基于结构化森林的前向车载视频图像的细节强化  

Detailed Enhancement of Forward Vehicle Video Images Based on Structured Forest

作　　者：金炜东[1] 胡燕花 唐鹏[1] 李伟[1] Jin Weidong;Hu Yanhua;Tang Peng;Li Wei(Southwest Jiao tong University,Chengdu 610031,China;Chengdu Metro Operation Co.,Ltd.Chengdu 610081,China)

机构地区：[1]西南交通大学,成都610031 [2]成都地铁运营有限公司,成都610081

出　　处：《系统仿真学报》2018年第12期4602-4609,4617,共9页Journal of System Simulation

基　　金：国家自然科学基金重点项目(61134002);中央高校基本科研业务费创新(2682014CX027);国家重点研发计划(2016YFB1200401-102F)

摘　　要：针对现有图像增强方法复杂度高,不能突出检测目标的问题,提出了一种基于结构化森林的前向车载视频图像细节强化的方法。该方法主要分为两个部分:结构化森林边缘检测和视觉增强。实现了接触网巡检图像的细节边缘强化,处理后的灰度直方图分布更均衡,增强后的巡检图像的细节区域与背景区域的灰度平均值之差、标准差之差更大,峰值噪声比以及结构相似性获得了提高,通过与其他方法对比,表明算法行之有效。可以更加直观的为铁路工作人员展示铁路基础设施的异常情况,具有很强的实践意义。In view of the complexity of existing image enhancement methods,which cannot highlight the detection target,this paper proposes a method for detail enhancement of forward vehicle video based on structured forest.This method is mainly divided into two parts:structured forest edge detection and visual enhancement.This paper implements the detail edge extraction and enhancement of the patrolling image of the catenary.The gray histogram distribution after processing is more balanced,and the difference between the average values of gray levels in the detail area and background area of the enhanced patrol image is the standard deviation.The difference is even greater,the peak-to-noise ratio and the structural similarity have been improved.By comparing with other methods,it shows that the algorithm is effective.It can be more intuitive to show railroad workers the anomalous conditions of the railway infrastructure,which is of great practical significance.

关 键 词：结构化森林 边缘检测 图像增强 图像融合 

分 类 号：U2[交通运输工程—道路与铁道工程]