EMPIRICAL EVALUATION OF AUTOMATICALLY EXTRACTED ROAD AXES

Internal self-diagnosis and external evaluation of the obt ained results are of major importance for the relevance of any automatic system for practical appl ications. Obviously, this statement is also true for automatic image analysis in photogrammetry an d remote sensing. However, so far only relatively little work has been carried out in this area . This is mostly due to the moderate results achieved. Only recently automatic systems reached a state in which a systematic evaluation of the results seems to be meaningful. This paper deals with the external evaluation of automatic r oad extraction algorithms by comparison to manually plotted linear road axes used as referen ce data. The comparison is performed in two steps: (1) Matching of the extracted primitives to the ref rence network; (2) Calculation of quality measures. Each step depends on the other: the less to l rant is matching, the less exhaustive the extraction is considered to be, but the more accurate it l ooks. Therefore, matching is an important part of the evaluation process. The quality measures pr oposed for the automatically extracted road data comprise completeness, correctness, quality, re dundancy, planimetric RMS differences, and gap statistics. They aim at evaluating exhaustivity as w ell as assessing geometrical accuracy. The evaluation methodology is presented and discussed in de tail. Results of a comparative evaluation of three different automatic road extraction approac hes are presented. They show the overall status of the road extractors, as well as the individual stre ngths and weaknesses of each individual approach. Thus, the applicability of the evaluation method is proven.

[1]  George Vosselman,et al.  Erkennung topographischer Passpunkte durch relationale Zuordnung , 1992 .

[2]  J. Chris McGlone,et al.  Projective and object space geometry for monocular building extraction , 1994, 1994 Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.

[3]  Olivier Jamet,et al.  Road network interpretation: a topological hypothesis-driven system , 1994, Other Conferences.

[4]  S. Airault,et al.  Détection et restitution automatique du réseau routier sur des images aériennes , 1995 .

[5]  William J. Christmas,et al.  Structural Matching in Computer Vision Using Probabilistic Relaxation , 1995, IEEE Trans. Pattern Anal. Mach. Intell..

[6]  Olivier Jamet,et al.  Distortion model in road networks from topographic maps: identification and assessment , 1995, Defense, Security, and Sensing.

[7]  Frederic Fol Leymarie,et al.  From Manual to Automatic Stereoplotting: Evaluation of Different Road Network Capture Processes , 1996 .

[8]  Helmut Mayer,et al.  A New Approach For Line Extraction and its Integration in a Multi-Scale, Multi-Abstractio n-Level Road Extraction System , 1996 .

[9]  Volker Walter,et al.  Zuordnung von raumbezogenen Daten: am Beispiel der Datenmodelle ATKIS und GDF , 1997 .

[10]  C. Steger,et al.  Multi-Resolution, Semantic Objects, and Context for Road Extraction , 1997 .

[11]  Carsten Steger,et al.  Removing the bias from line detection , 1997, Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[12]  Pascal Fua,et al.  Model-Based Optimization: An Approach to Fast, Accurate, and Consistent Site Modeling from Imagery , 1997 .

[13]  Christian Heipke,et al.  Context-Supported Road Extraction , 1997 .

[14]  R. Ruskoné.  Toward an automatic extraction of the road network by local interpretation of the scene , 2000 .