桥梁红外图像的自动识别跟踪方法探讨

AUTOMATIC RECOGNITION AND TRACKINGTECHNIQUES FOR INFRARED BRIDGE IMAGE

文中以实时采集到的桥梁红外图像为基础，探讨了复杂背景中桥梁的识别跟踪方法。文中先对桥梁红外图像进行了预处理，充分利用桥梁红外图像的二值化图和二值化方差图进行桥梁图像的分割，采用逐步渐进的方式进行桥梁的检测识别工作，并采用选定阈值的边缘增强手段突出了桥身的外缘线；利用可变，支数的线段模型去检测不同距离下不同大小的桥墩。文中还介绍了采用形心跟踪算法对识别出的桥梁进行自适应波门跟踪。

The methods for target recognition and tracking in complicated background based on the IR bridge image are described . Binary image and binary variance image are used to segment those IR bridge image in this paper, instead of the self-adapted accumulation which is used in traditional pro- cessing. Detecting and recognition bridge are performed step by step, including detection of river, bridge's arch and bridge , s pier . The edge-enchancing method of selected threshold is used to empha- size bridge' s arch , and the varied spot curve is used to detect bridge's pier . Finally, the bridge is rec- ognized and located . The bridge , s pier centre is tracked by a self-adapted gate technique according to the different distance.