Research On Defect Ultrasonic Image Mosaic Method Of Weld Of Plate

Ultrasonic testing and imaging technology is widely used in the detection of weld defects of plate to provide a basis for defect detection and evaluation according to the information of the defect such as the size and location.Due to the limited range of a single scan of the ultrasonic sensor,the defect distribution of the entire weld cannot be given,which affects the overall evaluation of the plate defects.Aiming at the problem of poor global defect images when large inspection areas or large defects in a single inspection,the dissertation has carried out a research on the method of ultrasonic image mosaic of weld defects,and provided reference and information for the overall evaluation of sheet weld.The dissertation uses the ultrasonic phased array total focus method to image the weld defects of the plate,which is used as the original image for image mosaic.By analyzing the color,texture,contour and frequency domain distribution characteristics of adjacent ultrasound images,an improved ultrasound image registration algorithm based on Scale Invariant Feature Transform(SIFT)is proposed.The algorithm realizes coarse matching by finding the feature points of the same level in different images,and then performs fine matching through the Euclidean distance histogram method to improve the correct matching rate of ultrasonic image feature points.Then,according to the adaptive threshold random sampling consensus algorithm,the transformation matrix between different images is obtained,and the matched images are transformed into the same coordinate system to realize the registration of ultrasound images.Based on the analysis of the principle and characteristics of the image fusion algorithm,a weighted fusion algorithm based on the best seam is proposed,which realizes the fusion of ultrasound images,and finally completes the image stitching,which is evaluated by the image stitching evaluation rule.The main research work and conclusions of the dissertation are as follows:1)Firstly,analyzed the principle and development status of ultrasonic detection and imaging technology for plate weld defects at home and abroad,and then the principle and characteristics of visual image and medical ultrasound image splicing and development status are analyzed.On this basis,according to the characteristics of defect ultrasound images in the industrial field,the purpose and significance of defect ultrasound image mosaic are explained.2)Analyzed the principle of ultrasonic phased array total focus method,es Tab.lished the ultrasonic phased array total focus method model of sheet weld defects,set the simulation parameters according to the characteristics of sheet weld defects,obtained the ultrasonic phased array total focus method image of the simulation and evaluate the imaging effect,and the ultrasonic phased array total focus method was determined to obtain the original ultrasound image for mosaic.3)The effect of image mosaic depends to a large extent on the amount of information contained in the image.Because the amount of information contained in the industrial ultrasound image is not rich,coupled with the low image resolution,the ultrasound image feature distribution is not obvious,increasing the difficulty of image mosaic.In order to select an appropriate image registration algorithm,analyze the color,texture,contour and frequency domain features of the ultrasonic defect image,which shows that the color and frequency domain as a global feature quantity is not sui Tab.le for the feature description of the ultrasonic total focus defect image;the correlation and contrast in the texture are differences in the ultrasound images of different defects,which are sui Tab.le as the basis for defect classification and recognition;the contours of ultrasound images can locate and extract the shape and location of defects in the image.Therefore,the geometric features of ultrasound images are more sui Tab.le for their registration basis.4)Through the analysis of the ultrasonic defect image characteristics,an improved SIFT-based ultrasonic image registration algorithm is proposed.In the feature matching stage,the matching points are first searched for on the same level to perform rough image matching to avoid mismatches between the same defects at different depths;Then,the matching pairs are fine-matched by the Euclidean distance histogram method to further improve the correct matching rate of the ultrasonic image feature points.Considering the low resolution of the ultrasound image itself and the few image matching points,the rigid body transformation matrix is selected as the transformation matrix between different images.Random sample consensus algorithm with adaptive threshold is proposed to obtain the rigid body transformation matrix,and different images are converted to the same coordinate.then the registration of ultrasound images is realized.5)The basic principles of the best seam search fusion method and the smooth filter fusion method are compared and analyzed,and the gradual in and out weighted fusion algorithm based on the best seam is proposed to fuse total focus ultrasound images,Combined the subjective and objective evaluation phases image mosaic evaluation criteria.After analyzing the image effects of different fusion algorithms,the results show that the performance of the fusion algorithm proposed in the dissertation is better than other fusion algorithms.6)An ultrasonic inspection experiment system for plate weld defects was built.The standard defect test blocks for grooves and through holes,and the plate weld test blocks were tested for defects and image mosaic,and the ultrasonic panoramic images were obtained.The results showed that the ultrasonic panoramic images were obtained can accurately locate defects,avoiding the negative impact of defect image dispersion on evaluation during a single scan,and can obtain the overall distribution of large areas or large defects.Through multiple test statistics,the success rate of splicing can reach98.3%.