Study on Forest Classification Based on Object Oriented Techniques

Since vegetation is an important indicator of global climate change,then the way to extract vegetation changing data should be put as the top priority.Especially,the extraction of sub-category information of forest vegetation has always been a difficult point in remote sensing image classification.And it is more difficult to extract sub-category information of the forest vegetation type only by taking advantage of the spectral information.As a widely-used method,object-oriented classification has been rapidly developed from the beginning of this century.Object-oriented classification method is mainly used in high-resolution remote sensing imagines,and it is applicable to medium resolution remote sensing images.This paper took Mentougou District,Beijing,which is mainly covered with forest vegetation,as the object of this research,and took HJ-1 image as the main data source then different buildings can be extracted by using the object-oriented classification method.By the reason of complicated terrain in this district,a hierarchical segmentation method was proposed in this research.Then different segmentation parameters could be set according to different buildings.Based on the spectral characteristic of the vegetation,appropriate characteristic parameters could be chosen and subordination function is constructed.After then,land cover types in this district could be extracted step by step and at the same time could be compared with those by the traditional maximum likelihood method.The result indicates that extraction accuracy of the forest vegetation sub-category data in this Mentougou District is 83% by using the object-oriented classification method.Compared with the traditional method,the extraction accuracy has been boosted a lot.