Wavelet Basis Selected Of Weld Defect Image Processing

Welding process is an important link in the process of shipbuilding, which is the key to guarantee the quality of ships. During the welding process affected by various factors, usually there are many welding cracks, slag, porosity, penetration, incomplete fusion appeared in the welding. If the welding defects exist in weld, they would cause the structure fracture, leakage, and even cause ship sunk. So it has the important practical significance and economic value to effectively detect weld defects. X-ray imaging method is common and very effective detection means. Traditional artificial detection technology was affected by the technical personnel qualify and experience, which detected low efficiency and operating complexly and was difficult to save film material etc. Computer-assisted assessment chips can not only improve work efficiency, effectively overcome artificial errors and missing pieces due to judging depending on technical personnel experience and the different external factor, but also make the evaluation process objectification, scientific and standardization.However, the defect of traditional method of image processing algorithm itself would exist, and it did not process ideal effect. So the new image processing algorithms were forced to appear, and wavelet analysis of its unique time-frequency localization characteristics, characteristics and directional characteristic scales are widely applied to various fields of science. Wavelet transform was also named the most powerful tool of signal and image processing. With the development of the theory of wavelet analysis, the wavelet base function species is also more and more, the wavelet base function directly affects the processing results. This thesis studies the theory application of wavelet analysis in the weld defect image processing system and how to select the wavelet base.This thesis mainly studied two aspects research work of the image de-noising and edge detection for weld defect image features. Firstly it introduced the basic theory of wavelet analysis, analyzed the basic theory of wavelet, multi-scale characteristic and Mallat algorithm. Then the analysis of the characteristics of weld defect image digitization, observed on weld image gray, a grey value, low contrast, blurring and defects in the small target. With the traditional image de-noising and edge detection, wavelet method was compared. Mainly studied wavelet function, analyzed the properties of orthogonal wavelet and tight teams, symmetry, regular and wavelet. It is summarized the selection principle of wavelet base and the selection of optimal wavelet base for weld image. And then there is better treatment effect in the weld image de-noising and edge detection. Finally it did some experimental analysis in the weld image de-noising and image edge detection, software system using common 39 species of wavelet function to do experimental analysis. In the final determination of weld defect image de-noising and edge detection, the optimal wavelet base was selected. Wavelet analysis method is used to detect more effectively and more accurately edge information better, improving the defect rate.