RURAL LAND-USE AUTOMATIC INTERPRETATION RESEARCH BASED ON
HIGH-RESOLUTION RS IMAGES
Y F Liu
yfliu610@163.com
Rural Land-use Automatic
Interpretation Research Based on High-resolution RS Images
-------Taking
Land-use classification of Yicheng City P.R. China as a case study
Y.F. Liu FanXia
(
The investigation
of land use condition which is the basic application of RS technology having
been researched for many years is also one of essential investigations about
the economic and natural situation of a country. To protect and utilize the
land resource of the nation, it needs to know the present land use condition authentically,
exactly and timely. Due to the existing limitations on remote sensing image
resolution and technology, the land use classification always only includes six
types, which is cultivated land, grassland, forest, residential, water,
unutilized land. In view of Nation land classification standard 2001, that is
usually only specific to a secondary category, far from being able to meet the requirement
of nation land use survey, which investigation is processing in more details.
Meanwhile, the investigation of land use in rural area attaches more important
on the recognition of cultivated land and some special residential types, such
as bleachery land for grain, diversity types of garden land and some cultivate
land and so on. Therefore, it is needed to put forward rural land use
classification standards which suitable for rural especially.
In this study,
some taste has been done to complete such task. Based on the characteristics of
rural land use, a classification process is promoted which can meet the requirements
of rural areas land use survey and be suitable for large-scale promotion at the
same time. The classification system based on "National Land
Classification," which mainly uses high-resolution remote sensing data
(SPOT) concomitant with some compliancy supporting data such as landscape
pattern indices(can treat surface features and its distribution as integration
and synthesis which would help much to the classification). Moreover, the land
cover classification may adopt the object-oriented image classification method
which can be used to reduce the amount of calculation and increase the classification
speed. The image segmentation method may be applying artificial ant colony
algorithm or optimization. Then, one or two landscape pattern indices, which
represent diversities of the differences surface features most, would be chosen
as complementary classification indices adding remote sensing image
classification. Lastly, based on above theoretical research, a sample district
of Yicheng city would be chosen as the demonstration site for the case study.
By using this classification
process, it can provide land use map automatically with more details, which
meets the requirements of the National Land Survey directly and can certain aid
nation economic and developing policy making much.
Key words: RS, Landscape pattern indices, ant
colony algorithm, object-oriented image classification