| Summary: | Land Cover object is one of the objects that are often mapped on earth�s surface.
Mapping in detailed scale typically uses visual interpretation method, which is time
consuming and requires consistency of the interpreter. Therefore,other methods are
needed that can replace visual interpretation method, and one of them is objectbased
classification. This study aims to assess : 1) land cover mapping on detailed
scale by using visual interpretation and object-based classification method, and 2)
to analyze the capablity of the object-based classification method in land cover
mapping compared with visual interpretation method.
This study uses GeoEye-1 multispectral pan-sharperned imagery, with spatial
resolution of 0.5 meters. Land cover classification scheme used is the spatial
dimension of land cover classification. As for the method of classification using
visual interpretation method which its result uses as reference data, and objectbased
classification using multiresolution algorithm for segmentation process, and
nearest neighbor algoritm for classification process. The result will be compared
based on per land cover classes land spatial patterns to understand the capability
of object-based classification.
The analysis showed that the object-based classification only able to map the
land cover objects with same class from visual interpretation�s result by 46.37%.
while the spatial pattern tend to produce different result between two methods. This
result indicate that the object-based classification has not been able to match the
ability of the visual interpretation method in land cover mapping.
|
|---|