CMOS image sensor: characterizing its PRNU (photo-response non-uniformity)

Abstract

To get higher resolution and higher SNR optical image data or digital picture, characterizing noise of image sensor helps to make better image sensor and better image signal processor. There are lots of manners with regards to characterizing image and noise quality of CMOS image sensors. And one of the most important performances is FPN(fixed pattern noise), because this kind of noise cannot be removed by temporal signal processing. This FPN can be divided into DSNU(dark signal non-uniformity) and PRNU (photo-response non-uniformity). PRNU depends on the uniform performance of sensitivity among pixels in general. Additionally, this uniformity of sensitivity is different from each color filter channels Red, Green and Blue in linear response region. And those performances were almost same in nonlinear response region for dim light condition. In this paper, in order to diagnose this tendency of PRNU regarding each color channel, special techniques are introduced to minimize other noise sources excluding PRNU from target images. And it also introduces that the different PRNU result between color and mono sensor by changing different color temperature light source. And it suggests ways to understand these results with simple manner as well.