Abstract | ||
---|---|---|
It is a conventional belief that line-based approaches perform better than point-based ones for homography estimation, as the line-fitting is generally more noise resistant than point detection. In this note, we show that blithely using line-based estimation is a risky business. More specifically, we show that when the image line(s) is (are) passing through or close to the origin, the line-based homography estimation could become wildly unstable whereas the point-based estimation performs normally. To tackle this problem, a new normalized method specially designed for line-based homography estimation is proposed and validated by extensive experiments. |
Year | DOI | Venue |
---|---|---|
2008 | 10.1016/j.patrec.2008.01.031 | Pattern Recognition Letters |
Keywords | Field | DocType |
data normalization,conventional belief,line-based homography estimation,point-based homography estimation,homography estimation,point-based estimation,visual metrology,noise resistant,image line,line-based approach,extensive experiment,line-based estimation,new normalized method | Point estimation,Computer vision,Normalization (statistics),Pattern recognition,Image processing,Homography,Artificial intelligence,Mathematics,Database normalization,Homography (computer vision) | Journal |
Volume | Issue | ISSN |
29 | 9 | Pattern Recognition Letters |
Citations | PageRank | References |
14 | 0.80 | 9 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Hui Zeng | 1 | 14 | 0.80 |
Xiaoming Deng | 2 | 68 | 7.59 |
Zhanyi Hu | 3 | 830 | 62.32 |