论文部分内容阅读
针对SPIHT(set partitioning in hierarchical trees)算法的编码过程具有重复运算、存储量大等问题,提出了一种适合于DSP(digital signal processors)处理的低内存并行SPIHT算法。该算法采用乒乓缓存策略,使得数据的传输和编码能够同时进行。通过引入基于行的整型提升方案,使得只需经少量行变换就能进行列变换,提高了小波的变换速度。根据DSP的并行特性和SPIHT算法的缺点,采用“改进的最大幅值求取方法”、“误差位数以及绝对零值和绝对零集合”、“最大值与零值图”和“单棵零树编码”等多种方法对其进行了改进,大大缓解了对内存的压力,减少了算法的运算量。该算法与LZC(listless zerotree coding)算法相比,重构图像的峰值信噪比相当,但速度提高了2倍,能满足一般的实时压缩要求。
The encoding process of SPIHT (set partitioning in hierarchical trees) algorithm has the problems of repetitive computation and large storage capacity, and proposes a low memory parallel SPIHT algorithm suitable for DSP (digital signal processors) processing. The algorithm uses ping-pong caching strategy, making the data transmission and encoding can be carried out simultaneously. By introducing a row-based integer lifting scheme, column transforms can be performed with only a small amount of line transforms and the transform speed of the wavelet is improved. According to the parallelism of DSP and the shortcomings of SPIHT algorithm, the improved method of maximum amplitude, the number of error bits, absolute zero and absolute zero set, maximum and zero value graph and single zero Tree coding "and other methods to improve it, greatly easing the pressure on the memory, reducing the computational complexity of the algorithm. Compared with the LZC (listless zerotree coding) algorithm, the reconstructed image has the same peak signal-to-noise ratio, but the speed is improved by two times, which can meet the general real-time compression requirements.