Abstract
This research attempted to reveal that utilization of aerial photographs image range feature for classification of forest crowns by Same-Color-Region Determining Method of Castanopsis cuspidate coppice woodland and Cryptomeria japonica and Chamaecyparis obtusa artificial forest in northern part of Kyusyu, Japan. This research consist from two parts, one is the method improving applied by image range features such as standard deviation and cumulative frequency of tree crown edge for parameter setting. And the other is crown width classification accuracy was proved by field survey data of each tree crowns. As a results, supervised classifications of this improved method were capable of correctly classifying coppice woodland canopies with radii greater than 1.5 m (r<+-1.0 m) to an accuracy of 34.4%. For artificial forest, 41.4% of trees with tree crown radii greater than 1.0 m (r<+-0.5 m) could be correctly classified.
