Noise Model Analysis for Sensor Arrays

This paper proposes a new possible noise model and its application to Bayer color filter array (CFA). We studied effects of several noise models on three Bayer patterns. For instance, current Bayer CFA uses RGGB pattern which contains two green pixels, a red and a blue pixels in a pair. However, one may consider other color combinations such as RRGB or RGBB patterns. To investigate RGGB pattern's superiority, we research each pattern in different noise model and its level. We also studied how to generate noisy image from noisy-free image. In addition, conventional noise models do not consider white balance, gamma and tone correction issues. In this paper, we suggest adding such components in the model. Experimental results show that the new noise model is reasonable than the conventional ones.