Design and FPGA implementation of automatic white balance algorithm for images from rearview electronic mirrors

In the fast-evolving realm of automotive intelligence, the advent of electronic rearview mirrors has imposed stringent demands on the real-time capabilities of image capture and display systems. Among these, the Automatic White Balance (AWB) algorithm plays a pivotal role in correcting color discrepancies due to varying environmental temperatures, thereby enhancing image authenticity and naturalness. This study addresses shortcomings of traditional AWB algorithms—namely, inflexibility, susceptibility to failure, and poor real-time performance—by implementing an improved Orthogonal Methodbased AWB algorithm on FPGA. The approach involves dividing 1920×1080 resolution images into 100 blocks and independently calculating gain factors and color bias corrections for each block. Comparative analyses with Greyscale World and original Orthogonal Methods demonstrate superior color correction effectiveness across diverse scenarios. Notably, the algorithm exhibits minimal color deviation in low-light and high-color-difference environments, outperforming alternative methods. Furthermore, harnessing FPGA's parallel processing capabilities accelerates algorithm execution. This study integrates the algorithm into a ZYNQ-based image acquisition and display system. Timing analysis using the ILA debugging tool confirms reduced latency to 0.47ms, showcasing superior real-time performance compared to adaptive AWB algorithms. This advancement meets the stringent real-time demands essential for applications such as electronic rearview mirrors in high-speed video surveillance.