In: Wagner W„ Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010,1 APRS, Vol. XXXVIII, Part 7B 
218 
spring crops at h . Obtained results are very satisfactory, as 
confirmed by both the high kappa and OA% values reported in 
Table 2 (always higher than 0.79 and 92, respectively). More 
over, by employing the majority voting ensemble it is possible 
to further improve the performances and obtaining for both 
indexes accuracies even closer to those obtained by PCCml and 
PCCsvm with a fully-supervised training at both dates. 
Dataset I 
Dataset II 
Dataset III 
K 
kappa 
OA% 
kappa 
OA% 
kappa 
OA% 
40 
0.8925 
96.59 
0.8403 
95.09 
0.7955 
92.61 
60 
0.8903 
96.53 
0.8458 
95.21 
0.8488 
94.97 
80 
0.8822 
96.25 
0.8412 
95.11 
0.8441 
94.80 
100 
0.8878 
96.47 
0.8470 
95.31 
0.8502 
94.93 
PSCTW 
0.9076 
97.08 
0.8649 
95.82 
0.8622 
95.32 
P CCml 
0.9339 
97.88 
0.9323 
97.83 
0.9330 
97.86 
PCC 5KM 
0.9501 
98.36 
0.9359 
97.90 
0.9703 
99.02 
Table 2. kappa coefficient of accuracy and OA% obtained for 
the “bare soil to spring crops” land-cover transition. 
While addressing the “alfalfa to com” transition with the pro 
posed PSCD technique, from all the training pixels reported in 
Table 1, we considered the only 2031 available for alfalfa at t\ 
and the only 2664 spatially-disjoint available for com at h . 
Such a transition is rather difficult to characterize, as only ex 
perienced by few fields in the considered area. Indeed, accord 
ing with the results in Table 3, this is confirmed by the very low 
accuracies obtained by PCCm, despite fully-supervised training. 
Instead, in the light of the high complexity of the problem, per 
formances exhibited by the proposed method are very promis 
ing, especially for Dataset II and Dataset III. Moreover, with 
the majority voting ensemble the gap with respect to PCC SVM 
becomes very small (Dataset II) or it is even possible to outper 
form results obtained with SVMs (Dataset III). 
Dataset I 
Dataset II 
Dataset III 
K 
kappa 
OA% 
kappa 
OA% 
kappa 
OA% 
40 
0.6725 
97.36 
0.8549 
98.79 
0.8117 
98.40 
60 
0.7110 
97.79 
0.8190 
98.49 
0.8343 
98.60 
80 
0.6530 
97.25 
0.8172 
98.45 
0.8206 
98.45 
100 
0.6553 
97.32 
0.7233 
97.92 
0.8367 
98.62 
PSCD WK 
0.7524 
98.16 
0.8896 
99.06 
0.9079 
99.22 
PCC mi 
0.5163 
96.87 
0.5134 
96.86 
0.5353 
97.05 
P CCsvm 
0.9003 
99.08 
0.9244 
99.32 
0.8969 
99.04 
Table 3. kappa coefficient of accuracy and OA% obtained for 
the “alfalfa to com” land-cover transition. 
5. CONCLUSIONS 
In this paper we presented a novel partially-supervised change- 
detection (PSCD) technique capable of addressing targeted 
change-detection problems where the objective is to identify 
one (or few) targeted land-cover transitions, under the assump 
tion that ground-truth information is available for the only (few) 
class(es) of interests at the two investigated dates. 
In this context, either supervised or unsupervised standard ap 
proaches cannot be effectively employed. The proposed 
method, instead, allows exploiting the only prior knowledge 
available for the specific land-cover classes of interest at the 
two times, while providing accuracies comparable with those of 
fully-supervised methods. In particular, the PSCD technique 
relies on a partially-supervised approach and jointly exploits the 
Expectation-Maximization (EM) algorithm and an iterative 
labelling strategy based on Markov random fields (MRF) ac 
counting for spatial and temporal correlation between the two 
images. Moreover, it also allows handling images acquired by 
different sensors at the two considered times. Experimental 
results on multi-sensor datasets derived from multispectral, 
hyperspectral and SAR data confirmed the effectiveness and the 
reliability of the proposed technique 
REFERENCES 
Besag, J., 1986. On the statistical analysis of dirty pictures. 
Journal of the Royal Statistical Society, Series B, 48, pp. 259- 
302. 
Bruzzone, L., and Femandez-Prieto, D., 1999. A Technique for 
the Selection of Kernel-Function Parameters in RBF Neural 
Networks for Classification of Remote-Sensing Images. IEEE 
Trans. Geosci. Remote Sens., 37(2), pp. 1179-1184. 
Coppin, P., Jonckheere, I., Nackaerts, K., and Muys, B., 2004. 
Digital change detection methods in ecosystem monitoring: a 
review. Int. J. Remote Sens., 25(9), pp. 1565-1596. 
Cristianini N., and Shawe-Taylor, J., 2000. An Introduction to 
Support Vector Machines. Cambridge University Press. 
Dempster, A., Laird, N., and Rubin, D., 1977. Maximum Like 
lihood from Incomplete Data Via the EM Algorithm. The Royal 
Statistical Society, Series B, 39(1), pp. 1-38. 
Duda, R. O., Hart, P. E., and Stork, D. G., 2000. Pattern Classi 
fication. Wiley, New York. 
Inaba, M., Katoh, N., and Imai, H., 1994. Applications of 
weighted voronoi diagrams and randomization to variance- 
based k-clustering. Proceedings of the 10 th Annual Symposium 
on Computational Geometry, pp. 332-339. 
Jensen, J. R., 2009. Remote Sensing Of The Environment. Pear 
son, Upper Saddle River. 
Jeon, B., and Landgrebe, D. A., 1999. Partially supervised clas 
sification using weighted unsupervised clustering. IEEE Trans. 
Geosci. Remote Sens., 37(3), pp. 1073-1079. 
Lu, D., Mausel, P., Brondizio, E., and Moran, E., 2004. Change 
detection techniques. Int. J. Remote Sens., 25(12), pp. 2365- 
2407. 
Radke, R. J., Andra, S., Al-Kofahi, O., and Roysam, B., 2005. 
Image Change Detection Algorithms: A systematic Survey. 
IEEE Trans. Geosci. Remote Sens., 14(3), pp. 294-307, 2005. 
Richards, J. A., and Jia, X., 2006. Remote Sensing Digital Im 
age Analysis. An Introduction. Springer, Berlin. 
Singh, A., 1989. Digital change detection techniques using 
remotely-sensed data. Int. J. Remote Sens., 10(6), pp. 989-1003. 
Solberg, A. H. S., Taxt, T., and Jain, A. K., 1996. A Markov 
Random Field Model for Classification of Multisource Satellite 
Imagery. IEEE Trans. Geosci. Remote Sens., 34(1), pp. 100-113.