In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
702 
(e)Result of (f)Result of 
Two-dimensional Otsu One-dimensional Fisher 
(g) Result of 2-D Fisher (h) Result of 2-D Fisher 
(before improvement) (after improvement) 
Fig.9 the results of real image change detection 
The time occupied by two-dimensional change detection 
method in the image change detection process is shown in Table 
2. 
Algorit 
hm 
2D-Otsu 
2D-Fisher 
( before 
improvement 
) 
2D-Fisher 
( after 
improvement 
) 
Time of 
detection 
14. 46 
min 
75.24 S 
0. 79 S 
Table 2 the time occupied by two-dimensional change detection 
method in the image change detection process 
Experimental results show the excellent of the change detection 
method based on the improved 2D-Fisher criterion function, 
whose superior anti-noise, and speed could have been fully 
embodied, whether in the simulation experiment or on the actual 
image. And, it’s able to meet the needs of practical application 
of remote sensing image change detection in effect of real-time 
and change detection. 
5. SUMMARY 
The classical Fisher criterion function is introduced into the 
remote sensing image change detection. The approach based on 
2D-Fisher criterion function method is proposed. Remote 
sensing image change detection 2D-Fisher criterion function 
method extended the one-dimensional gray value space of the 
classical Fisher criterion function to two-dimensional space, 
such as (G-Mean), (G-Medium), etc. Among them, the choice 
of two-dimensional space is based on the specific circumstances 
of the actual images. For example, G-Mean has a good effect of 
removing Gaussian noise, while G-Medium are obvious for the 
salt and pepper noise. In the process of the solution, the 
two-dimensional threshold, we split it into two one-dimensional 
thresholds, and the detection speed is greatly improved. 
The 2D-Fisher criterion function method takes into account of 
the anti-noise and detection speed of the change detection 
process, making the method much more suitable for practical 
application needs of remote sensing image change detection. 
Experiments show that the improved 2D-Fisher criterion 
function method of remote-sensing image change detection is 
superior to the classical one-dimensional Otsu method, 
one-dimensional Fisher criterion function method in the 
anti-noise, and is far superior to the Two-dimensional Otsu in 
the speed of change detection. 
6. REFERENCE 
[1] Richard O.Duda, Peter E.Hart, David G.Stork.,2003.Pattern 
classification, translate by Li Hong-dong, Yao TianOxiang, 
Beijing: China Machine Pre^’(Chinese). 
[2] Zhao Feng, Fan jiu-lun, 2007. One Image Segmentation 
Method Combining 2D Otsu's Method and Fuzzy Entropy. 
Application Research of Computers (Chinese), 24(6), 189-191. 
[3] Tong Ying, Qiu Xiao-hui,2004. A new algorithm of image 
segmentation using two-dimensional histogram based on Fisher 
criterion function. Telecommunications for Electric Power 
System (Chinese), 25(9), 36-39, 47. 
[4] Yu Jin-hua, Wang Yuan-yuan, Shi Xin-ling, 2007. Image 
segmentation with two-dimension fuzzy cluster method based 
on spatial information. Opto-electronic Engineering (Chinese), 
34(4), 114-119 
[5] Chen Guo, 2003. The Fisher criterion function method of 
Image Thresholding. Chinese Journal of Scientific Instrument, 
24(6), 564-567,576. 
[6] Xie Mingxia, Chen Ke, Guo Jianzhong, 2008. Research of 
FCM for image segmentation based on graph theory. Journal of 
Computer Applications (Chinese), 28(11), 2912-2914. 
[7] Zhong Jia-qiang,Wang Run-sheng, 2005. Multitemporal 
Remote Sensing Image Change Detection Based on Adaptive 
Parameter Estimation, Acta Geodaetica et Cartographica Sinica 
(Chinese), 34(4), 331-336.