International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
KERNEL-COMPOSITION FOR CHANGE DETECTION IN MEDIUM RESOLUTION
REMOTE SENSING DATA
Andreas Ch. Braun, Uwe Weidner, Stefan Hinz
Institute of Photogrammetry and Remote Sensing
KIT - Karlsruhe Institute of Technology
Englerstr. 7, 76131 Karlsruhe
andreas.ch.braun @kit.edu, uwe.weidner @kit.edu, stefan.hinz@kit.edu
http://www.ipf kit.edu
Commission VII/5
KEY WORDS: Change Detection, Classification, Landuse, multispectral, multitemporal, Landsat
ABSTRACT:
A framework for multitemporal change detection based on kernel-composition is applied to a multispectral-multitemporal classification
scenario, evaluated and compared to traditional change detection approaches. The framework makes use of the fact that images of
different points in time can be used as input data sources for kernel-composition - a data fusion approach typically used with kernel
based classifiers like support vector machines (SVM). The framework is used to analyze the growth of a limestone pit in the Upper
Rhine Graben (West Germany). Results indicate that the highest accuracy rates are produced by the kernel based framework. The
approach produces the least number of false positives and gives the most convincing overall impression.
1 INTRODUCTION
Although the availability of modern remote sensing datasets in-
creases — e.g. hyperspectral, high resolution optical satellite im-
agery, interferometric synthetic aperture radar — many change de-
tection applications continue to require traditional datasets. The
fact that e.g. Landsat data are available since 1972 makes them
a valuable source of information for four entire decades. They
can be seen as a way to recover information on past environ-
mental conditions which are not observable any more by field
campaigns. However, traditional methods like post-classification
change detection based on overlaying classification maps raise
accuracy issues (Serra et al., 2003). Therefore, more sophisti-
cated change detection methods have been proposed in literature.
Some approaches model the probability of transition from one
class to another. Another approach is change detection based
on kernel-composition (Camps- Valls et al., 2006b). These issues
are exemplified on a change detection application from Upper
Rhine Graben. Changes in the landuse are outlined with spe-
cial focus on the construction of a limestone pit which contin-
uously grows replacing near-natural ecosystems. Kernel based
classifiers — like the well know support vector machine (SVM)
(Boser et al., 1992), (Cortes and Vapnik, 1995) — work on kernel
matrices. These kernel matrices represent the similarity between
data points in high dimensional feature spaces (reproducing ker-
nel Hilbert spaces, RKHS). The SVM chooses the most suitable
points by optimizing a target function on the kernel choosing only
a few training data points as SVs. These points are used to de-
fine a separating hyperplane which is usually non-linear in the
input space. The conceptual advantage of kernel-composition
is, that different kernel functions can by combined e.g. by ad-
dition, thus performing data fusion in the RKHS during classi-
fication. This circumstance is usually employed for data fusion
(Tuia and Camps-Valls, 2009). However, it can also be employed
for change detection. In (Camps- Valls et al., 2008), (Camps- Valls
et al., 2006a) well know change detection techniques — like im-
age differencing — are adapted for kernel-composition based ap-
proaches. An entire framework for change detection and mul-
titemporal classification is presented. This framework is evalu-
Figure 1: Limestone pit near Heidelberg
ated and compared against traditional approaches herein. We ex-
emplify this framework analyzing the growth of a limestone pit
near the city of Heidelberg between 2001 and 2005. The kernel
based approaches are compared against traditional approaches
like post-classification change detection and an approach based
on stacking features before classification. Although traditional
approaches provide valuable information on the overall growth
of the limestone pit, kernel based approaches provide more ro-
bust results which are visually more convincing and quantita-
tively more accurate. The remainder of the paper is organized
as follows. Sec.2 provides the mathematical foundations of the
methods used. Sec.3 gives an overview of relevant literature in
change detection and kernel-composition methods. The results
of the comparison are presented in Sec.4 and discussed in Sec.5.
Finally, Sec.6 concludes the contribution.
