SWEET POTATOES QUALITY ANALYZER BASED ON NATIONAL PHILIPPINES STANDARDS USING DIGITAL IMAGE PROCESSING AND MACHINE LEARNING

Abstract

Sweet potato (Ipomoea batatas) is an important root crop for the Filipinos and farmers are still performing the grading manually and inconsistently by mostly relying on their judgment. This research introduces a computerized system using MATLAB for the digital image processing of sweet potatoes and grading per the Philippine Bureau of Agriculture and Fisheries Standards and an automated classification. The system consists of image acquisition, preprocessing, HSV color segmentation, morphological filtering, and feature extraction to identify defects on the surface, pest damage, and irregularities of shape. The user-friendly graphical interface presents individual analyses, batch summaries, and standard references for easy understanding. Evaluation of samples of sweet potatoes showed an accuracy of 92.6% when compared to manual grading which indicates the system as reliable, consistent, and human errors reduced. The system is a low-cost, standardized tool for post-harvest quality control that supports farmers, cooperatives, and agricultural technicians in their efforts to get a better market position. Future developments may consist of mixing machine learning, multispectral imaging for the detection of internal defects, and mobile platforms for real-time and in-field assessment thus encouraging wider adoption and digitalization of the Philippine agriculture industry.