Development of Algorithms for Automatic Separation and Extraction of Canopy Cover, using Laser and Multispectral Data

要旨

The automation of extracting objects from remotely sensed data has become more and more urgent as required by GIS (Geographical Information System) users from variety of industries such as telecommunication, commercial, real state, transportation and forest management etc. This paper presents an approach for generating GIS polygon of canopy structure including elevation and geometry from LIDAR (Light detection and ranging) and multispectral data acquired from same airborne platform. The approach is fully automated, efficient and especially good for producing wide forest area canopy model at minimal cost. Its accuracy is usually constant throughout the entire area covered. In first process manmade and natural objects were separated by automatic threshold of histogram of NDVI and nDSM. In second step individual trees were extracted using laser radar and multispectral data. The segmentation was done in two steps first using raindrop model and second ray algorithms with testing of compactness and anisomery with HDevlope and C++.