edit-reducing the co-efficient

this is a code

I=imread('dock.jpg');
I=rgb2gray(I)
threshold=0.624525
NewData = I;
idx = I >= threshold;
NewData(idx) = threshold - eps(threshold);

size(I);
size(NewData);

the size of I and new data are same,,the co-efficients are not yet reduced,and i get all values as 1,

else for example i have a image,which id dock.jpg,is it possible to remove the noise pixels ,so the co-efficients m,ay be reduced,please give suggestiongs for both ,the image is

http://imgur.com/Puw7o

Answer 1

[Walter Roberson]

After the line

I=rgb2gray(I);

add

I = double(I);

However, in previous discussions, I emphasized that this threshold is to be applied to the coefficients, not to the image. You are thresholding the image. You should be applying your T-cube algorithm (or whatever it was called) to your image to get the coefficients, and then after thresholding you would reconstruct the image based upon the altered coefficients.

[kash]

thanks walter ,i thought the pixles was the the co-efficients ,sorry to ask u can u plz help in getting the code for t-cube algorithm or other to get co-efficients please

[Walter Roberson]

I do not have any code for the decomposition routine you referred to earlier (whose real name I cannot locate at the moment, as you deleted that Question.)