Research On Mark Point Localization Algorithm Of Laser Processing Based On Digital Image Processing

With the development of the digital information technology,people start to research the automatic identification and positioning technology which is based on digital image processing,and make the technology applied to industrial production to improve the efficiency.In the process of acquisition and transmission the digital image is inevitably polluted by the noise,what is more serious is that the noise can affect the accuracy of the identification and positioning of image mark points.Aiming at denoising the mixed noise of gauss noise and the salt and pepper noise,three steps of the denoising methods are discussed in this paper.First,the existing algorithm of image denoising in frequency and airspace domain are studied,and found that gauss noise is effectively suppressed by wavelet threshold denoising and the salt and pepper noise is effectively suppressed by median filtering.And then,an improved median algorithm is put forward because the details of image after median filtering are still vague.Last,a hybrid denoising algorithm which combines wavelet threshold denoising with the improved median filtering is put forward to deal with the mixed noise,and it is proved to have better effect in denoising through the experiments.In order to realize the precise positioning of image mark points,edge detection is essential after denoising.The common edge detection operators are analyzed in detail in the paper,include the advantages and disadvantages of them.And then,a kind of progressive Canny edge detection based on Otsu threshold is put forward.It is proved to have better adaptability as well as keep the accuracy.In the paper,the basic idea of Hough transform and its improvement algorithms are studied,also the identification methods of the linear and circular mark are analyzed.Based on summarizing the predecessors' research,an improved Hough transform algorithm using the edge gradient information is put forward to detect the line,and an improved algorithm using limit strategy of initial points is put forward to detect the circular mark point,then in order to improve the positioning precision of the circular center,the least square method is used to get subpixel level positioning accuracy.