Design and implementation of two key image processing techniques for CMOS image sensor based on fpga

Automatic white balance and color filter array interpolation are two important function of image signal processor for CMOS image sensor. This paper describes the designs and implementations of new automatic white balance method and color interpolation method on FPGA. This white balance method can effectively get the better of the defects of gray world and gray world-retinex methods by calculating the average deviation values of Cr and Cb. This color interpolation method can greatly reduce the artifacts causing by interpolating across edges with comparing the differences of neighboring pixels with the predefined thresholds. These two methods have been realized on Altera¿s FPGA EP1C12Q240C8. Experimental results show that the white balance method can obtain better image quality than gray world and gray world-retinex methods, and the color interpolation method can get high PSNR value. They can be good candidate for VLSI implementation.