一种基于特征的动态目标检测与跟踪算法  

Algorithm of checking and tracking moving object based on target feature

作　　者：李国栋[1] 刘春阳[1] 柳长安[1] 

机构地区：[1]华北电力大学计算机科学与技术学院,北京102206

出　　处：《哈尔滨工业大学学报》2005年第7期927-929,共3页Journal of Harbin Institute of Technology

基　　金：华北电力大学青年教师科研基金资助项目.

摘　　要：为解决图像处理特征目标物易受环境影响的问题,采用图像处理和颜色匹配的方法,提出一种基于特征的动态目标的检测和跟踪算法.首先对图像进行灰度化处理并获取图像梯度,确定目标物的弱可疑区,利用非完全特征匹配确定强可疑区,再进行完全特征检测确定目标物的位置,并根据运动物体两次移动位置计算出跟踪摄像器件的旋转方向和角度,最终实现对动态目标物的检测与跟踪.实验证明,该算法能够较好地完成动态目标物的检测与跟踪.A tracking algorithm of moving object based on characters is proposed by using image disposal and image matching. First, the image has been grayed and the image grads has been got by which the short suspicious area can be decided. Second, the object's position can be decided by entirety character's checking after the long suspicious area been decided through non-entirety character matching. At last, the testing and tracking of dynamic objects have been realized by computing the rotation orientation and angle of tracking photography apparatus. The experiment led up to the fact that this algorithm can successfully finish checking and tracking dynamic objects.

关 键 词：完全特征像素矩阵 弱可疑区 强可疑区 图像匹配 动态目标跟踪 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]