Question

Hi, guys , I have a problem here regarding image processing . I am supposed to denoise an image by taking average of multiple image as suggested by http://www.cambridgeincolour.com/tutorials/image-averaging-noise.htm . I asked in stackexchange , the guys are just too polite to answer me directly and I do not want to disturb them ... So , if anybody want to collaborate me here .

Answer 1

Do you use any image processing program or not? Those programs might have an option to do it, or maybe the program Neat Image they suggest on their site. If you will program it yourself, we will have to look into the algorithms they use and how to interpret noise for an image.

A BMP image for example stores the pixel data in an array size(nr pixels x, nr pixels y) where every pixels has 4 bytes of color data (alpha, red, green, blue for example). Each byte has a value between 0 and 255 (or 00 to FF), to indicate the amount of each base colour in the pixel. Example from wikipedia:
808000FF would be Red and Green:50.2%, Blue:0% and Alpha:100%, a brown.
If you would add a black (000000FF), and just average the byte values, you would get:
404000FF, a darker brown.
I assume that they would simply 'average' the two colours from corresponding pixels in an image to average out the noise. However, they probably use quite some filters and algorithms for recognizing corresponding areas in an image (when you slightly moved your camera) for example, I'm not sure about how they fix that. Maybe you could try averaging two simple 10x10 pixel 1-bit monochrome binary image (black or white) with random black or white pixels?