Process Neural Network Method: Case Study I: Discrimination of Sweet Red Peppers Prepared by Different Methods

This study utilized a feed-forward neural network model along with computer vision techniques to discriminate sweet red pepper products prepared by different methods such as freezing and pureeing. The differences among the fresh, frozen and pureed samples are investigated by studying their bio-crystallogram images. The dissimilarity in visually analyzed bio-crystallogram images are defined as the distribution of crystals on the circular glass underlay and the thin or the thick structure of crystal needles. However, the visual description and definition of bio-crystallogram images has major disadvantages. A methodology called process neural network (ProcNN) has been studied to overcome these shortcomings.