Three-dimensional Measurement and Shape Recognition for Plant Factory

Abstract

A three-dimensional laser measurement system for crop was constructed for obtaining plant data for virtual plant factory. Also an algorithm for compressing range image with preservation of the shape characteristics was constructed.
The measurement system was composed of two units. One was a laser distance meter for measuring position of the sample along z axis, which had the resolving power of ±0.1μm. The other was an x-y axis robot for measuring the x- and y-position with an accuracy of ±10μm. The algorithm for determining the characteristic points of polygons was developed for compressing the size of the data of range image. The algorithm included a parameter, which was calculated based on the length of fruits in order to cope with different shapes.
Using compression algorithms, the size of three-dimensional measurement data was compressed to 1/1000 of the original. The shape of the polyhedron is comparable to the shape of fruits. The author suggested that this high accurate and useful measurement system is suitable for visual sensor of agricultural robot based on virtual reality.