Font Size: a A A

Research And Application The Improved SIFT Algorithm In The Marking System

Posted on:2016-04-20Degree:MasterType:Thesis
Country:ChinaCandidate:H H ZhangFull Text:PDF
GTID:2308330464958979Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
With the process of development and the education of the information society forward, OMR emerge as the times require marking. Such as high cost, high cost of equipment marking, the popularity is not high the chance to use OMR.Aiming at these deficiencies, in this paper, after image processing related to the technical content, it is concluded that The SIFT algorithm obtained good results in image matching. But SIFT algorithm as the image matching algorithm in the crowd, and not perfect adaptation all, according to the specific problems, sift or there is a large amount of calculation, lack of efficient, some threshold setting is not reasonable cause inaccurate results.Through the Improved SIFT algorithm a large number of literature reading, according to their specific application scenarios proposed improvement of SIFT algorithm two. Is through the reverse of feature points matching and reduce the redundancy of the wrong matching points, so as to improve the efficiency of the algorithm, and the next is to decide the threshold is no longer used in the feature vector matching process, but through a certain algorithm threshold dynamic adjustment, so as to improve the algorithm of matching accuracy.After experimental verification, the Improved SIFT algorithm in maintaining the stability of the original SIFT algorithm, at the same time, the performance and accuracy of the algorithm have obviously improved, increased from 60% to 95%. In order to view the improved algorithm in the actual implementation. In this paper, the design of the function of a relatively complete scoring system, and the improved algorithm is applied to the system of template matching link, on practical operation results are analyzed.
Keywords/Search Tags:Image matching, SIFT, Feature description, Reverse matching, Adaptive matching threshold, Marking
PDF Full Text Request
Related items