改进概率Hough变换在温度传感器焊接工艺中的应用

Application of improved probability Hough transform inwelding process of temperature sensor

为了解决温度传感器焊接中负温度系数(NTC)热敏电阻倾斜角度不一,无法实现自动化上料问题,提出一种改进概率Hough变换算法对电阻倾斜角度进行检测。首先通过直线边缘像素点梯度方向与倾斜角之间的关系,利用梯度方向中位数对概率Hough变换的检测角度进行限制,减少角度检测范围。其次在直线像素点的搜素过程中增加对直线方向邻域范围的像素点检测,同时改变间断点的判断条件,并利用最小二乘法对点集中的像素点进行直线拟合。实验表明:改进的概率Hough变换提高了直线的检测效率,保证了检测参数的统一,而且提高了检测精度,能够满足实际的焊接自动化需求。

In order to solve the problem of the different inclination angles of the negative temperature coefficient(NTC)thermistor in the production of temperature sensors,the automatic feeding cannot be realized,an improved probability Hough transform algorithm is proposed to detect the inclination angle of the resistance.Firstly,through the relationship between the gradient direction of the pixel point of the straight edge and the tilt angle,the median of the gradient direction is used to limit the detection angle of the probabilistic Hough transform to reduce the angle detection range.Secondly,in the process of searching for straight line pixels,the detection of pixels in the neighborhood of the straight line direction is added,and the judgment conditions for discontinuous points are changed at the same time,and the least square method is used to fit the pixels in the point concentration to a straight line.Experiments show that the improved probability Hough transform improves the detection efficiency of straight lines,ensures the unity of the detection parameters,and improves the detection accuracy,which can meet the actual welding automation requirements.