基于视觉激光切割厚板切割状态监控方法  

Method of Monitoring Cutting Point State in Laser Cutting Plank Process Based on Vision

作　　者：高士友[1] 马雷[2] 张润生[2] 

机构地区：[1]燕山大学机械工程学院,河北秦皇岛066004 [2]燕山大学车辆与能源学院,河北秦皇岛066004

出　　处：《中国激光》2012年第2期91-97,共7页Chinese Journal of Lasers

基　　金：国家重大科技专项(2009ZX04003-032);国家自然科学基金(50871022)资助课题

摘　　要：利用光纤镜头和彩色工业摄像机实时采集激光切割厚板中切割点的图像,从彩色图像中分别选取蓝色、绿色和红色通道图像,分析各通道图像的特点和切割点的几何形状特征。首先以激光焦点在图像中的位置为中心建立坐标系,以x轴方向为起始,45°为间隔向8个方向搜索激光切割区域的边缘点,根据边缘点到坐标原点的距离信息确定激光切割方向和切割顶点;建立边缘识别用抛物线模型,根据边缘处存在灰度特征、梯度特征和方向特征设计识别目标函数,识别切割顶点两侧边缘,进而识别整个切割点处的几何形状。实验表明识别方法具有良好的适应性、准确性和实时性。Fiber of optical lens and color industrial video camera are used to capture image of cutting point in laser cutting process.Blue channel,green channel and red channel images are adopted to analyze cutting point geometry shape from color image.The position of laser beam focal point is used as origin of coordinates and the coordinate system is set up.The x axis is original direction and interval is 45 degree,the edge points of laser cutting region are searched from eight directions.According to distances from origin of coordinates to edge points,the laser cutting direction and cutting peak are fixed.Parabola model is set up,and the objective function is designed according to gray feature,gradient feature and direction feature,and the objective function is used to identify edge at bilateral cutting peak,and then cutting point geometry shape is identified.Test shows that the method has preferable adaptability,accuracy and real-time quality.

关 键 词：激光技术 激光切割 厚板 边缘识别 目标函数 

分 类 号：O439[机械工程—光学工程]