Magnetic Tile Surface Defects Automatic Detection Of The Identification Method

To raising the quality of the product and the efficiency of the production,the on-line surface defects detection of the arc magnet seems to be important day by day.And because of the diversity and complexity of the surface defects of arc magnet, it's crucial to put forward a series of detection and recognization methods,and design a set of detection system to carry on the real time,on-line and nondestructive examination(NDE).This paper is on the basis of the production process of arc magnets and different part with different defects,and it carries out an intensive study of program designing and methods in terms of images acquisition,preprocessing,defects detection and classification of defects.The arc magnet surface defects detecion is on the basis of the computer vision theory,with the current image processing and recognization technology,selecting the detection and recognization technique and algorithms.And mean time,we improved the corresponding algorithms and designed several methods for the defects detection and recognization.Specific activities include the following:1. Study the methods of image acquisition, storage and management,so as to conduct the next image processing management;2. With the arc magnet image features,we did the experiments,to denoise,sharpen the defects,smooth the image,enhance the edge,region segmentation and carry out the dilation and erosion operations.3.Designed the algorithm of thresholding for the image segmentation,with the manual automatic or manual threshold based on the specific features to carry on the image binarization,to realize the automatic defect features extraction;4. Designed the algorithm of detection and recognization,carried out the simulation with MATLAB 7.1,based on the surface defects features;5.Making use of mixed programming by MATLAB7.1 and Visual C++ 6.0 to simulatie the defects detection algorthm,and realized efficient detection system.This paper is mainly to study how to acquire images,image partition,denoising,sharpening,ima-ge binary,mathematical morphology,defect features extraction,defects detection and recognition of the arc magnet in detail.We carryed out the simu-lation experiment,and obtained satisfied initial results finally.