Crystal Wafer Defect Detection Based On Digital Image Processing

Crystal, which is transparent and colorless, can be used as piezoelectric materials. Itcan function under certain frequency voltage. After being cut into single chip, it can vibrateat certain frequency. As the result of technological restrictions and the influences of otherfactors, it may have some defects under cutting process, such as crack, small angle, sometaint and so on. These small defects can damage surface properties of crystal materials andcause structural changes of the crystal materials, even have bad effects on the accuracy andreliability of the final products. In order to raise the passing rate, the crystal chips withsome kinds of defects should be inspected and removed before production pipeline.The subjects outline some methods commonly used in the inspection of crystaldefects, analysis and note advantages and disadvantages of these inspection methods.Considering the advantages and disadvantages of the method, the subject design a newkind of detection system of crystal chips' defects on the basis of the theory of digital imageprocessing. The article expounds the function theory of the hardware system in details,analysis and chooses the suitable camcorder, design a new kind of light source andimprove the lighting program. Finally the complete platform used for image acquisition iscompleted. The article expounds the basis theory of the message in Windows system,design a window used for sending messages and realizes the functions of the acquisition,judgment and return results. The camera captures the image signal and transfer to thecomputer under the function of USB. After a series steps, such as mode conversion,filtering, edge detection, image binary, feature extraction and classification detection, theimage of the crystal chips can be easily inspected and removed if it has some kinds of thedefects. Then the inspection results are returned. Feature extraction and classificationdetection is the key of the subject. The series steps belongs to the digital image processingand it can be realized with the Visual C++.NET.