This is a self-contained package including the 2-D DWT and inverse DWT routines. However, the Idwt2.m and Iidwt2.m codes are slow and inefficient. We did not try to optimize the codes at all. If you need faster routines, you should download and install the Rice Wavelet Toolbox, which is available via Rice DSP web page http://www.dsp.rice.edu INSTALLATION: download whmt.tar tar -xf whmt.tar decomposes the tar file into many matlab .m files. Main routines: hdenoise.m : denoise a 2-D image using HMT with training. You can also provide a pre-specified HMT model (no training). hmttrain.m : train HMT model for give 2-D wavelet transform. All wavelet coefficients at each scale within each subband are tied together to avoid overfitting of the model. hmtdeno.m : given 2-D wavelet transform and an HMT model, compute the posterior state probabilities. Then, use filter noisy wavelet coefficients to remove noise. hmtmodel.m : obtain an HMT model for real-world images without training. This model produces good denoising performance for many real-world images. Routines used by main routines: emhht.m : one step of EM algorithm for HH subband. Used by hmttrain.m emlht.m : same as emhht.m for LH subband. emhlt.m : same as emhht.m for HL subband. posthh.m : compute posterior state probabilities for HH subband. Used by hmtdeno.m postlh.m : same as posthh.m for LH subband posthl.m : same as posthh.m for HL subband 2-D DWT and inverse DWT routines: Idwt2.m : 2-D DWT Iidwt2.m : 2-D inverse DWT daubcqf.m : computes Daubechies wavelet filters makeh1.m, up.m : routines used by Idwt2.m and Iidwt2.m Other routines: gauss.m : gaussian density function testcon.m : compute convergence errors to stop HMT training vec2mat.m : used by hmtmodel.m Examples: ex256.m : denoising example of 256x256 Lena image generate a noisy 256x256 Lena. Train HMT and denoise. ex512.m : example using 512x512 Lena image. generate a noisy 512x512 Lena. Train HMT and denoise. example2.m : denoising example 2. generate a noisy Lena image. Use the HMT model in lenamodel.hmt (no training) to denoise it. example3.m : denoising example 3. generate a noisy Lena image. Use hmtmodel.m to obtain an HMT model for general real-world images. Then, use HMT denoising. Other files: lena256, lena512 : binary file containing original 256x256 and 512x512 Lena images. used in ex256.m, ex512.m, example2.m, and example3.m lenamodel.mat : HMT model (ES,PS,MU,SI matrices) for 512x512 Lena with \sigma=0.1 Used in example2.m

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：