基于图像的机器人指尖跟踪过程仿真分析  

Simulation Analysis of Robot Fingertips Tracking Process Based on Image

作　　者：卫朝霞[1] 唐波[1] 

机构地区：[1]电子科技大学成都学院,四川成都611731

出　　处：《计算机仿真》2015年第2期414-417,458,共5页Computer Simulation

摘　　要：对机器人指尖运动轨迹的视觉跟踪是智能机器人手势识别研究的关键。在多自由度机器人指尖跟踪过程中,由于自由度较高,指尖运动速度很快,指尖在运动终端的变化细微、灵活。利用传统的算法进行指尖图像跟踪时,需要连续、形状运动关联的特征才能完成跟踪,在较高自由度运动的情况下,指尖的运动终端图像特征的突变性因为无法以连贯的图像特征形式被捕获而被忽略,导致指尖跟踪图像与指尖终端的实际运动过程存在较大偏差,跟踪准确性下降。提出一种新的机器人指尖图像定位跟踪方法。针对采集到的指尖跟踪图像进行初始化处理,获取图像的高频部分和低频部分的向量,针对上述向量进行小波变换处理,实现指尖图像的去噪处理,提高了图像的质量。对指尖跟踪图像像素进行归一化处理,为指尖跟踪提供了准确的数据基础。利用线性分析的方法,对图像中的指尖运动轨迹区域进行准确分割,实现了指尖运动轨迹的准确跟踪。实验结果表明,利用改进算法进行指尖跟踪,能够有效提高跟踪的准确率。The visual tracking of the robot fingertips＇ trajectory is the key to intelligent robot gesture recognition research. In the paper, a new robot fingertips image locating and tracking method was proposed. The collected fin- gertips tracking images were initialized to obtain the vectors of the high frequency part and the low frequency part of the images, then the obtained vectors were processed with wavelet transform, to realize the denoising processing of fingertips images and improve the quality of the images. Pixels of fingertips tracking images were normalized to pro- vide the accurate data base for fingertips tracking. By using the linear analysis method, the areas of fingertips trajec- tory in the images were accurately segmented to realize the accurate fingertips trajectory tracking. The experimental results show that the improved algorithm can effectively improve the tracking accuracy.

关 键 词：图像处理 指尖跟踪 运动轨迹 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]