基于改进Retinex的城市交通图像增强

发布时间：2018-05-17 13:52

本文选题：城市交通 + 图像增强　；参考：《交通运输系统工程与信息》2017年05期

【摘要】：为提高在雾霾天气条件下获取的城市交通图像的对比度和清晰度,本文提出了一种基于引导滤波平滑约束和分数阶积分的改进Retinex的图像增强方法.Retinex理论把一幅图像视为照射分量与反射分量乘积的形式.首先,利用引导滤波作为Retinex算法的目标函数的平滑约束项,通过最小化该目标函数获取照射分量;反射分量为原始图像与照射分量的比值,其主要体现图像的局部细节,对反射分量进行分数阶积分掩膜,在去除噪声干扰的同时保留图像的细节信息;最后,将处理后的反射分量与照射分量相乘,得到最终的增强图像.实验结果表明,本文方法在图像平滑和细节保持方面得到了很好的平衡,有效地提高了雾霾天气条件下城市交通图像的对比度和清晰度,使得增强后的图像更加真实自然.
[Abstract]:To improve the contrast and sharpness of urban traffic images obtained under haze weather conditions, In this paper, an improved Retinex image enhancement method based on guided filtering smoothing constraints and fractional integral is proposed. The theory of Retinex regards an image as a product of irradiating and reflecting components. First of all, the guided filter is used as the smooth constraint of the objective function of the Retinex algorithm, and the radiation component is obtained by minimizing the objective function, and the reflection component is the ratio of the original image to the irradiated component, which mainly reflects the local details of the image. The fractional integral mask is used to remove the noise and keep the detail information of the image. Finally, the final enhanced image is obtained by multiplying the processed reflection component with the irradiation component. The experimental results show that the proposed method achieves a good balance between image smoothing and detail preservation, and effectively improves the contrast and sharpness of urban traffic images under haze weather conditions, and makes the enhanced images more realistic and natural.
【作者单位】：长安大学信息工程学院;
【基金】：国家自然科学基金(41372330) 长安大学中央高校基金优秀博士学位论文培育资助项目(310824165003)~~
【分类号】：TP391.41;U495

【相似文献】