netric 
(27) 
ırface 
Assuit 
y and 
bsigin 
sevier 
, New 
ective 
n pp. 
Sigma 
ydasie 
; Pub. 
inates 
g and 
8). 
of the 
) 149- 
  
THE ACQUISITION OF MAP INFORMATION FROM SCANNING MAP IMAGES FOR GIS 
Dr. Xiangyang Hao 
Zhengzhou Institute of Surveying and Mapping 
66 Middle Longhai Road 
PRC—450052 Zhengzhou, Henan 
Peoples Republic of China 
KEY WORDS: GIS, Recognition, Extraction,Algorithms, Graphics, Scanner. 
ABSTRACT 
The acquisition of geographic information is an indespensable process in the creation of GIS. The existed 
topographical maps are the most important source of geographic information. The automatic recognition 
and extration of map information from scanning map images are the key steps and have the great signifi- 
cance in the creation and application of GIS. In this paper, the methodology and algorithms of recognizing 
and extracting hatched polygons which stand for residential areas on maps are put forword and discussed. 
Run length smearing method in four directions is described and realized for the segmentation of hatched 
polygons. Other processes such as shrinking, expansion, thinning, open graphics deletion and distinguish- 
ing the hatched polygons from others also be discussed in detail. Experiments show that the correct recog- 
nition ratio reaches about 96%. 
1. INTRODUCTION 
It is indespensable to acquire geographic informa- 
tion in the creation of GIS. The existed maps are 
the most important one among various geographic 
information sources. Map digitizing is necessary in 
the process of inputting all sorts of map informa- 
tion into computers. The traditional way of map 
digitizing is sampling the feature points of the map 
symbols one by one on the board of digitizer man- 
ull. In order to keep the accuracy of digitized 
points, it is required to aim at each point as strictly 
as possible with the mouse. Because of the enor- 
mous amount of information on topographic maps, 
it is strenous and inefficient with this way of map 
digitizing. The appearance of scanner offers a nov- 
el way for the acquisition of map information. The 
digital map images, which can be obtained by scan- 
ning maps, make it possible to acquire map infor- 
mation intelligently or automatically. 
The strategy of automatic acquisition of map infor- 
mation is as follows:a)Obtaining the binary map 
images by directly scanning maps or transforming 
the gray map images. b)Recognizing and extract- 
ing map symbols one sort by one sort. c)Deleting 
the recognized map symbols from the copy of origi- 
nal binary map images for the recognition and ex- 
traction of the succeeding sorts of map symbols. 
d) Overlapped displaying the recognition results 
337 
with the original map images and interactive edit- 
ting. Because of the limit of pages, only the proce- 
dures and algorithms of automatic recognition and 
extraction of the residential areas are discussed in 
this paper. Residential areas with heavier popula- 
tions are usually shown as hatched polygons on the 
topographical maps. To automatically extract this 
kind of graphics directly form the complicated to- 
pographical maps, sophisticated image processing 
technology must be applied. The algorithms and 
procedures are described in the following sections. 
2. RUN LENGTH SMEARING ALGORITHMS 
One of the important steps of recognizing residen- 
tial sections is to segment the hatched polygons in- 
to blocks and then distinguish them from other 
sorts of map graphics. The run —length smearing 
(RLS) method is an effective way for this purpose. 
There are only two kinds of pixels, black pixels 
and white pixels, on binary map imges. By assum- 
ing a white pixel represented by 0 and black pixel 
by 1, we define the run length smearing tranforms 
in horizontal, vertical and two diagonal directions 
as follows. 
Defination 1 Suppose A (;,j) and B(i;,j) are two 
black pxels in row j, if |i,—i,|<n, then turn an 
arbitrary pixel C(i,j) between A and B (i.e. i,<i 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996