基于小波图像压缩技术的算法研究

Algorithm Research on Image Compression Technologies with Wavelet Transform

小波变换通过多分辨分析过程将一幅图像分解成近似和细节部分,细节对应的是小尺度的瞬间,在本尺度内很稳定。因此将细节存储起来,对近似部分在下一个尺度上进行分解,重复该过程即可。近似与细节在正交镜像滤波器算法中分别对应于高通和低通滤波器,这种变换通过尺度去掉相关性,在图像压缩中被证明是有效的。由于小波变换后高频部分小波系数的绝对值较小,而低频部分小波系数的绝对值较大,这样,在图像编码处理中,可以对高频部分大多数系数分配较小的比特以达到压缩的目的。

With the process of muhiresolution analysis, wavelet transform seperates an image into approximate and detail parts. The detail part corresponds with the small scaling transience which is stable in the small scale, and then wavelet transform analyzes the approximate part in the next scale. In the algorithm of quadrature mirror filters, the approximat part corresponds with high - pass filter and the other part corresponds with low - pass filter. This transform subtracts correlation through scale and it is efficient in image compression. After wavelet transform, the absolute value of wavelet coefficients in the part of high frequency is less, and that in the part of low frequency is more. In that case, the image can be compressed with less bits for in the most coefficients in the part of high frequency.