ISPRS Commission II, Vol.34, Part 3A „Photogrammetric Computer Vision“, Graz, 2002 
Automatic DTM Generation from Three-Line-Scanner (TLS) Images 
Armin Gruen, Zhang Li 
Institute of Geodesy and Photogrammetry, Swiss Federal Institute of Technology Zurich 
ETH-Hoenggerberg; CH-8093 Zurich, Switzerland 
Tel.: +41-1-633 31 57, Fax: +41-1-633 11 01 
E-mail: <agruen><zhangl>@geod.baug.ethz.ch 
Commission III, WG III/2 
KEY WORDS: Three-Line-Scanner (TLS), Relaxation Matching, Geometrically Constrained Multi-Image and Multi-point 
Matching, DSM 
ABSTRACT: 
This paper presents a matching procedure for automatic DSM generation from the Three-Line-Scanner (TLS) raw images. It can 
provide dense, precise and reliable results. The proposed method combines matching procedures based on grid point matching and 
feature point matching. Modified Multiphoto Geometrically Constrained Matching (MPGC) and Geometrically Constrained Multi- 
point Matching (GCMM) are used to refine the relaxation matching results in order to achieving sub-pixel accuracy on the grid DSM. 
We match three TLS images and provide the pixel and object coordinates for grid points simultaneously. 
In order to compensate the disadvantages of terrain modeling by grid points, an additional feature-point matching procedure is 
performed. The feature points are extracted by using an interest operator such as Moravec’s. Then we activate the modified MPGC, 
using three TLS images simultaneously, and achieve potentially sub-pixel accuracy. 
The sensor model used for the geometric constraints derivation is based on the collinearity equations appended by some trajectory 
models. 
The algorithms proposed in this paper have been applied to different areas with varying textures and terrain types. The accuracy test 
is based on the comparison between well-distributed semi-automatically measured feature points and the automatic extracted DSMs, 
and on visual inspection of the results. 
1. Introduction 
With the advent of large format digital aerial cameras an 
increased need for reliable automated image analysis functions 
emerges. The three-line-scanner concept provides for triple 
overlap in strip direction for every image point and as such 
basically for fairly good reliability characteristics. In addition, 
the basic capabilities of image matching techniques have so far 
not been fully utilized yet. This contribution aims at combining 
the new three-line-scanner sensor model with some novel image 
matching approaches, as multi-image and multi-point matching. 
As to the sensor we refer to the TLS system, developed by 
Starlabo Corporation, Tokyo. Our matching goal is DSM 
extraction. 
The TLS sensor model is based on the collinearity equation and 
expresses the relationship between the pixel and object 
coordinates. This sensor model is used for the recovery of the 
exterior orientation parameters for each scan line of the TLS 
images by a photogrammetric bundle adjustment, and for the 
derivation of the geometric constraints in our modified 
Multiphoto Geometrically Constrained (MPGC) matching and 
Geometrically Constrained Multi-point Matching (GCMM). 
This paper presents a matching approach for automatic DSM 
generation from the TLS raw images. It can provide dense, 
precise and reliable results. The proposed method is a combined 
matching procedure, which is based on both grid point matching 
and features point matching. After image pyramid generation 
and the extraction of approximations by using a simple feature 
point matching on the highest level of the image pyramid, grid 
point matching based on the relaxation technique is performed 
on a TLS stereo pair which can be any combination of two of 
the three TLS images. The important aspect of this relaxation 
matching that differs from other area-based single point 
matching is its compatible coefficient function and its 
smoothness constraint satisfaction procedure. With the 
smoothness constraint, poor texture areas can be bridged 
assuming the terrain surface varies smoothly over the imaging 
area. Modified MPGC and GCMM procedures are used to 
refine the relaxation matching results in order to achieve sub- 
pixel accuracy. Both can be used to match three TLS images 
and provide the pixel and object coordinates for object points 
simultaneously. 
The algorithms proposed in this paper have been applied to 
different areas with varying textures and terrain types. The 
accuracy testing is based on the comparison of well-distributed 
semi-automatically measured feature points to the automatically 
extracted DSMs and on visual inspection of the DSMs. 
2. The TLS System 
The TLS (Three-Line-Scanner) system is a new airborne digital 
sensor, developed by Starlabo Corporation, Tokyo. It utilizes 
the three-line-scanner principle to capture digital image triplets 
in along-strip mode. The imaging system contains three parallel 
one-dimensional CCD focal plane arrays, with 10200 pixels of 
7um each (see Figure 1). The TLS system produces seamless 
high-resolution images (5 - 10 cm footprint on the ground) with 
three viewing directions (forward, nadir and backward). In order 
to get highly precise attitude data and high quality raw image 
data from an aerial platform, a high quality stabilizer is used for 
the camera and outputs attitude data at 500 Hz. A Trimble 
MS750 serves as Rover GPS and collects L1/L2 kinematic data 
at 5 Hz and another Trimble MS750 serves as Base GPS on the 
ground. 
For the TLS sensor and imaging parameters see Table 1. 
The image data collected by the TLS imaging system is only 
useful under the condition that the geometric relationship 
between pixels and their corresponding ground coordinates, i.e. 
the sensor model is known. Thus, the sensor modeling is the 
most important problem to be solved firstly. 
Unlike with frame-based photography, the three-line geometry 
is characterized by nearly parallel projection in the flight 
direction and perspective projection perpendicular to the flight 
direction. Our sensor model for the TLS images is based on the 
A - 131