基于双边滤波和Retinex算法的货车图像预处理方法

Method of freight vehicle image based on bilateral filtering and Retinex algorithms

针对货车故障轨边图像检测系统(TFDS)中的图像光照不均匀问题,提出一种基于双边滤波和Retinex算法的货车图像预处理方法。该方法首先利用双边滤波法把图像分成两部分(粗略图像和细节图像),然后利用Retinex算法对图像的粗略部分进行增强,最后把细节部分与处理好的粗略部分进行合成,以达到最佳的图像视觉效果。实验结果表明,该方法在保留图像细节的同时很好地克服了光照不均匀对图像的影响。

Aiming at the problem of uneven illumination of images in Trackside Failure Detection System (TFDS),this paper proposes a preprocessing method of freight car images based on bilateral filtering and Retinex algorithm.The method firstly divides the image into two parts (rough image and detail image) by using bilateral filtering method,then uses Retinex algorithm to enhance the rough part of the image,and finally synthesizes the detail part and the processes the rough part to achieve the best image visual effect.The experimental results show that the method not only preserves the details of the image,but also overcomes the influence of uneven illumination on the image.