edge-directed bicubic interpolation detectsthe edge directions and interpolates along the detected edge directions. The reconstructed imagesare good and the running time is less than that of EBIBP.