基于曲波变换的图像去雾算法研究

Research on Image Defogging Algorithm Based on Curvelet Transform

图像去雾技术的目的是去除摄像头采集图像过程中雾、霾等复杂天气因素对图像的干扰,保留图像目标的细节信息,提高图像的整体质量。通过对比晴天图像和雾霾条件下图像,分析雾霾颗粒对图像的影响,并研究离散曲波变换理论,提出了一种基于离散曲波变换并结合高斯函数滤波的图像去雾方法。该算法对图像进行高斯函数滤波,优化图像质量,缩短算法处理时间,通过改进的离散曲波变换算法,使图像的清晰度和对比度达到正常图像水平。对比处理前后图像的信噪比和信息熵等客观参数,证实了基于离散曲波变换的图像去雾算法,对雾霾天气下的图像有明显的去除雾霾效果,得到的增强图像保留了较多的图像信息,图像清晰程度明显提升,扩展了户外视觉系统的应用范围。

The goal of image defogging technology is to remove haze interference of camera capturing images under complex weather con- ditions, such as fog, thus the detail information of image target has been retained and the overall quality of the image has been improved. By comparing the sunny image with the image under haze conditions, the influence of haze particles on the image has been analyzed and the theory of curvelet transform has been studied. The method of image defogging based on curvelet transform and combined with Gauss function filtering has been proposed. The algorithm has adopted Gauss function filtering dealing with images for optimizing the quality of image and shortening its transformation time, and then improving discrete curvelet transform algorithm makes the image clarity and con- trast to normal level. Compared with objective parameters of processing the image signal-to-noise ratio and information entropy before and after the images, it has been proved that the image defogging discrete algorithm based on discrete curvelet transform has obvious effect of removing haze in the haze weather image, and the enhanced image retains more image information, the image clarity has been obviously improved to extend to the outdoor application range of the visual system.