摘要
针对红外图像空间相关性强的特点,提出了一种红外图像上相邻区域灰度直方图有较高相似度的铁路跟踪算法。对红外图像进行中值滤波和直方图均衡化等数字图像预处理,提高了景物和目标的对比度,从而有利于准确跟踪,用相关系数比较相邻区域两直方图的相似度来确定跟踪目标区域。利用相邻帧间图像差别微小,采用多模板加权自适应更新的方法,大大提高了算法对环境的适应能力和跟踪的稳定性。此算法能准确获取铁路的位置,为铁路信号灯和路障检测提供前提,对铁路安全运输有很重要的意义。
Because of the obvious spatial similarity of the infrared image, a kind of railway tracking algorithm which based on the high similarity of the gray scale histogram of infrared image, is presented in this paper. By means of infra- red image median flitering and the histogram equabzing,we can enhance the scenery and the contrast of the objects. It determines the tracking target area through comparing the similarity cofficient between two adjacent region histograms and that benefits the accuracy of tracking. Besides ,considering the small difference of the adjacent interframe images and through multiple templates weight self-adaptive updating, the ability of this algorithm has been improved signifi cantly. The method can accurately locate the railway position and provide basis for the railway signal lamp and obstacle detecting. It has great significance for the safety of railway transportation.
出处
《激光与红外》
CAS
CSCD
北大核心
2012年第12期1385-1388,共4页
Laser & Infrared
关键词
铁路跟踪
相关系数
相似度
直方图
预处理
railway tracking
correlation coefficient
similarity degree
histogram
preprocessing
