基于整数小波变换和改进嵌入零树编码的图像压缩

Image Compression Based on Integer Wavelet Transform and Improved Embedded Zerotree Wavelet Encoding

整数小波变换(IWT)具有输入输出都是整数、只需进行位内运算、便于硬件实现等优点,特别适于图像的无损压缩;而在有损压缩方面,其效果通常稍逊于传统的离散小波变换(DWT)。为了提高IWT图像有损压缩的性能,该文采用基于提升格式的IWT,结合基于形态膨胀运算的改进嵌入零树小波编码(EZW)方法。实验结果表明,在没有增加运算复杂度的情况下,此算法与传统的DWT相比,提高了峰值信噪比(PSNR),具有较好的有损压缩效果。

The main advantages of Integer Wavelet Transform(IWT) are that the input and output values are all integers, all operations can be done in place, only a small memory is required and easy to be implemented in hardware. In the context of image coding, IWT is well suited for lossless compression. However, it performs a little worse compared to the conventional Discrete Wavelet Transform(DWT) for lossy compression. In this paper, a new algorithm is proposed for improving lossy compression performance of IWT. It is by means of the IWT based on lifting scheme combining with improved Embedded Zerotree Wavelet(EZW) based on morphological dilation operation. Simulation results show the proposed algorithm improves the Peak Signal Noise Ratio(PSNR) compared to conventional DWT without increasing computational complexity.