| Summary: | 碩士 === 國立臺灣大學 === 電信工程學研究所 === 88 === The motion estimation takes an important role in video coder. The purpose of this technology is to reduce the temporal redundancy between the frame and its neighbors. So the motion estimation has a big influence on the performance of a video coding system. In this thesis, we propose a new search algorithm named Elimination Search Algorithm (ESA), which is more efficient, effective, and robust than the previously proposed fast search ones. Experimental results show that the proposed algorithm can maintain its mean square error performance which is very close to the full search algorithm while achieving an average computation reduction of 12-15 times, with respect to the full search algorithm.
In conventional block motion compensation, a motion vector is estimated for each block in a picture, and the prediction for this block is given by the block in the previous decoded picture pointed to by the motion vector. If we encode the predicted frame, however, we are not sure whether the bit rate can minimize. Now we modify the motion estimation so that the best two motion vectors can be found. After the motion compensation, we can get two predicted frames. Then encoding these two, and recording the number of bits respectively. Finally we only transmit the information about the one using fewer bits to the decoder. By this method, the less bit rate can be obtained undoubtedly, which increases some computational complexity, however.
Motion uncertainty and image quantization noise are two important factors reducing the performance of block motion compensation in a standard hybrid video coder. The loop filter smooths the block boundary that exists as a result of motion compensation and coarse quantization. We modify the conventional loop filter to solve the block boundary problem that has no filter effect on four block boundaries. The modified loop filter can achieve better rate-distortion performance than the conventional one.
|
|---|
