Hass avocado ripeness classification by mobile devices using digital image processing and ANN methods
| Autor: | Carlos Augusto Meneses-Escobar, Gloria Edith Guerrero-Álvarez, César Augusto Jaramillo-Acevedo, William Enrique Choque-Valderrama |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
0303 health sciences
Engineering drawing biology 030309 nutrition & dietetics Process (engineering) Computer science Hass avocado Industrial chemistry Image processing Feature selection 04 agricultural and veterinary sciences biology.organism_classification Ripeness 040401 food science 03 medical and health sciences 0404 agricultural biotechnology Digital image processing Engineering (miscellaneous) Mobile device Food Science Biotechnology |
| Zdroj: | International Journal of Food Engineering. 16 |
| ISSN: | 1556-3758 2194-5764 |
| DOI: | 10.1515/ijfe-2019-0161 |
| Popis: | Proper farming, transportation, and storage processes of Hass avocado are important owing to its recent increase in production, export, and economic activity in Colombia. Since Hass avocado pricing and utility depend on its consumption ripeness, related to changes in skin color, sensory properties, texture, and nutritional value, developing an Android mobile application, namely iHass for smartphones and tablets, which estimates the number of days in which the Hass avocado reaches its optimal ripening level during post-harvest storage, contributes toward improving the fruit quality and decreasing the export costs and losses. This study aims to monitor the ripening processes of Hass avocados in complex backgrounds and indoor environments using various digital image processing techniques. The proposed study uses the red, green, and blue color model based on the physical and chemical changes that are observed during the ripening process. Herein, the color, shape, and texture characteristics of the fruits are obtained, and the fruits are classified using an artificial neural network, which features three layers, four input parameters, six hidden neurons, and four output parameters. Furthermore, ripeness was monitored in two crops, which provided 65 samples each. The results provided a ripeness estimate accuracy of 88% and a regression value of 0.819 during the post-harvest period. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|