论文部分内容阅读
用于电视信号帧内DPCM编码的非自适应预测器在图象的轮廓与边缘处经常导致大的预测误差,因此,它不适用于轮廓及细节丰富的图象。本文在轮廓预测的基础上提出一种新型的自适应预测器,图象的平坦部分及轮廓分别用不同的预测算法预测。预测器的自适应转换只利用前已传送的图象信息,不需向接收端另外传送附加的开关信息。该预测器的性能通过计算机模拟与其它已知的非自适应及自适应预测器进行了客观定量比较与主观试验。所建议的新的预测算法显著地降低了大预测误差的出现频数。与其它帧内预测器比较,采用这一新的预测器可以获得更大的码率压缩。
Non-adaptive predictors for intra-frame DPCM encoding of television signals often result in large prediction errors at the contour and edges of the picture and are therefore not suitable for outlines and detail-rich pictures. In this paper, a new type of adaptive predictor is proposed based on the contour prediction. The flat part and contour of the image are predicted by different prediction algorithms respectively. The adaptive conversion of the predictor uses only the previously transmitted image information without additional transmission of additional switching information to the receiving end. The performance of this predictor was objectively compared and subjectively tested by computer simulation with other known non-adaptive and adaptive predictors. The proposed new prediction algorithm significantly reduces the frequency of large prediction errors. Compared with other intra-frame predictors, this new predictor can achieve greater rate compression.