Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.


Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  Paper number  013137,  2001 ASAE Annual Meeting. (doi: 10.13031/2013.7305) @2001
Authors:   Ta-Te Lin, Wen-Chi Liao, Chung-Fang Chien
Keywords:   L-system, Machine vision, Seedling, Plant features, 3D reconstruction

An integrated system combining stereo machine vision and 3D graphical modeling of vegetable seedling structure was developed in this study. The system comprised an image processing subsystem and a computer controlled rotary stage on which a plant seedling was placed. Seedling features were extracted from top-view image and lateral images taken from two color CCD cameras. Following the calibration and image registration procedure, the seedling images were segmented from background using a back-propagation neural network algorithm. Seedling leaves were then located with their center and orientation determined using blob analysis. Several other geometric features, such as seedling height, average projection area, leaf area index, leaf and stem node number, coordinates of stem nodes and leaf endpoints were subsequently measured. To utilize the extracted features, a graphical simulation model based on parametric L-system and bracketed L-system was built. The 3D graphical model allowed for quantification of measurable characteristics associated with individual modules, such as the length of internodes and the magnitude of branching angles of vegetable seedling leaves. The leaf surface was modeled with a Bezier surface specified by 16 control points. The texture of leaves was directly acquired from segmented seedling image or generated from a leaf image template. The accuracy of the system was tested with various types of vegetable seedlings. Measurement of growth curves of pepper seedlings cultured in various conditions using the stereo machine vision system was demonstrated.

(Download PDF)    (Export to EndNotes)