Abstract
This paper treats the problem of detecting straight or circular pieces of road in noisy low-resolution aerial images. It first uses a local operator to detect pixels whose neighborhoods are line-like, and then applies (robust) estimation techniques to find sets of such pixels that lie on or near straight or circular loci. An (unbiased) ordinary least squares estimator cannot handle outlying data even for straight loci; on the other hand, conventional robust techniques for fitting circular arcs are severely affected by digitization effects and the fact that circular road segments are typically short and shallow. We therefore introduce an estimator that is both robust and statistically efficient. We also present a simple ad hoc technique that achieves comparable results and can handle road segments that are either straight or circular.
| Original language | English |
|---|---|
| Pages (from-to) | 1673-1686 |
| Number of pages | 14 |
| Journal | Pattern Recognition |
| Volume | 30 |
| Issue number | 10 |
| DOIs | |
| State | Published - Oct 1997 |
| Externally published | Yes |
Bibliographical note
Funding Information:The detection of roads in aerial (or satellite) images has been studied by many researchers over the past 20 years. (1-1°) Most of this work has dealt with pieces of road of arbitrary smooth shape, using nonlinear operators (of size of the order of the road width) to detect line-like or strip-like features in the image, which are possible road fragments, and then applying tracking methods to extract straight or smoothly curving sets of these fragments, which are presumed to be connected pieces of road. The shapes of the pieces of road are usually assumed to satisfy general-purpose models: straight, piecewise straight, smoothly curved \[with an upper bound on curvature (1 " tl) \]. However, highway engineering practice (11) specifies that on level ground, roads should ideally be straight, with the straight segments joined by curved pieces that, except for brief transitions, are essentially circular arcs. \[Road junctions are not treated in this paper; examples of their treatment can be found in earlier work, (12'13) which also (as do Geman and Jedynak °°)) make use of a probabilistic framework for the tracking process.\] In detecting pieces of road (on * Author to whom correspondence should be addressed. 1A condensed version of this paper appeared in Proceedings of the Thirteenth IAPR International Conference on Pattern Recognition, Vol. B, pp. 151-155. 2The support of the Advanced Research Projects Agency (ARPA Order No. A369) and the Air Force Office of Scientific Research under Grant F49620-93-1-0576 is gratefully acknowledged, as is the help of Sandy German in preparing this paper.
Keywords
- Circular arc fitting
- Line fitting
- Nonlinear regression
- Ordinary least squares
- Robust estimators