基于分形特征的PCB机器视觉定位

提出一种基于印刷电路板（PCB）兴趣区图像特性曲线分形特征的定位新方法，由PBC兴趣区图像生成一组可观测特征曲线，抽取曲线中最能反映图像转动角度信息的曲线段为研究对象，从曲线的形成机理分析其分形特征，运用信号处理和随机统计理论处理曲线，得到能反映图像转动角度的曲线分形维数，建立起了可观测特征曲线的分形维数和图像位姿转动角度间的对应关系。

Angle Locating of Printed Circuit Board Based on Fractal Theory

A novel approach of PCB (printed circuit board) image locating was initiated. The method focuses on the interesting area of rectangle mark in PCB. The white pixel number along each column in the interesting area is decided, then a curve is constructed using the image column number as x coordinate and the white pixel number of each column as y coordinate. Because of the limitation of CCD sensitive, some fractal features surely appear in the curve. The theory of signal processing and random statistics are used to process the special parts in the curve, and then the fractal dimension of the curve is acquired. The theory relation between the fractal dimension of the curve and the turning angle of the image can be described by using a straight line. According to the relation, whenever we get the fractal dimension of a curve, we can quickly and accurately know the angle of the image. The research shows the new method based on the fractal theory is powerful for the problems of angle locating of PCB and some other fields related.