WBCT变换的SPIHT图像压缩算法的改进

Improvement on a SPIHT image compression algorithm on the basis of wavelet-based contourlet transform

传统的基于WBCT变换的SPIHT图像压缩算法没有考虑低频子带与高频子带的关系,只是在高频子带之间寻找方向树的关系,并且需要对变换后的系数进行位置置换,这样必然会影响编码质量和效率.针对以上问题提出了构造虚拟低频的思想,通过构造虚拟低频建立低频子带与高频子带的关系,使方向树结构更高,压缩效果更好;同时又避免了系数位置置换,提高了编码效率.与传统算法和现有的WBCT算法相比,该算法既能有效保护图像细节和纹理,又节省了编解码时间,同时提高了压缩后图像的峰值信噪比(特别是在低比特率下),而且具有通用性.

The traditional wavelet-based contoudet transform （WBCT） algorithm does not consider the relationship between a low-frequency sub-band and a high-frequency sub-band ; it only seeks for a relationship of direction trees between high-frequency sub-bands. In addition, such an algorithm needs to adjust the location of the transformed coefficient, and this will inevitably reduce the quality and efficiency of coding. In view of the above issue, a thought on forming virtual low frequency was presented. By forming virtual low frequency, the relationship between a low-frequency sub-band and a high-frequency sub-band was established; this made the structure of the direction tree higher and the compression effect better, and in the mean time, avoided the adjustment of coefficient location while improving coding efficiency. Compared with the traditional algorithm and existing WBCT algorithms, the new algorithm not only effectively protects the texture and contour of an image, but also saves time. Additionally, it enhances the peak signal to the noise ratio of an image （especially at low bit rates）, and displays universality.