基于行的整数小波变换图像压缩算法

为克服小波图像压缩中内存开销大的问题，提出一种基于行的图像压缩算法．该算法是在完成行向小波变换后以累进方式完成列向小波变换，降低了对存储容量的需求，在大数据量图像压缩上优势明显．针对基于行的特点，采用整数实现的提升格式代替传统的Mallat塔式算法，实现CDF9／7的双正交小波变换，减小了计算复杂度，加快了算法执行速度．实验表明，该算法在比特率为0．5bpp附近时，恢复图像的峰值信噪比均高于30dB．根据人眼视觉特性及小波变换特点对变换系数进行感觉量化，提高了压缩比，适用于超大图像或内存敏感的应用场合．

Line-Based Integer Wavelet Image Compression Algorithm

A line-based image compression algorithm is put forward to get over the problem of high memory utilization in the process of wavelet transform. Column transform is performed progressively after each row transform , so this algorithm lowers the requirement of memory. The algorithm has an obvious advantage over the image compression of data in large number. According to the characteristics of line-based wavelet, the CDF9/7 filter is completed by integer lifting scheme instead of the traditional Mallat pyramid algorithm. The integer-lifting scheme reduces the computational complexity and accelerates its implementation. The experiments show that all the peak signal to noise ratio (PSNR) values of the reconstructed images are higher than 30 dB when the bit-rate is near 0. 5 bpp. According to the features of human visual system and characteristics of wavelet transform, the wavelet coefficients are quantized to improve the ratio of compression. It is applicable to large images and the situation of the limited memory.