Search academic texts

Information × Registration Number 0224U032633, (0124U004787) , R & D reports Title To develop a pattern analysis and recognition system of grain materials using machine vision popup.stage_title Розробити процес аналізу та розпізнавання зернових матеріалів з використанням машинного зору Head Stepanenko Serhii P., Доктор технічних наук Registration Date 27-11-2024 Organization Institute of mechanics and automatics of agroindustrial production of the National Academy of agrarian sciences of Ukraine (imaap naas) popup.description1 The purpose of the research is to develop a machine vision system that could effectively obtain images suitable for further processing and to develop an algorithm for recognition grain materials using image analysis, which will enable effective control and determination of grain quality popup.description2  The study of the physical, mechanical and optical characteristics of grain materials was conducted. The dependences of the main physical and mechanical properties of the processed wheat material on its humidity were obtained: coefficients of friction on different surfaces, angle of natural slope, coefficients of shear resistance. Based on the results of the study of the optical characteristics of grain materials, the dependences of the change in the spectral coefficients of reflection and transmission of radiation in the visible and near infrared ranges by samples of various grain materials were obtained. The contour sizes and average color coordinate of grain contours were studied by photographing and processing digital images using the OpenCV library. The dependences of the influence of sample illumination and camera parameters on the values of the scale pixel coefficient and deviation and deviation from the real color were obtained. The current state of machine vision systems was analyzed by reviewing literature sources. Wheat seed samples were selected and analyzed, and samples of wheat grain material components were photographed to form an image database. A algorithm was developed for the analysis and recognition of grain materials by sequential processing of digital images. For its implementation, computer vision and deep learning technologies were used based on OpenCV, TensorFlow and Keras libraries. Fragments of program codes were written in Python for: forming a database of grain mixture components; creating and training a convolutional neural network (CNN); image segmentation and classification of components. Based on the results of processing digital photographs, a data set was collected, consisting of over 3,000 photographs of wheat grains and over 600 photographs of impurities. A software structure has been developed for the analysis and recognition of grain materials. Product Description popup.authors Borys Andrii M. Volyk Daryna A, Dnes Viktor І. Yovenko Nataliia Yu. Kuzmych Alvian Ya. Popadiuk Ihor S. Shvydia Viktor О. popup.nrat_date 2024-11-27 Close