| Summary: | 碩士 === 國立臺灣大學 === 農業機械工程學系研究所 === 87 === The measurement of plant growth is helpful in understanding the interaction between plant growth and environmental factors. Therefore, we developed an intelligent machine vision system to determine seedling morphological characteristics non-destructively in this research. The image segmentation of seedling from background natural lighting was discussed and an operating algorithms to measure seedling morphological characteristics was developed. Preliminary information was obtained from the top view image and side view images of the seedling. Seedling characteristics such as leaf number and leaf angle on the top view image were first determined. According to these characteristics, the side view images were acquired by rotating the rotary stage to specified angles for later analysis. From the side view images, other seedling characteristics, including seedling height, span, projection area and total leaf area were then determined by calculation and calibration. For seedling within 6 leaves, the error of leaf number was within 1 leaf, and for seedling within 5 leaves, the accuracy of leaf area was 92%. The measured data can be further processed for 3D reconstruction of graphical display of the measured seedling.
|
|---|
