1109
Multi-Image Matching for DTM Generation from SPOT-5 HRS/HRG and IRS-P5 Imagery
- For the Project of West China Topographic Mapping at 1:50,000 Scale
ZHANG Li 1 , ZHANG JiXian 1 , Wang ShaoCheng 2
{ zhangl, zhangjx }@casm.ac.cn
'Chinese Academy of Surveying and Mapping, Haidian, Beijing, 100039
2 Beijing East Dawn Information Technology Inc., Haidian, Beijing, 100080
Theme Session ThS-3
KEYWORDS: The Project of Western China Topographic Mapping (WChTM) at 1:50,000 Scale, High-Resolution Satellite Imagery
(HRSI), Multiply Primitive Multi-Image (MPM) Matching, Geometrical Constrained Cross-Correlation (GC 3 ), Digital Terrain Model
(DTM)/Digital Surface Model (DSM)
ABSTRACT:
High-resolution satellite images (HRSI) at sub-5m footprint such as IKONOS, IRS-P5 (CartoSat-1) and SPOT-5 HRG/HRS images
are becoming increasingly available to the earth observation community and their respective clients. HRSI is one of the main data
sources for the Project of Western China Topographic Mapping (WChTM) which has been approved by the State Council of China in
year 2006. New advanced technologies in field of photogrammetry and remote sensing, such as the block-adjustment with sparse
GCPs, automatic DSM/DTM generation and 3D mapping with HRSI are very necessary to the project. In this paper, we present a
matching approach for automatic DSM/DTM generation from HRSI. It can provide dense, precise and reliable results. The method
matches multiple images simultaneously and it uses a coarse-to-fine hierarchical solution with an effective combination of several
image matching algorithms and automatic quality control. The DSMs/DTMs are generated by combination of matching results of
feature points, grid points and edges. The new characteristics provided by the high-resolution imaging systems, i.e. the multiple-view
terrain coverage and the high quality image data, are also efficiently utilized in this approach.
The proposed approach has been applied to SPOT-5 HRS/HRG images over a testfield in Zone of headstream of Three rivers, eastern
Tibet Plateau, China with variable terrain géomorphologie type. The accuracy study was based on the comparison between as many
as 160 accurate GPS check points, more than 1400 manually measured check points and the automatically extracted DTMs. The
RMS errors for the whole area are 2-7 m, while for flat/hilly areas the accuracy is about 2-3 m or even better. The proposed approach
has also been applied to 23 IRS-P5 stereo pairs over Beijing city, the resulted 12.5 m DTM reproduced quite well not only the
general features of the terrain relief but also small geomorphological and other features visible in the IRS-P5 images. From these
experiments, it’s shown that with the proposed automatic DTM generation approach, by using SPOT-5 HRS/HRG and IRS-P5
imagery, satisfactory 1:50000 DTMs can be completed with a better accuracy than those requirements from Chinese Surveying and
Mapping regulations.
1. Introduction
In recent years, space-borne CCD linear array sensors have been
widely used to acquire panchromatic and multi-spectral imagery
in push-broom mode for photogrammetric and remote sensing
applications. HRSI at sub-5m footprint such as IRS-P5, SPOT-5,
IKONOS, and QuickBird/WorldView-I provide not only for
high-resolution and multi-spectral data, but also for the
capability of stereo mapping.
SPOT-5 HRS/HRG, IRS-P5 and other HRSI images are main
data sources for the project of Western China Topographic
Mapping (WChTP) at 1:50,000 scale. At the current time, in
western part of China up to 2.02 million km 2 are not mapped at
1:50,000 scale, it includes Sorthem-XinJiang desert area,
Qing-Tibet Plateau area and Heng-duan mountain ranges. This
unmapped area covers about 20% of all areas of China; contains
total number of about 5,032 sheets of 1:50,000 scale topographic
maps. This situation greatly hiders the socio-economic
development of this region, also posed potential threat to
national security. The project has been approved by the State
Council of China in year 2006, and is dedicated to complete
1:50,000 scale topographic map and construct the national
geo-spatial database for the region within next 5 years through
year 2006 to 2010. The unmapped area related to the project
covers most unmans area of China, the very harsh natural
conditions (average terrain elevation is more than 4000m) and
difficult transportation conditions provide a very difficult
situation for field works. In many areas, it’s even impossible for
accessing or collecting enough GCPs, therefore, the project of
WChTP must across traditional surveying mode, utilize
innovative aerial and satellite remote sensing techniques, such as
DGPS/IMU assisted aerial photogrammetry, high resolution
satellite imagery mapping technique and radar image mapping
technique to achieve surveying for unmapping area in Western
China.
The project of WChTP is the importance and complicated
project which first uses HRSI cosmically for topographic map
generation in China. Automatic DSM/DTM generation is one of
the key techniques for 3D mapping from HRSI images.
Although automatic DTM/DSM generation through image
matching has gained much attention in the past years, a wide
variety of approaches have been developed, and automatic DTM
generation packages are in the meanwhile commercially
available on several digital photogrammetric workstations, the
accuracy performance and the problems encountered are very
similar in the major systems and the performance of commercial
image matchers does by far not live up to the standards set by
manual measurements (Gruen et al., 2000). The main problems
in DTM/DSM generation are encountered with:
(1) Little or no texture
(2) Distinct object discontinuities
(3) Local object patch is no planar face
(4) Repetitive objects
(5) Occlusions
(6) Moving objects, incl. shadows
(7) Radiometric artifacts like specular reflections and others
The key to successful and reliable matching is the matching of a
dense pattern of features with an appropriate matching strategy,
making use of all available and explicit knowledge, concerning
sensor model, network structure and image content such as the
epipolar geometry constraints and a piecewise smooth surface
model. For an appropriate matching strategy, we have to
