为了解决汽车抗晕光场景中，融合图像质量的客观评价与人眼视觉效果不一致的问题，提出了一种新的可见光与红外融合图像质量评价方法，该方法通过设计自适应迭代阈值法自动确定融合图像的晕光临界灰度值，并将融合图像自动分为晕光区和非晕光区。针对晕光区，设计晕光消除度指标评价晕光消除的效果；针对非晕光区，从多方面评价色彩、细节信息的增强效果，并甄选出合适的指标构成完整的图像质量评价体系。为验证该方法的合理性，对采用 4种不同算法的融合图像进行评价，实验分析表明，该方法的主客观评价结果一致，适于评判不同可见光与红外融合的抗晕光图像质量及算法的优劣。

Owing to the problem that objective evaluation of fusion image quality is inconsistent with human visual effects in vehicle anti-halation, a new quality evaluation method is proposed in this paper for visible and infrared fusion images. The method can automatically determine halo critical gray values to divide fusion images into halo and no-halo regions by designing an adaptive iterative threshold algorithm. For the halo region, the halo limination effect is evaluated by designing a halo limination index. For the no-halo region, enhancement effects of color and detail information are evaluated from various aspects. Appropriate indexes are selected to constitute a complete image quality evaluation system. Fusion images of four different algorithms are evaluated to verify their rationality. Experimental analysis show that subjective and objective evaluation results of this method are consistent, rendering it suitable for evaluating anti-halation image quality and the algorithm of different visible and infrared fusion.