TY - JOUR AU - Tuates, Andie M. AU - Ligisan, Aileen R PY - 2016/10/18 Y2 - 2024/03/28 TI - Development of a Computer Vision System for Brown Rice Quality Analysis JF - Asian Journal of Applied Sciences JA - AJAS VL - 4 IS - 5 SE - Articles DO - UR - https://www.ajouronline.com/index.php/AJAS/article/view/4130 SP - AB - <h1>Conventional brown rice analysis is done by visually inspecting each grain and classifying according to their respective categories.  This method is subjective and tedious leading to errors in analysis.  Computer vision could be used to analyze brown rice quality by developing models that correlate shape and color features with various classification.  The objective of the study was to develop a computer vision system (CVS) for predicting quality parameters of brown rice.</h1><h6>Brown rice training samples were collected in Nueva Vizcaya, NFA Binalonan, Pangasinan, and SM supermarket. An ordinary flat bed scanner was used as image acquisition device coupled to a laptop computer equipped with image processing and analysis software developed at PHilMech. The CVS set-up was tested using samples collected at the regional NFA warehouses. The performance of the CVS was compared to human inspection based on their capability to classify brown rice samples.</h6><p>An artificial neural network using probabilistic neural network (PNN) model was developed. Sensitivity analysis revealed a true positive proportion ranging from 0.8792 to 1.00. Likewise, a weight prediction model based on the projected area was made using linear regression. The developed equation is y = 0.00148A – 0.00018 with a R<sup>2</sup> of 0.854.</p><p>The results of performance testing revealed that the CVS could predict the weight of brown rice and detect color-related quality of brown rice such as: sound, damaged, chalky/immature, yellow fermented, red, and paddy. Processing time for classification using the developed CVS has an average of 18.53 minutes and sixty percent of its time (equivalent to 11.24 minutes) was consumed in the manual arranging of grain samples. If a digital separation could be developed, the total time can be reduced to 7.11 minutes compared to 40.07 minutes of manual assessment. Moreover, CVS classification is more accurate compared with the human inspection.</p> ER -