In: Wagner W., Sz6kely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
577 
after a certain CPU time; when a minimum error function is 
reached; after minimum gradient is reached and learning per 
epoch is only marginal; when the error value of validation 
datasets starts to increase (cross-validation). The highest 
classification accuracy for the validation data was found when 
the networks were trained with early stopping (Mehner et al., 
2003). The stopping criterion was defined to be the maximum 
number of allowed iterations, which was defined as 300. 
3. RESULTS AND DISCUSSION 
In Figure 5, the beach features/pattems map obtained through 
the application of the developed ANN to the IKONOS image is 
given, with the discrimination of each identified class. 
Figure 5. Beach features/pattems map for the IKONOS image, 
obtained through the implemented ANN. 
The network performance was assessed by estimating the 
accuracy with which the validation data were classified. The 
ANN presented a very good performance, demonstrated by the 
results of the individual class accuracy (Table 4) and overall 
accuracy (98.6%). 
Class 
S 
SS 
BZ 
BF 
B 
Accuracy (%) 
S 
1748 
23 
0 
0 
0 
98.7 
SS 
27 
1727 
0 
0 
0 
98.5 
BZ 
0 
2 
139 
0 
0 
98.6 
BF 
0 
1 
0 
35 
2 
92.1 
B 
0 
0 
0 
2 
427 
99.5 
Table 4. Individual class accuracy 
This approach conduced to better results than the traditional 
classification methodologies. For instance, for the same dataset, 
the best result for the supervised classifications was achieved 
with the parallelepiped classifier, with a value of 97.5% for 
overall accuracy. The better result using an objected-oriented 
approach was found for the pan-sharpened true color imagery 
with an overall accuracy of 65.8%. 
4. CONCLUSIONS 
The ANN presented a very good performance, demonstrated by 
the results of the individual class accuracy and overall accuracy 
(98.6%). The ANN applied in this work have been shown to be 
useful in the recognition of beach features/patterns. Given the 
results obtained, some conclusions can be drawn: 
1. The spatial resolution of the IKONOS image suggests that it 
is adequate for the identification of the considered beach 
features/pattems, through ANN techniques. 
2. The use of ANNs for beach classification from remotely 
sensed data resulted in an increased classification accuracy 
when compared with traditional classification methods. 
3. When applied to the validation dataset, ANNs performed 
successfully, with high classification accuracies observed. This 
fact justifies the future application of this methodology for 
others locations. 
In the future, we are interested in the recent developments in 
classifier design with the introduction of the support vector 
machine classifier. 
In conclusion, the ANN’s have been shown to be useful in the 
recognition of beach features/pattems. 
REFERENCES 
Atkinson, P.M. and Tatnall, A.R.L., 1997. Neural networks in 
remote sensing. International Journal of Remote Sensing, 18, 
pp. 699-709. 
Atkinson, P.M., Cutler, M.E.J. and Lewis, H.G., 1997. Mapping 
sub-pixel proportional land cover with AVHRR imagery. 
International Journal of Remote Sensing, 18, pp. 917-935. 
Benediktsson, J.A., Swain, P.H. and Ersoy, O.K., 1990. Neural 
network approaches versus statistical methods in classification 
of multisource remote sensing data. IEEE Transactions on 
Geosciences and Remote Sensing, 28, pp. 540-551. 
Bishop, C.M., 1995. Neural Networks for Pattern Recognition, 
1st edn. Oxford: Clarendon Press. 
Chen, D. G., and Ware, D. M. ,1999. A neural network model 
for forecasting fish stock recruitment. Canadian Journal of 
Fisheries and Aquatic Sciences, 56, pp. 2385-2396. 
Dayhoff, J. E., 1990. Neural network architectures - An 
introduction. New York: Van Nostrand Reinhold. 
Dunne, R.A. and Campbell, N.A., 1994. Some practical aspects 
of pruning multi-layer perceptron models applied to remotely 
sensed data. Technical Report 94/06, Murdoch University. 
Haykin, S., 1999. Neural Networks: a comprehensive 
foundation, 2nd edition. Prentice Hall, pp. 117-173. 
Lippmann, R.P., 1987. An introduction to computing with 
neural nets. IEEE ASSP Magazine, April, pp. 4-22.