Study On The Image Detection Of Ceramic Water Valve Plate's Surface Defects

A ceramic water valve plate,as an important part of the water tap,is a key component in determining the quality of water tap.The quality detection on the ceramic valve plate's surface has become a key step to control the quality of water tap.Traditional detection of the ceramic valve plate's surface defects is done artificially by workers.Its efficiency is low and the work is too hard for workers.Furthermore,some minor defects maybe are skipped over.The detection of the ceramic water valve plate's surface defects based on the image processing method is rapid intelligent,accurate,reliable,and can improve the detection efficiency.Therefore,to carry out the study of high accuracy and efficiency ceramic water valve plate's surface defects detection method has important significance and practical value.The detection system of ceramic water valve plate's surface defects is constructed which consist of the light photos,CCD camera,image acquisition card and computer image processing software by using digital image processing technology.The image pre-processing,image segmentation and image recognition are carried out on the ceramic valve plate's image in detail.And the software and hardware design of the detection system of ceramic water valve plate's surface defects have been finished.The image filtering algorithm adapting to the ceramic water valve plate is researched by analyzing the noise characteristic of ceramic water valve image,and then the denoising scheme is proposed.The morphological edge extraction operator is chosen to extract the edge of ceramic water valve plate by analyzing different edge detection operators.The Hough transform is adopted to calculate the tilt angle of the image of ceramic water valve plates to correct the possible tilt of the ceramic water valve plate in the defects detection process.The Otsu threshold binarization from different binarization threshold algorithms is selected as a reserve algorithm in the following detect detection process.In order to solve the problem that the binarization threshold algorithms cann't distinguish the scratch defect,a binarization processing method based on support vector machine was proposed,which can be used to calculate the characteristic parameters of the scratches.The Hu invariant moment feature,gray level co-occurrence matrix and Gobar texture feature of ceramic water valve plates are extracted to research the featureextraction methods of ceramic water valve plate's image.The support vector machine model is designed to detect clipped edge defect of ceramic water valve plates by extracting Hu invariant moments of samples.And the support vector machine model is designed to detect the scratch defect of ceramic water valve plates by extracting the Gobar texture features of samples.The parameter analysis method of defects on the ceramic water valve plate surface is designed to quantificationally calculate parameter values of the clipped edges and scratches.Finally,the experimental device of the detection system of ceramic water valve plate's surface defects is constructed,and the hardware and software is jointly debugged.The following experiments have been conducted: 1.Under different light intensity,the images of ceramic water valve plates are collected and the experiment is carried out to test the accurate rate of clipped edges defect.The results show that light source has less influence on the accuracy rate of clipped edge defect detection;2.Under different light intensity,the images of ceramic water valve plates are collected and the experiment is carried out to test the accurate rate of scratches defect.The results show that light source has great influence on the accuracy rate of scratches defect detection;3.The other type of ceramic water valve plates is adopted for testing applicability of the system.The results show that the system has good applicability for the detection of different type of ceramic water valve plates.