Non-linear Growth Modeling of Greenhouse Crops with Image Textural Features Analysis

ABSTRACT: Nowadays, Machine Vision and Image Processing have become two important techniques in micro-precision agriculture. Vector extraction in textural features analysis is one of the principle methods in image processing. Entropy (randomness of gray-level distribution) and homogeneity (determination of the related gray-level pixel distribution amongst the surrounding pixels in the plant image) are two features in numeral image texture analysis. In this article, results of measurement of entropy and homogeneity are presented for greenhouse crop leaves’ image with a computer image processing method in an experiment. The objective of the current study was growth modeling with a machine vision system for tomato, cucumber and eggplant crops. The leaf samples were brought to the laboratory from a hydroponic greenhouse to measure the textural features. Results showed that the values of entropy and homogeneity were dependent to the plant growth for these three types of crops. The older plant leaves had more entropy and lower homogeneity than younger plant leaves. Finally, the relationships between the age of plant (in days) and textural features were modeled