Detection methods for surface micro defection onsmall magnetic tile based on machine vision
                                                                   HU Hao1, LI Junfeng1, SHEN Junmin2,3
(1.Department of Automation, Zhejiang SciTech University, Hangzhou 310018, China; 2.Department ofElectronic Information Engineering, Zhejiang SciTech University, Hangzhoug 310018, China;3.Jinhua yibo technology Co., Ltd., Jinhua 321015, China)

Abstract: Aiming at the characteristics for unclear image,difficult detection,low comparison degree, complex texture background, uneven brightness, small defect area and many defection types for small magnetic tile, a new visual detect method for surface micro defection in small magnetic tile was proposed . Firstly, exterior defect categories were classified into three categories through analysis for gray level between defective area and normal area, variance between grayscale and defect form based on image influence from arc surface, angle of chamfer and defect area. Secondly, respective defection abstraction methods were designed based on three kinds of image characteristics for surface defect, defect morphological characteristics and relationship with background. At last, experimental analysis were carried out with developed experimental devices based on various sunlight, specification and defection type. The results indicate that the surface defection extraction algorithm for small magnetic tile is in good stability and strong robustness. Defective area on surface of small magnetic tile can be abstracted correctly and promptly and related accuracy rate as 93.5%.

Key words: small magnetic tile; detection on micro defection; grayscale; surface defection; machine vision