Research On Visual Inspection Of Powder Metallurgy Gear

The powder metallurgy gear has the advantages of low cost,simple manufacturing processes and diverse shapes,which is widely used in industrial transmission.In this paper,in order to detect the defects and the size of the inner bore,the gear image inspection and sub-pixel edge location method is studied,has a practical significance for automatic detection of gear.The gear contour is extracted by image segmentation and boundary tracking method.The contour classifier is constructed by analyzing the related parameters(roundness,perimeter and area).Based on the principle of least square and RANSAC algorithm,the center point calculation formula is derived.the calculation formula of center point is deduced.The causes and forms of gear defects(notch,surface and residual)are analyzed,and the testing requirements.Aiming at the defect of notch,the method of extracting is discussed,and the method of subdivision gray area analysis is proposed.For the surface defection,the surface area is expanded into rectangle by geometric relation,the edge detection and area growing method are used to inspect the target area.For the defection of residual powder,the one-dimensional function representation of gear contour is discussed.The relevant characteristic parameters of the function is extracted,the normalized discriminant expression is constructed.The characteristics and errors of the algorithm are analyzed according to the data results.According to the grayscale imaging characteristics of the ideal step edge,a sub-pixel localization method of constructing fourth-order curve in the local area is proposed.According to the gradient and gray relation of the pixels,the curve expression is deduced.The deviation between the actual target contour and theoretical model is analyzed.A method of smoothing the sub-pixel coordinates of B-spline curve is proposed.Compared with fitting method and ZOM method,the experimental results show that the accuracy of this algorithm is 0.5?m,the error of repeated measurement is within 1?m,which is better than the other two methods and has better local edge expressive ability,can realize the detection of gear inner bore discrimination.