how to compress image using wavelet transform???give code....but don't use direct function like wavedec2,WPDENCMP

8 次查看（过去 30 天）

pratik chavda 2013-12-26

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/110755-how-to-compress-image-using-wavelet-transform-give-code-but-don-t-use-direct-function-like-wave

评论： Walter Roberson 2013-12-26

I. Decompose
  Choose a wavelet; choose a level N. Compute the wavelet.
   Decompose the signals at level N.
II. Quantizer 
a. Threshold detail coefficients
  For each level from 1 to N, a threshold is selected and hard thresholding is applied to the detail coefficients.
-Hard thresholding
-Soft thresholding
In ,hard thresholding ,a tolerance is selected. Any wavelet whose absolute value falls below the tolerance is set to zero with the goal to introduce many zeros without losing a great amount of detail.
Another type of thresholding is soft thresholding. Once agin a tolerance, h, is selected. if the absolute value of an entity is less than the tolerance,than that entry is set to zero. All other etries,d,are replaced with sign(d) ||d|-h|.
Soft thresholding can be though of as a translation of signal toward zero by the amount h.
   b. Entropy encoding
  Wavelet and thresholding help process the signal, but until this point, no compression has yet occurred. 
III. Reconstruct
  Compute wavelet reconstruction using the original approximation coefficients of level N and the
  modified detail coefficients of levels from 1 to N.[5]