Author : Tuates, Andres Morales Jr.

Summary : Responding to the need of an automated milling quality assessment for brown rice, a computer vision system was implemented to reduce the tedious and subjective manual method of evaluation. An ordinary flat bet scanner and a digital camera were used as image acquisition devices coupled to a laptop computer equipped with image processing and analysis software. The performance of the scanner and camera were compared on their capability as acquisition devices. The artificial neural network using probability neural network (PNN) model was developed and generated a true positive proportion for the scanner ranged from 0.8792 to 1.00 while the camera ranged from 0.8409 to 0.9851. Results of the training and verification revealed that the test images acquired using the scanner and camera attained above 90 percent efficiency in all classification parameters. The performance of CVS using two image acquisition devices and four different varieties of brown rice attained an average accuracy of 94.21 percent. Processing time for classification using the developed CVS average to 13.35 minutes compared to 32.36 of manual assessment. The estimated investment cost in putting up a computer vision system developed in this study ranged from P94,000 and P105,000.