To find sensor/mechanism in my following logic

In a grey scale image we take a sliding window of 3*3matrix .In the particular window constitutes of a pixel and its neighbouring pixels (in this case 8 neighbouring pixels of 3*3 matrix). If the middle pixel surrounding by neighbouring pixels is noised then in that sliding window by averaging the intensities of neighbouring pixel and that middle pixel intensity is replaced .If the middle pixel is not noised then there is no updation in that sliding window then sliding window go to another selected window. In order to find that the middle pixel is noised or not we should a mechanism and that mechanism is a sensor or degradation model. If that pixel is matched with that a sensor or degradation model then it is noised pixel. If not matched then it is not.

I want to know that mechanism/sensor Can any one help

Answer 1

[Walter Roberson]

Example:

some sensors can have pixels that are stuck completely off, or stuck completely on. If you model your sensor as being one of those varieties, then if you find a pixel that is 0 or the maximum and the pixels around it are all notably different than that, the pixel is probably stuck.

Now, read up on Schottky Noise. Are you modeling your sensor as being prone to it? If so then figure out how to detect it.

Next... are you modelling a sensor that is prone to Dark Current? How about to Quantum Noise

Okay, now is the sensor you are representing prone to blooming ? And other sensors have difficulties, in that they bleed pixels

Different kinds of sensors have different physical properties and so are more or less prone (or not at all) to different kinds of noise. Your sensor model should, ideally, know all the different types (and amplitudes) of noises of the system you are modelling.