关于二维小波变换的程序 [精华]



说明：此算法重在概念，速度并不是很快。因为FOR循环的缘故。此程序从循环矩阵的观点出发，把圆周卷积和快速幅里叶变换建立了联系。实现了分解和无失真重构。它只做了一层分解，即将256x256图形分解成为64x64的四个图形，避免了使用WKEEP()的困惑。主要思想为用小波滤波器族构造正交阵W,变换写为B=W*A*W ,反变换为：A=W *A*W,这与所有正交变换无异。W为循环正交矩阵，因此可用FFT实现快速运算，难点就在重构矩阵上。若用矩阵概念明确，一个共扼转置可以搞顶。但FFT的使用必须找到与分解序列的关系。-on Wavelet Transform procedures [best] Note : This algorithm important concept is not speed quickly. Because FOR cycle nowadays. This program cycle from the point of view of Matrix, circular convolution and fast Fourier transform is established links. To achieve the decomposition and reconstruction without distortion. It only had one decomposition, 256x256 graphics will be 64x64 decomposition of the four graphics, avoiding the use of WKEEP () confusion. The main idea of using wavelet filter generator orthogonal array W, write to transform B = W * A * W, anti-Transform : A = W * A * W, which is orthogonal transformation all the same. W cycle orthogonal matrix, can be used to achieve rapid FFT computation, on the difficult reconstruction matrix. If a clear matrix concept, a home to a total o