Study On Automatic Inspection And Recognition By X-ray For Cathode Carbon Block Defect

The station of carbon material will get more and more important with the development of industry technology. There is an urgent need for methods in the testing of interior defect for many manufacturers and expenditure departments to control and improve their product quality. However, the SNR of detective signal is low, which make it difficult to detect effectively interior defect. So, it is need to look for effective methods and means. The key technology of automatic inspection and recognition is deeply researched taking cathode carbon block as the research object in the thesis. Automatic inspection system is exploit. Major content of thesis is as follows:In chapter 1, the author gives a comprehensively summarizing to present research status of interior defect nondestructive testing, the contents of dissertation is introduced and the key problem of automated detection of defect is pointed out.In chapter 2, various detective methods of carbon product are analyzed, ultrasonic and x-ray detection are emphasized. The author puts forward that Ultrasonic impact-echo method is not suitable to detect interior defect of carbon product, Ultrasonic speed method is fit for inspecting homogeneous structure, but it can't appraise quantificationally defect. At the same time, the author advances that x-ray detection method is fit for appraising quantificationally interior defect of carbon product by successfully using image processing technology. Exposure quantity curve is erected by means of experiment, the curve has important practical significance to how to choose voltage based on the thickness of carbon product.In chapter 3, the construction method of the compactly supported biorthogonal wavelet is studied, and the biorthogonal wavelet is constructed successfully that suit for x-ray image of cathode carbon block. Object boundary extraction algorithm is designed in order to remove the disturb of background. The denoising and enhancement methods of images based on wavelet transform are studied, the algorithm of histogram and wavelet fusion enhance and the adjust method of image asymmetry based on...