一种基于整数小波和SPHIT算法的静态图像编码方法

A static image compression algorithm based on integer wavelet transform and SPHIT

自1993年Shapiro提出了EZW编码后,小波图像编码得到了快速发展.整数小波变换不仅使信息无损表示,而且减少了运算时间,使算法更易于软硬件实现.为了能够快速访问影像数据并保证图像的高清晰度,笔者在SPHIT算法的基础上,结合使用5/3整数小波变换并改进了SPHIT算法的一些不足,提出了一种可行且高效的图像编码方法,主要思路是保留LL子带,对高频部分采用类似于零树小波中的Z字型扫描方法,依照非线性最优的原理,通过3个存储数组完成了由原SPHIT算法中的3个队列反复迭代完成的图像位小波系数的嵌入式码流.实验结果表明:该方法不仅减少了复杂度和占用的内存量,而且易于硬件实现.

Since Shapiro presented the Embedded Zerotrees of Wavelet Transforms in 1993,wavelet transforms had been developed rapidly.Integer wavelet transform can represent information losslessly and reduce the operation time.It also can make the algorithm easier to be implemented by hardware and software.In order to access image data rapidly and guarantee image with high quality,a feasible and efficient image compression method based on SPHIT,is presented,which combines 5/3 integer wavelet transform and improves SPHIT algorithm.The main idea is to reserve the LL part and use the Zig-Zag scanning method for the high frequency part.Three arrays instead of three queues are used to realize the embedded code of wavelet coefficients in image bits according to the nonlinear optimal principle,The experiment shows that this image compression algorithm can reduce the complexity and use less memory.It can be implemented easily by hardware.