Our experiments show that combining wavelet transform and 
histogram-based intersection for image comparison can 
effectively remove motion blurring images, but the number of 
images picked out by this method is much greater than actual 
number. For instance, if 30 images out of 1,000 images are useful, 
then more than 100 image or even several hundreds images can 
be chosen out by using this method. Though all the useful images 
are included, this method is actually of no use (Yan Zhou, 
personnel corresponding). 
Using our algorithm described in section 3, we solved the 
problem successfully. If a pick-out-ratio is defined between the 
number of useful images out of all the picked out images and the 
number of all the picked out images itself, then the assessment of 
the effectiveness of the proposed algorithm includes the 
following two aspects: 
(1) whether all the useful images are included in the picked out 
images? 
(2) How about the pick-out-ratio? 
Our results using train images under different illumination 
conditions show that all the useful images can be chosen out, 
With a pick-out-ratio steadily around 9096. For images of a train, 
all the procedure can be finished within 6 minutes, which is 
completely satisfied by practice. 
S. CONCLUSIONS 
Starting from the primary mathematical definition of 
semivariogram, this paper fully uses the properties of 
semivariogram which describes both the randomness and 
structure of a data set to define a new parameter for image 
comparison. Compared with other image distance functions for 
image comparison, the algorithm proposed in this paper has three 
merits: high sensitivity to image structure, low computational 
complexity and no special requirements of imaging condition. 
The case study show the effectiveness of the algorithm, 
providing a new method for content-based image retrieval. 
Though more than 10 years have been past since the first 
application of semivariogram to remotely sensed image analysis, 
there is no report on its application to close-range photographic 
image processing. This paper gives out the pioneering work in 
this field. 
However, more case studies are still needed to test the 
effectiveness of the algorithm, and some computational 
strategies also should be considered to make the algorithm more 
fast. All these have been listed to the agenda of the authors 
further studies. 
REFERENCES 
l. Addink, E.A , Stein, A, “A comparison of conventional and 
geostatistical methods to replace clouded pixels in 
NOAA-AVHRR images," INT. J. of Remote Sensing, Vol 
20, No 5, pp961-977, 1999 
2. Berman A.P, Shapiro L.G., “A flexible image database 
system for content-based retrieval." Computer vision and 
image understanding. Vol. 75, Nos. 1/2, pp: 175-195, 1999 
3. Carr, JR, Miranda, F.P, “The semivariogram in 
comparison to co-occurrence matrix for classification of 
image texture,” IEEE transaction of geoscience and remote 
sensing, 36(6), pp1945-1952, 1998 
12. 
14. 
17. 
18. 
20. 
—508— 
Cha Guang-Ho, Chung Chin-wan, “An indexing and 
retrieval mechanism for complex similarity queries in image 
database ,” Journal of visual communication and image 
representation. 10, pp268-290, 1999 
Cheng K.S., Yeh H. C., Tsai C. H., “An anisotropic spatial 
modeling approach for remote sensing image rectification," 
Remote sensing of environment. 73, pp 46-54, 2000 
Chica-Olmo, M., Abarca-Hernandez, F., "Computing 
geostatistical image texture for remotely sensed data 
classification." Computers & Geosciences. 26, pp373-383, 
2000 
Chiles, J. P & Delfiner, P., Geostatistics: modeling spatial 
uncertainty. P. 650, Jonhn Wiley & Sons. Inc, New York, 
1999 
Collins, J.B, Woodcock, C.E., *Modeling the distribution of 
cover fraction of a geophysical field." In: Advances in 
remote sensing and GIS analysis, edited by Atikonson,P. M. 
and Tate, N.J. pp119-133, John Wiley & Sons, Chichester, 
1999: 
Costa, J.P., Pronzato, L., Thierry, E., “Nonlinear prediction 
by kriging with application to noise cancellation, * signal 
processing. 80 , pp553-566, 2000 
Cressie, N.A.C, Statistics for spatial data analysis. P. 900, 
Jonhn Wiley & Sons. Inc, New York, 1991 
Curren, P. J., “The semivariogram in remote sensing: an 
introduction.” Remote sensing of environment. 24: 
pp493-507, 1988 
Curren, P. J., Dungan, J.L., “Estimation of signal-to-noise: a 
new procedure applied to AVIRIS data,” IEEE transaction 
on geoscience and remote sensing. 27, pp620-628, 1989 
Decenciere,E., Fouquet, C.D., Meyer, F. “Applications of 
kriging to image sequence coding,” signal processing: 
image communication. 13: pp227-249, 1998 
Di Gesu V., Starovoitov V. “Distance-based functions for 
image comparison,” Pattern recognition letters. 20, pp 
207-214, 1999 
Green, A.A., Berman, M., and Switzer, P, et al, “A 
transformation for ordering multispectral data in terms of 
image quality with implications for noise removal.” IEEE 
transactions on geoscience and remote sensing, 26: pp65-74, 
1988 
Lew M.S., Huijsmans D.P., Denteneer D. “Content based 
image retrieval: optimal keys, texture, projections, or 
templates.” In: Image database and multimedia search, 
edited by Smeulders A.W.M, Jain R., pp39-47 World 
Scientific Publishing, Singapore, 1998 
Mortensen, R. E., Random signals and systems. P. 231, 
Jonhn Wiley & Sons. Inc, New York, 1987 
Myers,D.E., “Statistical models for multiple-scaled 
analysis.” In: Scale in remote sensing and GIS, edited by 
Goodchild, M.F. et al. pp273-293, CRC Press, Raton, 1997 
Rossi, R.E., Dungan, J.L., Beck, L.R., “Kriging in the 
shadows: geostatistical interpolation for remote sensing." 
Remote sensing of environment. 49, pp32-40, 1994 
Rui Y., Huang T. S., Chang S.F, “Image retrieval: current 
techniques, promising directions, and open issues." Journal 
of visual communication and image representation. 10, pp