摘要
为了满足自动化装配线中工件自动上料作业的需求,提出了一种基于改进形状上下文特征的工件识别和定位算法。首先对图像进行预处理,然后进行阀值分割提取轮廓图像,利用改进的形状上下文特征检测算法在已检测出的轮廓中识别工件轮廓。与传统形状上下文特征算法不同的是改进形状上下文特征检测算法采用直方图结合Harris角点进行特征生成与匹配,加快了匹配速度。实验结果表明,本方法具有较高的精度,对环境变化有一定的适应能力,能较精准地对工件进行定位。
To meet the requirement of automation assembly line,in this paper,we use the six DOF industrial robot and stereo vision system to recognize and locate the object.We use the Harris to extract the corners as the sampled points,and then build the shape context descriptor based on fuzzy rules.The shape recognition is achieved by evaluating the cost function.In this paper,we use the four-parameter affine model to find the corresponding pixels,which is simple to implement.The experimental results show that the proposed method can satisfy the need of the vision of assembly robot and keep robust against environment change.
出处
《机械工程与自动化》
2016年第6期40-41,44,共3页
Mechanical Engineering & Automation
基金
杭州市创新链产业链重大科技创新项目(20132111A04)
杭州市重大科技创新项目(20142013A56)
关键词
形状上下文
工件识别
工件定位
3D重构
shape context
object recognition
object localization
3D reconstruction
