Smart Potato Grading using Image Processing and Fuzzy Grading System

Horticultural commodities have distinctive and unique characteristics, such as different colors, shapes, and sizes for each harvest. Grading is currently being done, generally based on these characteristics. The grading person must be an expert and maintain the respective consistency for each commodity characteristic, which requires more time, cost, and risk. This study aims to design a smart potato grading structure by combining different parameters: color, shape, and size. We use image processing with a fuzzy grading system. The camera correctly captures the potato image and is processed by the image processing room. Captured images and database images provide an output rating by comparison. The system offers 2-D image parameter detection—color, shape, and size count. We also design knowledge base fuzzy rules. This system gives better results compared to the current grading. The results showed that by using the fuzzy logic grading system, the grading accuracy was 86%. There are still errors in reading the color of potatoes (14%). For further research, researchers can develop this smart grading to determine potatoes' quality.

I. Introduction

Indonesia is the largest producer in Southeast Asia and the 14^th worldwide, with the most prominent producers being West Java (277,187 tons), Central Java (269,476 tons), and East Java (241,180 tons) [1][2]. However, nationally, there was relatively no increase in potato production, 1,094,232 – 1,235,180 tons/year, with low productivity of 16.4 tons/ha [3].