On-machine Inspection Method Of Laser Cleaning Quality Based On Machine Vision

Laser cleaning technology has many advantages such as non-contact processing,high cleaning accuracy,small damage to the substrate,and green environmental protection,and its rapid development and application fields are more and more extensive.In order to achieve automatic precision cleaning,the demand for accurate and rapid detection and evaluation methods of laser cleaning quality is becoming more and more urgent.The existing detection methods are difficult to evaluate the overall cleaning quality of large parts,and most of the detection methods are offline detection,which consumes time and manpower.Therefore,an on-machine detection method for laser cleaning quality is urgently needed to evaluate the quality after cleaning.Aiming at the practical problem that the laser cleaning quality detection method is complex and difficult to realize on-line detection,a laser cleaning quality on-line detection method based on machine vision is proposed.Construction of laser cleaning on-machine vision detection system;Using image enhancement technology to improve image quality;The detection image is spliced by image stitching technology,and the pollutant area is extracted by color extraction technology and the proportion area is calculated.The large area detection and quality evaluation on the machine are realized.By extracting the regional contour of residual pollutants,zoning path planning and cleaning.The specific research contents are as follows:(1)Aiming at the problem of low image quality caused by uneven illumination in laser cleaning environment,median filtering is used to remove image noise.After analyzing the shortcomings of existing image enhancement algorithms such as color deviation,distortion and detail ambiguity,a Retinex image enhancement algorithm based on two-dimensional gamma function is proposed,which improves the quality of the image,solves the problem of image loss caused by insufficient illumination,and is more suitable for laser cleaning detection.(2)In order to realize the large-scale detection of laser cleaning and quantitatively evaluate the detection results,SURF algorithm is used to splice the detection image,and the color extraction based on HSV color space is adopted.The pollutant area is accurately extracted,and the median filtering method is used to remove the interference pixels.This method realizes the visual detection of the surface quality of laser cleaning,improves the detection accuracy compared with the gray threshold extraction method,and provides method support for the quantitative evaluation of the laser cleaning quality of pollutants on the surface of steel.(3)In order to improve the cleaning quality,the edge extraction method for the residual pollutant area in laser cleaning was studied.The edge extraction was regularized and the partition path was generated by using the partition envelope regularization method.The temperature field of the partition path planning was simulated by using the finite element analysis software.The experimental results show that the residual pollutant area is well removed and the surface cleanliness is significantly improved,which verifies the feasibility of this method and provides method support for path planning of laser cleaning.