海上薄雾条件下退化图像的快速Retinex增强算法

目前基于Retinex 理论的各种算法对海上薄雾条件下拍摄到的退化图像有很 好的增强效果，但是这些算法主要是采用在RGB 色空间对3 个颜色通道进行高斯卷积方 法，导致处理较大尺寸图像速度慢，现针对上述现象，提出一种改进的Retinex 快速增强 算法来处理海上薄雾条件下的退化图像，此方法改变了传统Retinex 算法利用二维高斯核 进行卷积的办法， 并且考虑到HSV 空间更适于人眼视觉特征， 先将RGB 色空间转换到 HSV 空间，再将二维高斯函数简化成一维迭代高斯函数，然后应用一维迭代高斯函数对 亮度V 分量进行卷积，并对饱和度S 做相应的拉伸。根据时间复杂度理论对改进的方法 进行了分析，并经过实验验证，此算法的处理速度有明显的提高，去雾效果较明显，图 像质量得到了改善，为视觉导航系统后续的工作提供了保证。

A Rapid Retinex Enhancement Algorithm of Images Degraded in Mist

Currently, the algorithms based on the theory of Retinex have effective enhancement on images captured under the conditions of sea mist. However, these algorithms are mainly used in the RGB color space with the method of Gaussian convolution on three color channels, which result in slow processing of larger size image. Now for the phenomenon above, a rapid enhancement Retinex algorithm is proposed to improve the degradation images in the sea mist. This method changes the traditional idea of Retinex algorithm using a two- dimensional Gaussian kernel convolution approach. Besides having regard to the character of HSV space, Retinex in HSV space is more suitable for human visual characteristics. Firstly, make conversion from RGB color space to HSV space and simplify to one-dimensional differential equations from two-dimensional Gaussian function. Then make the application with V component and stretch saturation S accordingly. This article analyzes the advanced algorithm based on the time complexity theory and experimental verification, the processing speed of this algorithm is significantly improved. Defogging effect is obvious and the image quality is improved, providing visual navigation system in follow-up work with a guarantee.