how to reduce noise in image.

i have an image A (A=orginal image + secret message). now i add noise to A such as Gaussian noise or Salt and Pepper noise. how can i detect A and noise. i used a lot of filter module but i can not cover my original image. please help me. thanks a lot.

Answer 1

[Image Analyst]

Unless you were the one who inserted the secret image, you most likely will not be able to remove it. There is a whole field of steganography dedicated to inserting secret images and making them difficult or nearly impossible to remove. The good algorithms can survive many kinds of attacks such as noise, rotation, cropping, etc. But you can reduce noise and get close to the original image+secret image. For example, here's some code I posted a few days ago: http://www.mathworks.com/matlabcentral/answers/56515#answer_68417 There are several variants of that concept, but basically you filter the image and then replace only the noise pixels with filtered pixels. If a pixel is determined to be noise-free, then it is left alone.

[Tran Loc]

thank you so much but it filter most of my secret message. my original image is binary image and secret message is binary information. I can get back my (original image + secret message). Please help me.

[Image Analyst]

I thought I did help you. The link I gave you removes salt and pepper noise. Did you try it? That should get you back a good estimate of the "pre-noise" image, which you say is the "(original image + secret message)" image.