Font Size: a A A

Research On Reconstruction Method Of Cylindrical Workpiece Topography Based On Multi-focus Image Sequence Fusion

Posted on:2021-04-05Degree:MasterType:Thesis
Country:ChinaCandidate:Q Q WangFull Text:PDF
GTID:2428330602970207Subject:Electronic and communication engineering
Abstract/Summary:PDF Full Text Request
As automotive,aerospace,weaponry,and other fields require higher and higher reliability of tube-like tubular parts,quality problems such as corrosion and scratches on the inner wall of the tubular workpiece will directly affect the performance of industrial products and life and property.Safety,therefore,regularly carry out quality inspection and evaluation of the inner wall of the cylindrical workpiece.At present,imaging detection methods based on folding mirrors and structured light are affected by the narrow and long internal characteristics of the cylindrical workpiece,and the imaging device is difficult to design and process and has low reliability.Direct imaging through a fisheye lens,due to the short imaging focal length and large field of view distortion,affects the detection effect.To solve this problem,the paper is based on a zoom lens and uses multi-focus vision imaging technology to study the reconstruction method of the inner wall shape of the cylindrical workpiece based on the multi-focus image sequence fusion,and to study the panoramic imaging detection of the inner wall of the cylindrical workpiece.First,for the problem of difficult extraction of focus region features in multi-focus image sequences,combined with the advantages of deep learning adaptive extraction of deep features,a multi-focus image sequence fusion algorithm based on Siamese Network(Siamese Network)is studied,and a detection tube based on Siamese Network is used.The focus area of the inner wall of the workpiece is determined by the score map output by the network to determine the focus area of the image sequence of the inner wall with different depths of field,and a fully focused inner wall panoramic view is merged.Because the panoramic image is to compress the 360-degree inner wall information of the cylindrical workpiece in the three-dimensional space into a two-dimensional planar ring,the vision is not intuitive.To solve this problem,according to the principle of perspective transformation,the method of expanding the panoramic image of the inner wall of the cylindrical workpiece is studied,and the mapping interpolation method of the coordinate system during the expansion process is improved to improve the quality of the expanded image.Finally,the thesis uses a zoom accuracy of 6.5mm-42.5mm,and designs an imaging experiment device for a cylindrical workpiece of 80mm-130 mm.The experimental results show that the morphology reconstruction method proposed in this paper can directly obtain the panoramic morphology of the inner wall of the cylindrical workpiece,and the imaging quality is high,and the inner wall quality can be effectively detected.
Keywords/Search Tags:panoramic vision, convolutional neural network, multi-focus image fusion, panoramic image development, cylindrical workpiece
PDF Full Text Request
Related items