H. Takizawa, N. Ezaki, S. Mizuno, and S. Yamamoto (Japan)
Agricultural application, Ubiquitous computing, Recogni tion of plants, Stereo vision, Reconstruction of plants
This paper describes a new method for recognizing plants by use of stereo vision for ubiquitous computing systems distributed over a farm. The proposed method can automat ically and non-destructively take three-dimensional mea surements of plants, such as height, width and leaf areas, which are important cues for determining the condition of plants. In addition, the method needs only two cameras and minimum preparations for plants, which are suitable for small computing systems like ubiquitous systems. In this method, first, a pair of color images for a plant is ob tained by a binocular stereo camera system. Then, the re gions of leaves and stalks are extracted from the images, range data is calculated from the (stereo) images, a three dimensional reconstruction for the plant is generated by use of the Delaunay triangulation, and the plant is taken three dimensional measurements of. Two experimental results are shown for actual plants.
Important Links:
Go Back
