Using matching algorithms for improving locations in cadastral maps
Safra, Eliyahu and Yerach Doytsher
When integrating two different cadastral maps, points represent the same real-world location should be identified and matched. The commonly used Nearest-Neighbor (NN for short) join may be applied in order to match between objects from the two sources. Yet, this method has several evident drawbacks. First, the NN join is not symmetric, i.e. it depends on the matching direction. Second, in order to make the integration more accurate, one needs way to evaluate the quality of each match. Integration of regular vector datasets was studied in the past. This research investigates the specific features of integration of cadastral maps. When integrating cadastral maps, the main issue after identifying the matching points, is the points locations accuracy. The matching process can be used to improve the locations accuracy by interpolating the locations from the two sources. However, typically part of the points from each source does not appear in the other source, so their location cannot be easily corrected. Moreover, each pair of matched points is in some confidence, and it is not clear which matches should be included in the result. By defining pairs of matched points and points appearing in one source only, improving the locations of not matched points by finding a systematic distortion between the sources in some area is becoming fissile. Correcting the locations of not matched points is based on linear and/or non-linear rubber-sheeting transformations. Samples containing result of the integration of two large cadastre maps are presented and discussed. The results show that, without correcting the not matched points, the topology relations between neighboring nodes can be damaged. Furthermore, by applying new matching methods instead of the commonly used NN (Nearest-Neighbor) method the quality of the resulting map can be significantly improved.
Event: XXIII International FIG Congress : Shaping the change
Only personal, non-commercial use of this document is allowed.