In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
AUTOMATIC REGISTRATION OF AIR-BORNE AND SPACE-BORNE IMAGES BY 
TOPOLOGY MAP-MATCHING WITH SURF PROCESSOR ALGORITHM 
Anna Brook, Eyal Ben-Dor 
Remote Sensing Laboratory, Tel-Aviv University, Israel 
anna.brook@gmail.com 
Commission VI, WG VI/4 
KEY WORDS: Automatic registration, Multi-sensors airborne and space-bome fusion, Change detection, Weight-based 
topological map-matching algorithm (tMM), Scaling and image rotation 
ABSTRACT: 
Image registration is widely used in many Remote Sensing applications. The existing automatic image registration techniques fall 
into two categories: the intensity-based and the feature-based methods, while the feature-based technique (which extracts structures 
from both images) is more suitable for multi-sensors fusion, temporal change detection and image mosaicking. Conventional image 
registration algorithms have greatly suffered from quantity and spatial distribution of extracted control points. In this study, we 
propose a novel method for automatic image registration based on topology rules (AIRTop) for temporal change detection and multi 
sensors (airborne and space-bome) fusion. In this algorithm, we first apply the SURF (Speeded Up Robust Features) method to 
extract the landmarks structures (roads and buildings) of the given images, and then they are expressed by a topology rules, which 
define the permissible spatial relationships between features. The defined rules for a weight-based topological map-matching 
algorithm (tMM) manage the relationships between features in different feature classes (roads and buildings) and present a robust 
method to find a control points in both reference and sensed images. The main focus in this study is on scale and image rotation 
invariant the quality of the scanning system. These seem to offer a good compromise between feature complexity and robustness to 
commonly occurring deformations. The skew and the anisotropic scaling are assumed to be second-order effects that are covered to 
some degree by the overall robustness of the sensor. Experimental results show that our method can provide better accuracy than the 
conventional registration process. 
1. INTRODUCTION 
Image registration is a critical preprocessing procedure in 
all remote sensing applications that utilize multiple image 
inputs, including multi-sensor image fusion, temporal change 
detection, and image mosaicking. The recent interest in 
temporal change detection and modeling transform make the 
automatic image registration to important stage of preprocessing 
the data (Moigne et al., 2002). The automatic registration of 
images has generated extensive research interests in the fields of 
computer vision, medical imaging and remote sensing. 
Comprehensive reviews include Brown (1992) and Zitova and 
Flusser (2003). 
The existing automatic image registration techniques fall 
into two categories: the intensity-based and the feature-based 
methods (Zitova and Flusser, 2003). The feature-based 
technique extracts salient structures from sensed and reference 
images by invariance and accuracy of the feature detector and 
by the overlap criterion. As the significant regions (e.g. roofs) 
and lines (e.g. roads) are expected to be stable in time at fixed 
position, the feature-based method is more suitable for multi 
sensors fusion, temporal change detection and image 
mosaicking. The method generally consists of four steps (Jensen 
et al., 2004): 1) control points (CPs) extraction, 2) 
transformation model determination, 3) image transformation 
and re-sampling, and 4) registration accuracy assessment. 
Among the four steps, the first is the most complex, and its 
success essentially determines the registration accuracy. Thus, 
the detection method should be able to detect the same features 
in all projection and different radiometrical sensitivities 
regardless of the particular image / sensor deformation. 
The search for discrete CPs can be divided into three main 
steps: 1. selection of "interesting points", 2. Description of 
nearest points or features, 3. matching between images. The 
most valuated property of CPs detection is its repeatability. The 
description of nearest points has to be distinctive but robust to 
noise, potential displacements as geometric and radiometric 
deformations. In order to succeed, the matching technique has 
to be accurate and sufficient while detection scheme has to 
simplify the above requirements. 
This paper presents a novel method for automatic image 
registration based on topology rules (AIRTop) for temporal 
change detection and multi-sensors (airborne and space-bome) 
fusion. 
2. AUTOMATIC IMAGE REGISTRATION 
The AIRTop algorithm (Figure 1) consist four stages as any 
conventional registration method. First, the significant features 
extracted by applying SURF (Speeded Up Robust Features) 
method on both sensed and reference images and converted to 
vector format. The spatial distribution and relationship of these 
features expressed by topology rules and converts them to 
potential CPs by determine transformation model between 
sensed and reference images. The defined rules for a weight- 
based topological map-matching algorithm (tMM) manage, 
transform and re-sampling features of sensed image according 
to reference. Since AIRTop has a sufficient number of CPs the 
registration accuracy can be estimated with test point error 
(TPE) technique. 
* Anna Brook, the Remote and GIS Sensing Laboratory, Tel-Aviv University. Ramat Aviv P.O. Box 39040 Tel Aviv 69978, Israel, 
Tel: 972-3-6407049 Fax: 972-3-6406243 Email: anna.brook@gmail.com. 
98