矩在复杂构件射线图像识别中的应用

The Application of Moment Features in Digital Image Identification of Complex Components

在工业射线成像在线检测中,由于装配引起的复杂构件在线位置的随机性,传送系统的机械误差等因素的影响,容易使被检测构件处于不同的状态,例如,倾斜状态,这些都成为影响检测精度的重要因素。针对这一问题,提出了一种矩特征检测构件状态的方法,该方法首先计算出图像中构件的质心坐标和二阶中心矩,然后求出构件的倾斜角度,根据倾斜角对处于倾斜状态的构件进行校正。此方法实现了自动识别前的被检测构件图像位置的标准化处理。

In the industrial on-line test of digital image, some factors such as complex component position＇s random caused by assembly and mechanical error of delivery system can easily make the tested components been in different states, like inclination, which are important ones affecting test accuracy. A method of moment features testing components′ states is put forward to solve this problem. First the components′ barycentric coordinates and second-order centered moments are calculated. Then the inclined angles are obtained, on the basis of which the inclined components can be rectified. This method realizes standardization of the tested components′ digital image.