基于改进光流算法的机器人目标跟踪系统  

A Moving Objects Tracking System of Robot Based on Modified Optical Flow Algorithm

作　　者：华贵山[1] 梁栋[2] 葛浩[1] 

机构地区：[1]滁州学院机械与电子工程学院,安徽滁州239000 [2]安徽大学电子信息工程学院,合肥230039

出　　处：《控制工程》2014年第5期634-637,共4页Control Engineering of China

基　　金：国家自然科学基金(61172127;11071002);安徽省高等学校自然科学研究项目(KJ2012A212)

摘　　要：基于计算机视觉的机器人运动目标检测与跟踪,就是建立起一种机器人视觉与电机驱动相关联的系统。光流算法在此类系统中有着广泛应用,但是求取所需稠密光流场的运算量过大是其明显的缺点。为减少计算时间,提高跟踪系统响应速度,引入均值平滑算法对传统H-S(Horn和Schunck)光流算法进行改进,并在此改进算法基础上建立起目标跟踪系统。通过此系统,机器人可以根据采集到的图像的光流场变化来检测运动目标,再通过对光流场的奇异值分解,对跟踪系统模型所需参数进行估算,并驱动机器人做出相应动作,保持对目标的跟踪,从而使机器人对周围环境变化做出及时、准确的动作。经过实验证明改进后的光流算法有效的减少了计算时间,增强跟踪系统的实时性能。Moving objects detection and tracking based on the computer vision means to build a system with robot vision part and motor driver part. Optical flow algorithm is widely used in such systems, but it has an obvious disadvantages： complicated calculation. According to the combination of H - S （ Horn ＆ Schunck） optical flow algorithm with smoothing average algorithm, we set up a target tracking system of the robot to reduce the operation time. According to the optical flow field＇s changes derived from the collected images, the robot detects moving targets, and estimates the targets＇ parameters with singular value decomposition of optical flow field, then drives motor, to realize the target tracking. This system can make the robot move timely, accurate to the changes of environment. The modified optical flow algorithm overcomes the shortcoming of calculation, and makes the robot move timely.

关 键 词：运动目标检测与跟踪 H—S光流算法 均值平滑算法 计算机视觉 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]