Design And Implementation Of High Precision Industrial Spindle Size Detection Based On Machine Vision

With the continuous development of industrial modernization and the successive introduction of Industry 4.0 and National Intelligent Manufacturing 2025,industrial intelligent automation has become the general trend of industrial development.In order to realize the intelligent automation of the dimension inspection in industrial production,the machine vision-based dimension inspection technology came into being and began to be applied to the industry.The features of machine vision size detection are convenient,fast and accurate,which greatly liberate the labor force in industrial production,reduce the production costs and promote the overall industrial production efficiency.Therefore,research on machine vision size detection is receiving more and more attention.In this paper,the actual industrial axle core is taken as the research object,and the detection accuracy of minimum 0.002 mm is as the goal.In order to achieve a good overall design,the basic process of machine vision size detection and common algorithms are introduced in detail in Chapter 2.Through the overall analysis of machine vision size detection,the image processing algorithm is optimized and improved,which is the research focus of this paper to achieve high-precision size detection.In the image processing algorithm,this paper optimizes the key algorithms in the detection process to improve the final size detection accuracy.The optimization of the key algorithm is as follows: PCA combines the image contour feature to accurately target the target.After the positioning is completed,the optimized eight-neighbor tracking method is used to extract the edge based on the Moore boundary tracking algorithm.Then,based on the hyperbolic tangent edge model,use the hyperbolic tangent fusion correction linear model that is optimized by considering the actual factors to position the sub-pixels of the edge points to improve the data precision.Finally,when the straight line is fitted to the edge,the distance-weighted straight line fitting method based on the weighted idea optimization design is used to achieve the accurate fitting of the edge line in the image.For the common thread features existing in industrial axle cores,this paper discards the traditional corner detection method with large error,and extracts the inflection point of the thread based on the shape characteristics of the thread and the best value in mathematics to realize accurate calculation of the thread feature.In the design and implementation of the system,the two-sided telecentric lens and the 5 million CCD industrial camera constitute the lower position device.Using Windows system as the host computer development platform,using the Qt Quick framework of Qt development library to design and implement the host computer,and the C++ language combined with the OpenCV visual algorithm library implements the algorithm optimized in this paper,and uses Visual Studio 2015 to encapsulate it as a background call algorithm dynamic library.Finally,the test system is tested in terms of detection speed,detection accuracy and system stability,and detailed data analysis is carried out on the test results.Through optimizing the algorithm and analyzing the test results of the final implementation system,this design proves that it can meet the requirements of 0.002 mm detection accuracy and the actual industrial application conditions while achieving rapid and stable size detection.The original intention of proposing this design was realized.