Research And Implementation Of Fast-NCC Registration And PCB Printing Correction Algorithm

The cost and stability of new energy,power equipment and other products are closely related to the level of PCB manufacturing technology.As a new type of PCB processing technology,digital printing technology has significant advantages in shortening the production cycle,improving product quality,reducing production costs,and reducing environmental pollution,and gradually replaces traditional manufacturing processes.However,large-format substrate processing leads to low efficiency of the visual positioning algorithm in the system,which seriously affects the system output;the plate has different degrees of distortion during the processing process,which leads to poor printing accuracy of the system and affects the output quality.To solve the above problems,firstly,the visual positioning algorithm in the system is improved to improve the accuracy of PCB image segmentation and reduce the computational complexity of the registration algorithm;and the image deformation algorithm is introduced to correct the processed image and eliminate the printing error caused by the distortion of the substrate.The main work of this paper is as follows:(1)Aiming at the problem of poor robustness of traditional PCB image segmentation algorithms and easy misjudgment of edge regions,resulting in low registration efficiency,a PCB image segmentation method that combines CRF and Gabor is designed.The Gabor filter is used to enhance the edge texture features,combined with the CRF model to link the feature context information with the spatial distribution,to overcome the local blurring of the image caused by noise,and to improve the robustness of the algorithm.Experiments show that this method has higher discrimination than traditional PCB image segmentation methods.(2)Aiming at the high time complexity of the traditional NCC registration algorithm applied to high-resolution PCB images,a fast normalized cross-correlation registration method(referred to as Fast-NCC)is proposed.Use statistical filtering to optimize the candidate point set;and combine dynamic programming and breadth-first search methods to adaptively select feature blocks.Experiments show that the registration accuracy of this method is about 99%,and the efficiency can be increased by 1.6 times compared with the traditional NCC algorithm.(3)Aiming at the projection deformation characteristics of hard PCBs during processing,design a silk screen layer image correction algorithm based on the moving least square method;for FPC showing local flexible deformation characteristics,design a correction algorithm based on thin plate spline deformation,and based on space the sparsity redundant control point optimization algorithm improves the correction efficiency.Experiments show that the accuracy of PCB printing is effectively improved after the correction algorithm is added.(4)On the basis of the above algorithm research,a set of PCB digital printing system is realized,which mainly includes hardware selection and software design.After field testing,the system's printing accuracy can reach±30?m;the processing time in the efficiency mode is 124m~2/h,and the processing time in the quality mode is 33m~2/h.