双正交重叠变换的整数实现算法与图像压缩

Integer Algorithm of Lapped Biorthogonal Transform and Image Compression

本文利用提升原理 ,以二进制整数DCT为基础 ,提出了双正交重叠变换的整数实现算法 ,其计算只需要少量的移位和加法运算 ,因而软件实现和硬件实现都非常简单 .实验结果表明 ,它是一种有效的图像压缩变换 ,压缩质量大大超过了运算量更多的近似二进制整数小波变换和DCT ,与Daubechies 9/ 7小波也非常接近 .

This paper describes the implementation of integer lapped biorthogonal transform (binLBT) using lifting scheme and binDCT. The implementation is in the forms of shifts and adds, so it is very fit for software and hardware implementations. Image coding experiments show that the binLBT is far superior to the DCT and the binary digit 9/7 wavelet which needs more operations, moreover, in subjective coding performance the binLBT is very competitive with the 9/7-tap biorthogonal wavelet.