Determination of optimal imaging parameters of hyperspectral images using the identification module

Federal Scientific Agroengineering Center VIM, build. 5, 1st Institute proezd, 109428, Moscow, Russia

^* Corresponding author: gwinnyandrei@yandex.ru

Abstract

This paper is devoted to imaging with hyperspectral camera on the developed prototype of optical system for identification of micro- and macrodamage of plant tissues. The presented device is used for imaging process to detect fruit diseases at an early stage. Hyperspectral monitoring is a proven non-invasive method for detecting foetal damage. Hyperspectral imaging is a collection of spatially resolved spectral images representing an object or landscape. This data can be used to determine characteristics and parameters of fruit that indicate the presence of damage. This study reports on a triplicate experiment to determine the optimum parameters at which a hyperspectral image suitable for further detection and analysis of apples is obtained without the detection process slowing down the conveyor belt. This paper presents an optical identification device, which is a set of guides and racks on which a platform with a holder and a hyperspectral sensor attached to it moves. The recording and analysis of the hyperspectral cube was carried out using BaySpec’s SpecGrabber and Gelion. Data processing of the planned experiment was carried out in the MatchCad 15 environment and the PlanExp B-D13 v.1.0 programme. To find the combination of factors giving the optimum value of the output parameter. In the course of calculations the results of statistical processing of the obtained data and the conclusion on the adequacy of the mathematical model were obtained, on the basis of which it was decided that the model is applicable for solving production problems and the extrema of the response function were obtained, which correspond to the values of optimal factors. The main parameters affecting the acquisition of hyperspectral images suitable for subsequent analysis are the speed of movement of the bracket with the camera, the illumination of the biological object and the height of the camera installation. These parameters are variable and their combination directly affects the image quality expressed in pixels.