Milkfish freshness detection utilizing Coiflet wavelet transform method

Abstract

Fish is a perishable good, which can get easily putrefied in a short span of time. To maintain the fish’s quality freshness must be preserved, so as not to cause harm to the consumer. This study is aimed at determining a classification technique on which to base the milkfish’s freshness using a modified set of features like wavelet transform coefficients and support vector machine. The Support Vector Machine (SVM) is the tool in which forms a link between feed-forward neural networks and Coiflet wavelet filter. The Coiflet filter uses six scaling functions to increase the pixel averages and differences, resulting in a smoother wavelet and increased performance capability. As a result, the system’s output showed that the accuracy result of the eyes is 89.9%, gills 90.9% and body 88.8%. This framework can significantly be beneficial to milkfish growers, dealers and consumers. In general the study shows that the method gains better performance in terms of freshness percentage. However, the system’s performance may still be optimized by increasing the number of data sets and classifiers to be used for a specific purpose.