针对普通压缩算法和基于DSP或PC机的实现方法已不能满足高速和小体积图像压缩的要求,本文以硬件实现高速图像压缩为目的,提出一种基于5/3小波变换的低复杂度图像压缩算法。该算法首先采用3级二维小波变换去除图像相关冗余,并根据小波子带的变换增益对其进行最佳量化,然后对量化后的LL子带进行二维预测,最后针对小波系数概率分布的特点,采用自适应零游程编码联合指数哥伦布编码实现图像压缩。该方法在保证较高压缩质量的同时,具有低复杂度和硬件易实现的特点,通过FPGA最快可实现高达175 Mpixel/s的超高速图像压缩,为高速图像压缩器件的研制提供了算法基础。

Because ordinary image compression algorithms and other compression methods based on Digital Signal Processors(DSPs) or PC machines can not satisfy the demands of high speed and small volume image compression applications,a low complexity image compression scheme based on the 5/3 Discrete Wavelet Transform (DWT) is proposed in this paper.Firstly, the correlation redudancy of image is excluded with 3-level z-dimension DWT. Secondly, the DWT coefficients are quantized according to the conversion gain of wavelet subband and the predict of LL subband is quantized. Finally,the image compression by combining adaptive zero run-length coding with Exp-Golomb coding is realized based on the probability distribution. Experimental results show the scheme has advantages in low computational complexity and hardware implementation,while it maintains relatively high performance in compressing various images.Moreover,it offers a high speed image compression of 175 Mpixel/s by FPGA,which provides an algorithm base for developing high speed image compression devices.