I need to write a code to detect and remove the fixed pattern noise in an image .Can you please help me on how to write the code.

I have questions like: Fixed pattern noise removal apply on the Bayer image from sensor or RGB image? How to write a code for detecting the amount of fixed pattern noise in the image? How to write a code for reducing the fixed pattern noise in the image? This process i want to do for continous frames(vide) Can you please help on below issue as soon as possible?

Answer 1

[Anitha Thomas]

Fixed pattern noise can be measure in the absence of illumination.

Answer 2

[Image Analyst]

You remove the lens from the camera so that the light is uniformly on the sensor. Then set the exposure time so that you get a nice image with no saturation (clipping) at the dark or bright end. Then convert that image to a percentage image where every pixel is the percentage of the brightest value in the image. Now that image represents the relative gain of each pixel. Now, to fix an arbitrary image, you can divide the actual image by the fixed pattern percentage image. Why? Well if a pixel is only 90% as responsive as some other pixel (your best pixel), then you'd want to divide that pixel's value by 0.9, wouldn't you? Of course.

If you want to be super accurate, you can completely block all light from hitting the sensor and measure the signal at the same exposure. That is your dark noise. You can subtract the dark noise image from the test image before you do the division.

Needless to say you need to acquire lots of fixed pattern and dark noise images and average them together first because you're dealing with very noisy images.