基于融合策略的单幅图像去雾算法

Single image defogging based on fusion strategy

为了有效增强雾天衰退图像,提出了一种基于融合策略的单幅图像去雾算法。该融合策略无需依靠大气散射模型或场景结构信息,只需通过原始衰退图像来获取其输入图与权重图。其中2幅输入图的作用主要是对原有雾图像进行颜色校正和对比度增强,而3幅权重图则突显了图像雾气较浓区域的细节信息。融合策略将上述输入图与权重图相融合以生成对比度高、色彩丰富的去雾图像。此外,还从人类视觉感知的角度提出了一个新的去雾效果评价指标,从而为图像去雾与去雾效果评价问题提供了新的解决思路。实验结果表明,与已有方法相比,提出的基于融合策略的去雾算法能更好地提高各种雾天图像的清晰度。

To effectively enhance the degraded foggy image, a single image defogging algorithm based on fusion strategy was proposed. The strategy does not require the atmospheric scattering model or scene structure, only needs the original degraded image to derive the inputs and weight maps. The function of the two inputs is to correct color and enhance contrast, and the three weight maps improve the detail information of the regions in dense fog. All the inputs and weight maps are fused by using fusion strategy to produce a high-contrast and vivid fog removal image. Besides, an index for defogging effect assessment is also proposed from the perspective of human visual perception, which provides a new so- lution to the problem of single image defogging and its effect assessment. Experimental results show that the proposed fusion strategy method can better improve the image quality for various foggy images.