346 
cct 
Fig. 5 Spectral signature for the designated classes 
Table 1 Estimated mixing ratios of road, paddy field 
and water body classes for the designated pixels 
No. Mixing ratios Remarks 
Road Paddy Water 
1 0.198 0.612 0.190 Paddy field 
2 0.193 0.674 0.133 apart from road 
3 0.145 0.622 0.243 
4 0.287 0.612 0.101 Paddy field 
5 0.034 0.582 0.384 beside road 
6 0.078 0.552 0.370 
7 0.510 0.399 0.001 Road 
8 0.484 0.395 0.121 
9 0.637 0.231 0.132 
a 0.333 0.536 0.131 Disconnected 
b 0.343 0.517 0.140 segments of road 
c 0.443 0.377 0.180 
6. Conclusion 
This study is at the first stage of the whole 
study on the Image Classification Artificial 
Intel1 igence(ICAI) system so that further study is 
required for proportion estimation, on accuracy 
assessment for proportion estimation, in the case 
that not only mean but also variance of the 
spectral response are taken into account, and for 
the system, on the integration of multi-information 
etc. 
Reference 1 2 3 
(1) Inamura, 1987, Analysis of Remotely Sensed Image 
Data by Means of Category Decomposition, Journal of 
Electronics and Communication Society of Japan, 
Vol. J70—C, No. 2, pp. 241-250. 
(2) Rikimaru, Oshima and Kamijo, 1988, Development 
of simple method for extract pixel inside spectral 
information, Journal of Photogrammetry and Remote 
Sensing Society, Vol. 27, No. 6, pp. 23-34. 
(3) HallF.G., 1982, Satellite Remote Sensing: An 
integral tool in acquiring global crop production 
information. Proceedings of the 1982 Machine 
Processing of Remotely Sensed Data Symposium, pp.10- 
22. 
Fig. 6 Designated pixels for investigation of 
proportion 
(4) Kettig, R. L. and D. A. Landgrebe, 1975, 
Classification of multispectral image data by 
extraction and classification of homogeneous 
objects, Proceedings of the 1975 Machine Processing 
of Remotely Sensed Data Symposium, pp. 2A-1 - 2A-11. 
(5) Towmey, S., 1977, Introduction to the Mathematics 
of Inversion in Remote Sensing and Indirect 
Measurements, Elsevire Scientific Pub. Co. 
(6) Ozaki and Taniguchi, 1988, Image Processing, 
Kyouritsu Pub., Co., Ltd. 
(7) Terayama, Y., K.Teramoto and K. Arai, 1990(in 
preparing), A method for estimation of mixing ratio 
of mixed pixel(MIXED by means of Inversion Problem 
Solving method. Proceedings of the Fall Symposium of 
the Japan Society for Photogrammetry and Remote 
Sensing, E-2.