10 January 2018 A robust embedded vision system feasible white balance algorithm
Author Affiliations +
Abstract
White balance is a very important part of the color image processing pipeline. In order to meet the need of efficiency and accuracy in embedded machine vision processing system, an efficient and robust white balance algorithm combining several classical ones is proposed. The proposed algorithm mainly has three parts. Firstly, in order to guarantee higher efficiency, an initial parameter calculated from the statistics of R, G and B components from raw data is used to initialize the following iterative method. After that, the bilinear interpolation algorithm is utilized to implement demosaicing procedure. Finally, an adaptive step adjustable scheme is introduced to ensure the controllability and robustness of the algorithm. In order to verify the proposed algorithm’s performance on embedded vision system, a smart camera based on IMX6 DualLite, IMX291 and XC6130 is designed. Extensive experiments on a large amount of images under different color temperatures and exposure conditions illustrate that the proposed white balance algorithm avoids color deviation problem effectively, achieves a good balance between efficiency and quality, and is suitable for embedded machine vision processing system.
© (2018) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Yuan Wang, Feihong Yu, "A robust embedded vision system feasible white balance algorithm", Proc. SPIE 10616, 2017 International Conference on Optical Instruments and Technology: Optical Systems and Modern Optoelectronic Instruments, 1061607 (10 January 2018); doi: 10.1117/12.2287363; https://doi.org/10.1117/12.2287363
PROCEEDINGS
9 PAGES


SHARE
RELATED CONTENT


Back to Top