Calculation method of surface shape feature of rice seed based on point cloud

Abstract

Surface shape feature is a very important index of seed classification. A method to calculate the three dimensional characteristics of the rice seed based on its surface point cloud was investigated. First of all, a laser scanner was used to acquire the three dimensional point cloud (TDPC) of a rice seed, then based on the principal component analysis (PCA) the TDPC were translated and rotated to normalize coordinate system. The length, width and thickness of rice seed, and furthermore its elongation, flakiness, shape factor and sphericity could be calculated by the oriented bounding box (OBB). We can generate mesh surface of the rice seed surface point cloud by the greedy triangulation algorithm, then calculate the surface area of the rice seed. On the basis of the slice method, an improved method based on rice seed TDPC was proposed to calculate the volume of rice seed. Compare the calculated length, width thickness and volume of different rice seeds with manual measurement. It is found that the average error is less than 1.5%. A ball whose radius is known instead of a rice seed to be used to verify the validity of the surface area by comparing the calculated value with their theoretical value, the mean error is 0.55%. The results showed the calculation method could acquire the surface shape feature of rice seed based on the TDPC and it is valid enough for application, this method is applicable for measurement of other irregular object as well.