Journal of Researches in Mechanics of Agricultural Machinery

Journal of Researches in Mechanics of Agricultural Machinery

Spectral reaction investigation of healthy and fungi infected wheat kernels using imaging in visible and non-visible range and utilization of neural network

Authors
saman zohrabi alibeyglou
Seiied Sadegh Seiiedlou
ebrahim safari
yasaman hampanejad
Abstract
Wheat is one of the most important and strategica crops in Iran and the world. It is rich in protein and has a high economic value. Therefore, it is very important to determine the safety of wheat. The objective of this study was to identify wheat kernels infected by fungi, such as Penicillium Expansum and Fusarium Graminearum using setup of monochromator and imaging in visible and non-visible range. Alvand and Sardari wheat cultivars that allocated more cultivation area in the country were chosen for this purpose and three following comprehensive systems were used. The basis of the first system was employing detector and imaging visibly and non-visibly in the range of 630-680 nm of laser light. Setup of monochromator and image analysis in the visible and non-visible ranges were used in second and third system, respectively. For this purpose, CCD visible camera, visible-infrared and infrared cameras used for imaging of healthy and unhealthy wheat samples in ultraviolet, visible and infrared lights. In this system, eighteen features in RGB, LAB, HSV, HSI, YCbCr, and YIQ color spaces were extracted and used as neural network inputs. Wheat samples are classified into six classes of healthy Sardari, healthy Alvand, Sardari infected with Fusarium, Sardari infected with Penicillium, Alvand infected with Fusarium and Alvand infected with Penicillium.
Keywords
Wheat.-Neural network-Detection-Fungi-Near Infrared-