基于感知的右视点质量可分级编码算法

A Perception Based Algorithm of Right View Quality Scalable Video Coding

为了减少立体视频传输如此庞大的数据量,根据人眼的立体掩蔽效应,提出了基于感知的右视点质量可分级编码算法.该算法将宏块分为3类:平坦块、纹理块和边缘块,通过不同的量化步长使得右视点图像块的质量可分级.实验证明根据左视点的图像质量,右视点在人眼感知范围内码率减少25%～37%.该算法能在人眼主观察觉范围内很好地提高码率,进一步消除人眼立体视觉冗余.

Based on the masking effect of human eyes, a quality scalable video coding algorithm is proposed for right view aiming at reducing large amount of data in stereo video transmission. The algorithm is derived from the perception of human visual system, and the macro blocks are used to classify fiat blocks, edge blocks and texture. These blocks of different types have the different addition QP which can be inferred from that of the left view. As a result, the right view features in scalable quality that can be identified from one area to amother. The result shows that according to the quality of left view images, the hit rate for right view can be reduced by about 25%-37%. The proposed algorithm can well increase hit rate within the human subjective perception range and further eliminate eye stereo vision redundancy.