Font Size: a A A

The Image Processing And 3D Reconstruction Based On Shape From Focus

Posted on:2022-07-10Degree:MasterType:Thesis
Country:ChinaCandidate:X LingFull Text:PDF
GTID:2518306734479444Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
The role of digital technology in the protection and restoration of cultural relics has become increasingly prominent,and has attracted more and more attention from cultural relics protection workers and researchers.However,for cultural relics with small undulations on the surface such as murals,scanning technology can only obtain two-dimensional planar images.Therefore,it is difficult to obtain the three-dimensional feature information of such planar cultural relics with three-dimensional features.In view of this,this paper proposes a three-dimensional reconstruction method based on shape from focus(SFF)that takes macroscopic objects as the research target.This method provides a new idea for the study of obtaining the three-dimensional feature information of planar cultural relics.SFF is a passive vision 3D reconstruction technology favored by researchers and scholars.This technology only requires a camera with focus control function,which reduces the cost of image acquisition and avoids the stereo matching problem in binocular vision and multi-eye vision systems..The fruitful research results of this technology in the field of microscopy show that this technology is a widely studied method,but the application research of this technology in the macro field is minimal.The large number of cultural relics in my country and the harsh preservation environment make the research of cultural relics have the following characteristics:the wide variety of cultural relics requires versatile technology,the storage of cultural relics in warehouses or potholes should not use huge equipment,and the structure of cultural relics is not suitable for long periods of time.Exposed.The SFF technology has the advantages of simple principle,high accuracy,good real-time performance and low requirements for experimental scenes,which is in line with the demand for real-time high-precision acquisition of three-dimensional information of cultural relics under non-ideal scenes.Therefore,this paper takes macroscopic objects as the research target.Explore the three-dimensional reconstruction method based on SFF technology.In order to obtain high-precision three-dimensional reconstruction results,most of the relevant research of SFF focuses on improving the accuracy of depth estimation.The quality of the acquired image and the focus evaluation function for detecting the focus level have the most critical impact on the depth estimation.However,there are few literatures on the depth estimation.Special research is carried out on preprocessing work to improve image quality.This paper researches and improves the method of reducing the influence of illumination and noise in image preprocessing,and combines the experimental results to establish the preprocessing process of adaptive Gamma correction—improved median bilateral filtering—custom cropping algorithm—geometric correction—field of view adjustment.The improvement of the focus evaluation function is another research focus of this article.After research and testing,it is found that the spatial frequency function has better focus level detection capabilities than the commonly used SML,Tenegrad variance,gray variance,DCT and Sum of Wavelet Coefficient.This paper proposes a focus evaluation method based on weighted spatial frequency,which further improves the accuracy of detecting the focus level.The improved median filtering algorithm,the established image preprocessing process and the weighted spatial frequency function proposed in this paper can effectively improve the accuracy of the 3D reconstruction of macroscopic objects It provides necessary technical support for the subsequent use of SFF technology to restore the three-dimensional shape of cultural relics.
Keywords/Search Tags:Three-dimensional reconstruction, SFF technology, focus evaluation function, Image preprocessing in SFF technology
PDF Full Text Request
Related items