论文部分内容阅读
用 H.2 63标准对低码率图像编码时 ,编码器的复杂性主要取决于运动估计、离散余弦变换等运算 ,为了减少编码器的运算量、提高编码速度 ,提出一种有效的频域块匹配 +预判全零系数的运动估计算法。这种算法充分利用离散余弦变换的特点 ,在频域内使用块匹配法进行运动估计 ,并将预判全零系数和运动搜索相结合。实验表明该算法不仅可以有效地提高编码速度 ,而且可以较好地保持 H.2 63高压缩比及运动图像的质量等特性。本文用该算法研制了甚低码率视频编码器的软件 ,该软件编码器可以在 PSTN上实现实时编解码 ,具有较大的使用价值。
When coding the low bit rate image by H.263 standard, the complexity of the encoder mainly depends on motion estimation and discrete cosine transform. In order to reduce the computation amount of the encoder and improve the encoding speed, an effective frequency domain Motion Estimation Algorithm for Block Matching + Predicting All Zero Coefficients. This algorithm takes full advantage of the characteristics of discrete cosine transform, using block matching in the frequency domain for motion estimation, and the combination of predictive all-zero coefficients and motion search. Experiments show that this algorithm not only can effectively improve the encoding speed, but also can keep the high compression ratio of H.263 and the quality of moving images well. In this paper, the algorithm has been developed very low bit rate video encoder software, the software encoder can achieve real-time encoding and decoding on the PSTN, with great value.