Research On Inspection Technology Of ICF Target Assembly Based On Machine Vision

Inertial Confinement Fusion(ICF) is one of the most promising way to resolvethe energy crisis. Currently, Laser Inertial Confinement Fusion is the hottestresearch topic of ICF. The manufacturing and assembly of ICF target are keytechnologies of ICF. The target is very small, complex and its materials are special.as a key component of ICF, the accuracy of manufacturing and assembly must bevery high. Because of the high cleanliness requirement, and the materials that areeasy to deform, it cann’t use the traditional contact method for detection. This thesisdeeply research the inspection technology of ICF target assembly at home andabroad, proposes a non-contact, high precision, high efficiency testing way in theprocess of ICF target assembly based on multi-dimensional machine vision, anddevelopes a set of ICF target assembly inspection system.The main research includes:(1) With multi-dimensional machine vision and theLaser displacement sensor, design and develop a set of ICF target assemblyinspection system that can give the real-time images, as well as angle and positionof ICF target, to achieve the target multidimensional detection. Analyze theworkflow of the system and complete the measurement software. Finally, achievethe target’s automated assembly.(2)This thesis give a systematic study of the target’simage processing methods, include: study the image enhancement algorithms, usingHistogram equalization method to improve the contrast of images; study the regionsegmentation method, using Threshold segmentation with iterative algorithm tosegment images for the first time; study the edge detection algorithm, using Cannyoperator to achieve the precise edge detection in the images; study the contourrecognition method, using Hough transform edge extraction algorithm to detect thelinear and circular contours in the images; study the precise positioning method,using Sub_pixel subdivision technique that can improve the precision ofmeasurement to fit lines and circles with a sub_pixel high precision; study the NCCtemplate matching and Shape-based matching, so we can complete automatic targetrecognition and positioning, even in the situation that exists noisy interference,illumination change or parts occlusion.(3)With the Laser displacement sensor, thesystem can give the angle of thin-walled and the blocked target, but the method ofimage processing can’t deal with these problems.The experimental results and the analysis of accuracy showed that the accuracyof ICF target assembly inspection system can meet the high-precision assemblydetection, its overall positioning accuracy is up to2.7m, even up to1.7m whenthe motors aren’t moving. The angle accuracy with image processing method is up to0.056°, and the angle accuracy with the Laser displacement sensor is up to0.191°.