Generalization of cadastral map based on graphics matching

Generalization of cadastral map based on graphics matching

Xiaoyong CHEN, Jie DU, Takashi DOIHARA^*
Asian Institute of Technology, P.O. Box 4 Klong Luang,
Pathumthani 12120, THAILAND Email: xychen@ait.ac.th
^*Asia Air Survey Co., LTD., 8-10, Tamura-Cho,
Atsugi-Shi, Kanagawa 243, JAPAN

Abstract:
Old cadastral map records the geographic shapes of land partials and their related thematic information. It is generally drawn on an A3/A4 size paper and only presents the cadastral information of small city blocks. During the recent years, Japan government plans to make a kind of new cadastral map that is a typical formal map with the scale 1:2,500. So the editing works are needed to convert or transfer the old cadastral map to new one. The objective of this paper is to present the algorithms for automated generalization of old cadastral map based on graphics matching. Some practical examples are also given to show the efficiency of our algorithms.

Introduction
Old cadastral map records the geographic shapes of land partials and their related thematic information (such as land ID number, owner name, and prices). It is generally drawn on an A3/A4 size paper and only presents the cadastral information of small city blocks. During the recent years, Japan government plans to make a kind of new cadastral map that is a typical formal map with the scale 1:2,500. So the editing works are needed to convert or transfer the old cadastral map to new one. The purpose of this research is to develop the algorithms and a system for automated editing old cadastral map, which mainly contains feature point detection, feature point matching, conflicted line detection, automated conflicted line generalization.

Figure 1. An example of an old cadastral map

As shown in Figure 1, there are four kinds of map data for our research. These data are cadastral line map, road line map, building line map and boundary line map. Cadastral line map is or target map, it will be overlapped on base maps and adjusted by existing base lines, such as road lines, building lines and boundary lines.

There are two kinds of problems in our research. One is automated matching old local cadastral map with road base line map. Another one is automated detection of conflicted lines (such as cadastral lines with boundary lines, cadastral lines with building line, and cadastral lines with local small road lines) and make related automated adjustment.

Methodology
1 Cadastral Map Pre-matching
The method of cadastral map pre-matching with road maps is based on Affine transformation by using four given points on each map. Why we select Affine transformation as our cadastral map pre-matching method is that two maps are not exactly similar and sometimes we need matching with several city blocks together. So free matching will be very difficult and reliability will be very low. Giving two points on each map is only for scaling and rotation transforms. So that giving four points on each map is the minimum numbers for reasonable complicated transformation.

2 Cadastral Map Sequential Auto-matching
After above pre-matching, two map layers may still have small differences. We designed an algorithm called sequential auto matching based on sequential Affine transforms.

Figure 2. An example of a cadastral map sequential auto-matching

The procedure of our algorithm is as follows:

Checking four control points used by pre-matching on each map. If the coordinates still have some differences on cadastral map, we move them to the related points on road map. Then fix all these points (it means that these points will be no change in following transforms).
Automatically searching one similar feature point one each map based on the given distance parameter and point sequence. If one feature point has been found, we select three control points within four given points at beginning. The rule for selecting three control points is based on the maximum area generated by final four points.
Affine transforming the cadastral map by using the selected four points. Checking the coordinator differences between two maps. If the contribution (the coordinator differences) of this step processing is too big, the pre-result will be no modifying. Otherwise, changing the pre-result with new coordinates.
If the coordinator differences between the automatically searched feature point and its related points on road map in smaller than given limited, moving and fixing two points.
Sequentially processing the step 2-4 until the end.

The algorithm has three good points. First is simple by using Affine transform only. But the sequential simple processing will approach the complicated result. Second is easy to control because we only select one new point for processing. Third is high reliability since these three points are selected by operators and they have high accuracy and good distribution.
3 Boundary Line based Automated Adjustment
The algorithm for automated adjusting cadastral lines based on given boundary lines includes following three parts:

Searching the conflicted cadastral lines based on given boundary lines;
Moving the conflicted cadastral lines to given boundary lines;
Related cadastral line adjustment.

1). Searching the conflicted cadastral lines
We use three parameters (distance between two lines, angle between two lines, and distance between end points) to search the conflicted cadastral lines. That means we only want to modify the cadastral line, which is almost parallel to one of boundary line and has limited distance between two lines and one of end point. The distance between two lines can be calculated by averaging the two distances from end points to lines.

2). Moving the conflicted cadastral lines
If one cadastral line has been checked as a conflicted line by above three parameters, we move it to the given boundary line based on the end point which has minimum distance between two lines. We sequentially move all conflicting cadastral lines until the end.

3). Related cadastral line adjustment
The processing includes following four parts:

Searching the related cadastral lines based on the end pints;
Moving each one of end pint on the conflicted cadastral lines to new points;
Adjusting each another end point on the modified lines checked by the direction changing between new and old lines.
Sequentially doing above processing until the end.

4 Building Line based Automated Adjustment
The algorithm for automated adjusting cadastral lines based on given building lines includes following three parts:

Searching the conflicted cadastral lines by cross-point calculation. Calculating cross-point number and moving distance and direction for each conflicted cadastral lines;
Moving the conflicted cadastral lines based on calculated cross-point number and moving distance;
Related cadastral line adjustment.

1). Searching the conflicted cadastral lines
We use the cross-point between cadastral lines and building lines to search the conflicted cadastral lines. For each cadastral line, we calculate the cross point between cadastral lines and building lines. The cross-point must be a real cross-point within the given line segments and not on their extension lines. Then we count the cross-point numbers and calculate the maximum distance (and its direction) of each minimum distance between cross-points and end points of building lines.

2). Moving the conflicted cadastral lines
If the cross-point number is big or equal to two and the moving distance is smaller than the given limited distance, we move this cadastral line by calculated moving direction and distance plus alpha. If the cross-point number is equal to one (it means that the end point of cadastral line is located inside a building nd the moving distance is larger than the given limited distance (it means that the crossed building has some problem, such as merged or generalized by several small buildings), we do not move this cadastral line. These problems will be checked and modified by operator editing.

3). Related cadastral line adjustment
The processing is similar with boundary line processing and includes following four parts:

Searching the related cadastral lines based on the end pints;
Moving each one of end pint on the conflicted cadastral lines to new points;
Adjusting each another end point on the modified lines checked by the direction changing between new and old lines.
Sequentially doing above processing until the end.

2.5 Road Line based Automated Adjustment
The algorithm for automated adjusting cadastral lines based on given road line is similar with boundary line processing and includes following three parts:

Searching the conflicted cadastral lines based on given local road lines;
Moving the conflicted cadastral lines to given local road lines;
Related cadastral line adjustment.

2.6 Editing Methods
It is very difficult to get the 100% correct result for automated processing. So we need develop some editing tools for modifying the mistakes caused by automated processing. Here we developed a tool for moving cadastral points and related lines based on mouse click. Since other map layers, such as boundary lines, building lines and road lines, are benchmark lines, we do not need to modify them at all.

Result
According to above presented algorithms, we have developed a system on Windows 98/NT for automated old cadastral map editing. The system main window is shown in Figure 3. The whole processing only need inputting 4 points, then can be automatically working as the following sequence:

Cadastral map inputting;
Inputting 4 control points and make cadastral map pre-matching;
Feature point detection and automated cadastral map matching;
Boundary line based conflicted line detection and automated adjustment;
Building line based conflicted line detection and automated adjustment;
Road line based conflicted line detection and automated adjustment;
Editing and output.

Figure 3(a). The original cadastral map overlapped with other base maps.

Figure 3(b). The final editing result.

Reference

Burrough, P.A., 1986: Principles of Geographical Information Systems for Land Resources Assessment, Oxford Science Publications.

Chen, X. (1991). Image Analysis and Mathematical Morphology --- the Development of application Models for Computer Mapping. Surveying & Mapping Press, Beijing.
Chen, X. and K. Ikeda (1994). AUTO-2D --- A Super Map Digitizing System Assisted by Mathematical Morphology. Asia GIS/LIS AM/FM and Spatial Analysis Conference, Hong Kong.

Chen, X. and S. Murai, (1992). Morphology based an automated approach for editing raster digitized contour maps. The 13th Asian conf. on Remote Sensing, Uramlbatall.

Dori, D. and et.al. (1993). Sparse-Pixel recognition of primitives in engineering drawings, Machine Vision and Applications, 6: 69-82.

Douglas, D. H. and T. K. Peucker (1975). Algorithms for the reduction of the number of points required to represent a digitized line or its caricature. The Canadian Cartographer, 10(2): 112-122.

Lawrence, R. L. and et.al. (1996). An Automated Method for Digitizing Color Thematic Maps, PE&RS, 62(11): 1245-1248.

Lee, L. and T. Su (1996). Vision-Based Image Processing of Digitized Cadastral Maps, PE&RS, 62(5): 533-538.

Li, D. and X. Chen (1990), Automatic Generation of DEM from CCD-Scanned Topographic Contour Maps. Festschrift Gottfried Konecny zum 60 Gebrutstag, Hnnover University, Heft 13.

Li, D. and X. Chen (1991). Automatically Generating Triangulated Irregular Digital Terrain Model Networks by Mathematical Morphology, ISPRS J. Potogramm. Remote Sensing, 46, pp. 283-295.

Longley, P.A. and et.al., 1999: Geographical Information Systems (Vol.I),John Wiley & Sons, Inc..

Morain, S.and S. L. Baaros, 1996: Raster Imagery in Geographical Information Systems, ONWORD Press.

Robinson, A. H. and et.al., 1984: Elements of Cartography, John Wiley & Sons, Inc..

Serra, J. (1982). Image Analysis and Mathematical Morphology, Academic Press, London.