基于监督颜色恒常性的低照度亮度校正算法

提出了一种新的监督颜色恒常性算法,主要解决在机器视觉识别中由于色温变化所带来的光照校正问题。利用人类视觉对比度分辨率的限制规律,采用抛物线函数作为亮度增益函数,将图像亮度映射于实验确定的最佳识别范围。实验结果表明,该算法能够获得较理想的视觉校正效果,有效地提高识别率;不需要进行颜色空间转换,计算耗时少。

Correction algorithm based on supervised color constancy for low-illumination images

A new surveillance color constancy algorithm is proposed. It is used in machine vision recognition to solve illumination adjustment due to color temperature variation. According to the human visual contrast resolution limit rule, the parabola function is adopted to take brightness gain. The image illumination values are mapped into superior experimental scope. Experiment results show that the algorithm can achieve the desired visual adjustment effect, and enhance the recognition rate effectively. It does not need to transform the RGB to another color space, so the computation can be faster.