ICIST 2019 Conference

Development of a Non-destructive image processing System to Improve Rice Quality
Sidiq Syamsul Hidayat*); Jumi; Roihatin; Abdul Syukur Alfauzi

Politeknik Negeri Semarang, Indonesia, Jl. Prof Soedarto SH, Semarang, Indonesia

*) corrosponding author e-mail: sidiqsh[at]ieee.org


Post-harvest problems faced by farmers in the form of grain drying, have been overcome by previous studies using a Wireless Sensor Network (JSN) dryer. The use of JSN in dryers is able to provide solutions to improve the accuracy of temperature and water content measurements in the drying chamber. Problems that arise after field trials are (1) the actual moisture content in the grain cannot be determined precisely and quickly (real time), (2) the separate dryer and grinder system has some disadvantages a.l. have to move the dried grain to the grinding machine, adding power, time and cost. In addition, many grains are wasted due to material handling, causing losses. For this reason, it is necessary to develop an integrated system (two-in-one) and that is equipped with quality sensors to determine the quality of grain and rice produced automatically. Grain and rice have different colors at different humidity levels. The use of humidity sensors based on grain color and rice color allows the measurement of water content to be done accurately and faster (real time). Grain and rice quality level sensor in this study uses image recognition based on its features. To determine the level of grain moisture recognition in this study using color features with the method of preprocessing edge sharpening and histogram equalization. Whereas the rice quality level classification sensor uses a combination of shape, color and texture features using the Invariant moment extraction method for shape features, color moments for color features and GLCM for texture features and K-means as descriptor clustering.

Keywords: rice quality; machine learning; sensor network

Topic: Electrical Engineering and Computer Science

Link: https://ifory.id/abstract-plain/Ge7JzmxL69cw

Web Format | Corresponding Author (Sidiq Hidayat)