A novel method for spatial relation extraction based 
quadtree has been present. The method is applicable 
for remotely sensed image interpretation. 
The main contributions of this paper are; (1) 
representing a hierarchical structure of quadtrees, 
which is composed of quadtree segmentation quadtree 
, homogeneous region quadtree and spatial relation 
quadtree. (2) describing a quadtree structure called 
binding that contains three parts, location code , region 
number and attribute value. (3) introducing the 
notions of control quadtree, feature quadtree, feature 
node and feature line quadtree into the spatial relation 
extraction. 
In addition, A comparetively normalized way of 
repre senting image and interpreting image based on 
quadtree are described in this paper. It can exclude 
the ambiguity of spatial relation extraction and 
interpretation even when some objects spectrum are 
the of 
segmentation and interpretation errors involve edges 
very approximate. Because problems 
and inperfect interpretation etc.in the low — level 
phase , which are difficult to be detected in the high — 
562 
level phase.some interpretation errors can still be 
found from results . We have tried to solve them by 
repeating some proper steps to revise the errors under 
current best interpretation based on DGs. But the 
further improvement of this method is being done by 
authors. 
6. ACKNOWLEDGEMENT 
This paper was sponsored by the National Natural 
Science Foundation of China .This support is 
gratefully acknowledged. 
REFERENCES 
(1) Jean P. L, “Two — dimensional run — encoding 
for quadtree representation” ,CVGIP 30,1985, 56 — 
69. 
(23 AM, 
Comput. Vision Graphics Image Process. VOL. 43, 
NO. 3,1988. 
(3). V.S. Frost likelihood 
classification of systhetic aperture radar imagery", 
Jan * Discrimination vision ”, 
etc * , ^ Maximum 
Comput. Vision Graphics Image Process. VOL 32, 
NO. 3,1985.