ImageAnalyst <firstname.lastname@example.org> wrote in message <email@example.com>... > superZZ: > You can try to subtract a median filtered version from the noisy > original and check for values that exceed some threshold. > > Or you can try Brett Shoelson's deleteoutlier function: > http://www.mathworks.com/matlabcentral/fileexchange/3961
Thank you for your suggestion.
But it does not work well with my test data. I tried many alpha value, but only a small part of noisy points are detected. Below is my test code.
Sorry for my duplicated post. I sent my post and went to sleep without shutdown computer. In the tomorrow morning, I want to see whether my post are replied, so I pressed F5 to refresh the page. But I really did not realise that action will cause the post is sent to newsletter group again at that time. Sorry!