| Budget Amount *help |
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2017: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2016: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2015: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
|
|---|
| Outline of Final Research Achievements |
We developed the method for automatic classification of individual tree species from laser scanned point cloud using deep learning. In our method, given a point cloud of an individual tree, point subset at breast height is selected, and then branches and leaves are removed by RANSAC-based circle fitting. Next the images are created which clearly represents bark texture using bi-cubic surface fitting or curvature estimation. These images are finally used in deep learning for species classification. From various experiments using point clouds of Japanese cedar and cypress trees within 15 meters from the scanned position, our method achieved high classification performance over 90%.
|
