Computer Control of Plant Growth by Image Processing

Abstract

The feedback information about plant growth is needed in the computer control system to optimalize the environment for maximalizing production. The present paper reports a method for evaluating physiological and pathological states of cucumber plants, by means of image processing of reflectance of plants or leaves in different wave length regions of light. From the analysis of spectral reflectance of cucumber leaves, it became clear that states of plants can be characterized by respective reflectances in wave length regions of 650-690 nm (R) and 800-1000 nm (I), and evaluated by the ratio of two reflectances. So, R- and I-images of reflectance were taken by the vidicon camera through filter systems having transmitting peaks at 670 nm and at 900 nm, respectively, and digitized in a mesh of 240×256 words (6 bits/W) . The physiological and pathological states were observed clearly in the digital display of those images. In order to obtain an index for evaluation of those various states of leaves, the ratio (Br) of reflectances in I- and R-images was calculated from Eqs. (2) and (3) . In general, the reflected light intensity varied with the angle of incidence of light, owing to the specular and spread reflections. However, this disturbance caused by the angle of incidence was eliminated enough to characterize the spectral reflectance of leaf by using Br. So, the Br was examined in various states of cucumber leaves. On the basis of the fact that the frequency distribution of Br of normal leaves was approximately normal, the interval containing 99% of total of Br's of normal leaves was given as 7.6≤Br≤12.7, and Br= 7.6 and Br= 12.7 were used as the classifiers for discrimination between normal and abnormal (damaged and diseased) leaves. Thus, the present method for image processing of reflectance of plants made it possible to obtain a feedback information about vigor of plants, which can be used in an on-line computer system for environmental control of plant growth.