一种快速的纹理预测和混合哥伦布的无损压缩算法

摘要

为了进一步降低芯片内无损压缩的运算复杂度和编码时间,该文在保持高压缩率的基础上,提出一种基于方向预测和混合熵编码的快速无损压缩算法.该算法首先采用自适应方法进行纹理方向的预测,以获得当前像素的参考像素,并计算预测残差;然后对预测残差进行混合哥伦布编码,最终大幅度地提高了无损压缩的压缩性能.实验结果显示,与基于梯度预测和变长编码的无损压缩算法相比,该算法在平均压缩率略有提升的前提下,平均编码时间减少了36.86％.%A fast-lossless compression using texture prediction and mixed golomb coding is proposed to reduce the computational complexity while keeping high compression ratio. First, the reference pixel of the current pixel is gotten by texture direction prediction, meanwhile, the pixel difference is calculated. Then, the pixel difference is entropy coded through mixed Golomb. Thus, the compression performance is improved greatly. Simulation results show that compared with lossless frame memory compression using pixel gain prediction and dynamic order entropy coding, the proposed algorithm reduce the average coding time by 36.86％. Moreover, the average compression ratio is increased slightly in the proposed algorithm.