Application of Machine Vision and Machine Learning to Predict the Focal Position in Laser Machining

A new approach is proposed for focus positioning on a curved surface based on image processing method during laser machining. By analyzing the image in the CCD camera, the defocal direction and interval along z axis, rotation about x and y axes are obtained simultaneously. The images from the CCD camera are collected into a data set with the appropriate labels. In laser machining, the data from the CCD camera are acquired and matched with the above-mentioned data set. Subsequently, the one that is the closest to reality will be chosen. The more data in the data set, the higher the accuracy of the method will be. This is a fairly modern approach. It is expected that this method can be broadly employed for laser processing.