Outline shape measurement of rice seed is key to non-destructive testing of rice seed. On the basis of point cloud on rice seed surface acquired by laser sensor

present an outline shape measuring method of rice seed based on main plane of projection. By principal component analysis method

calibrate coordinate of point cloud model

obtain six main plane of projection of seed rice seed using principal plane projection method

apply Alpha Shape algorithm to extract edge outline of the projected point cloud model

and calculate nine shape feature parameters of perimeter

area

length

width

aspect ratio

maximum radius

minimum radius

radius ratio

circularity of rice seed projection model. Select three categories of rice seed

200 grain per each class as a test sample

using RBF neural network model for training recognition extracted shape feature

recognition rate reach 96 %

94 %

98 %

respectively. The results show that the outline shape measurement method of rice seed based on laser sensor can well be applied to identification of rice seed.