Research And Application Of Sub-pixel Edge Detection Algorithm For Precision Shaft

Precision shaft is the core components of various rotating equipment,and its quality is decisive for the reliability and durability of equipment operation.At present,most manufacturers still use traditional contact measurement methods to detect,the detection effect is greatly affected by subjective factors,and is not suitable for production automation.The detection technology based on machine vision has the advantages of non-contact,high speed,moderate precision,strong anti-interference ability and online non-destructive testing.It is very suitable for the needs of modern manufacturing automation and has been widely used.In the precision shaft visual detection,accurate extraction of the precision shaft edge is the key to achieve high precision and rapid detection.With the development of precision manufacturing technology,the industrial field has higher and higher requirements for detection accuracy,and the traditional edge detection method can only achieve pixel-level accuracy,it is difficult to meet the needs of higher-precision detection in actual production.Therefore,it is urgent to study various sub-pixel edge detection methods with higher precision,and continue to improve according to actual needs.Based on the previous researches,this thesis combines the development of pixel-level edge detection algorithm,and proposes a sub-pixel edge detection algorithm combining the coarse and precise location,and it has been initially applied in the National Natural Science Foundation Project of 'Research on precision shaft high precision vision measurement method based on compressed sensing'.Experiments show that the algorithm can not only improve the detection speed,but also achieve higher detection efficiency.The main research work of this thesis is as follow:(1)The domestic and international research status of machine vision technology and edge detection algorithm are reviewed.The chapter content and structure of the thesis are given.(2)The image preprocessing algorithms necessary in the thesis are introduced,and some simple simulation analysis is carried out.At the same time,the characteristics of various algorithms are expounded and compared.(3)The existing common pixel-level edge detection algorithm and the implementation principle and characteristics of sub-pixel edge detection algorithm are introduced.(4)A sub-pixel edge detection algorithm based on coarse and precise location is proposed.Firstly,aiming at the shortcomings of Roberts algorithm sensitive to noise,an 8-neighbor improvement method is proposed,which effectively overcomes the defect that Roberts algorithm don't have the anti-noise capability.Then,owing to the Zernike method requires manual selection of thresholds,which affects the detection efficiency,etc.Combined with Otsu method,the threshold select method is improved,and the improved method can effectively improve the detection efficiency and also achieve a better detection effect.(5)A machine vision-based edge detection hardware platform was built,and a.machine vision-based edge detection system was developed.