DIGITAL IMAGE PROCESSING OF PLANTS IN TWO WAVE BANDS OF RED AND INFRARED FOR EVALUATION OF GROWTH

概要

論文の詳細を見る
In an attempt to extract clear pattern of plant image from the background, two wave bands of red (R; about 670nm) and infrared (I; about 900nm) regions were selected on the basis of spectral characteristics of leaf reflectance and used for taking R and I images in phytotron glass room by a TV camera. These images were digitized for computation and denoted as respective matrixes of M_R and M_I where each matrix element was expressed as reflectance (%). From M_R and M_I, difference in reflectance between R and I bands was given as M_s=[c_<ij>]=M_I-M_R, where the element, c_<ij>≦25% was treated as '0', and the element, c_<ij>>25% was treated as '1'. In M_s, the plant image was clearly separated from the background, and the plant growth appeared in increase in the number of digits of '1'. So, sum(P) of the elements of M_s was used as a parameter for evaluation of the growth: The fresh weight was represented reliably by a regression equation on P even when the plants were grown at different air temperatures of 22℃ and 25℃. The temperature effect on the plant growth was observed in increase in the fresh weights calculated from the regression equation in course of time. Thus, digital image processing of plants in red and infrared regions could be used as a tool of nondestructive and on-line measurement for the evaluation of plant growth.
九州大学の論文