Machine Vision System for Plant Morphogenesis Analysis

Abstract

An image processing system for investigating plant morphological characteristics of two or more plants was developed and demonstrated. A CCD camera was mounted on a stepping motor with ten plants equally spaced around it. A computer captured and stored the image of the first plant, rotated the stepping motor 18 degrees and captured the image of the second plant. This operation was repeated until all plant images were captured. In order to achieve similar gray levels during day and night photography, image acquisition was performed using an 850 nm infrared LED as a light source as well as an infrared filter attached to the CCD camera. The stored images were analyzed using sub-pixel estimation with area-based pattern matching techniques. The reference image size within the sub-pixel pattern matching process was evaluated. It was clear that the reference image greater than 15×15 pixels in dimension was capable of detecting a 0.1 pixel displacement in plant images used as part of this research. This system was employed to measure stem elongation in ten Chrysanthemum morifolium Ramat ‘Reagan’ plants every 10 min for three days. Daily variations in stem elongation were clearly observed and it was found that stem elongation was greater during the night than during the day. Substantial increases occurred during the first two hours of night conditions.