Key Techniques For Defect Detection Of Small Hole Inner Surface Based On Machine Vision

In the recent years,a variety of defect detection techniques in industry have been developed continuously and especially with the improvement of image sensor and computer hardware,the machine vision technology is more and more widely used in the intelligent detection of industrial field.There are still some problems and difficulties in the defect detection of small hole inner surface,owing to the limitation of the dimension and the complexity of the cylindrical inner surface.Hence,most of the traditional detection methods are no longer applicable.At present,there are two main methods to detect the small hole inner surface.One is artificial observation method with a magnifying glass,and the other method is to insert the industrial endoscope into the small hole.However,the manual measurement is time-consuming and laborious,and has a low detection efficiency.The general industrial endoscope can't be inserted into the small hole with the diameter below 4mm owing to the limitation of the structure size,and the optical fiber endoscope has a low resolution so that we can't get the clear image of the small hole surface.Therefore,in this paper,a kind of micro-optical system at oblique incidence based on the machine vision is designed to detect and identify the defects on the inner surface of small holes.The defect detection system of the small hole inner surface we proposed mainly includes two parts: the overall design of the defect detection system and the research of the small hole inner surface detection algorithms.The former includes: the design of system structure;the choice of key hardware;the determination of light pattern;the design of system software and so on.The latter includes: the multi-focus image fusion algorithm to obtain the fully focused image of the small hole inner surface under the constraint of the micro-optical system with a limited depth of field;the correction algorithm for images of the small hole inner surface which are collected at oblique incidence;the extraction algorithm of defects and recognition algorithm according to the geometry characteristics.Based on the above algorithms,we can obtain the fully focused and clear flat image of the small hole inner surface with little distortion,and realize the extraction and classification of defects on the inner surface of small holes.On the basis of the above research,we realize the defect detection of small holes with diameter of 2~4mm and medium ratio of length to diameter.Experimental results show that the proposed method is feasible,and it avoids the problem of low efficiency of manual inspection,and breaks the limitation of traditional inner surface detection method for such small holes.