Identification of Mustard Greens Freshness Level Based on RGB Leaf Color and Stem Shape Features using Image Thinning Morphology

Abstract

Fruit and vegetables are the commodity with high commercial value and have many benefits. One of the popular and favorable vegetables in Indonesia is Mustard greens (Brassica juncea) since it has good taste and easily gets at an affordable price. Indonesian and Chinese food often used this vegetable as soup or stir fry. Like other leaf vegetables, Mustard greens can easily damage. It can easily wither since it contains much water. Wide and large leaf surface make withering and evaporation process faster. Keep the freshness level in vegetable storage is one of the important criteria in the market to maintain high-quality products. To sort out a high volume of products, stores need an automatic, fast, and efficient Mustard Greens system for sorting out their freshness level. The improvement of digital image processing research and technology makes it possible to identify agricultural and plantation products. Combining this technology with some mechatronics automation, the sort out system could work automatically. The purpose of this research is to identify Mustard Greens Freshness by visual analysis of their images acquired using Smartphone camera. The feature extracted from the image is the shape of the stem and their leaf color. The feature extraction method for stem shape is done using Binary Morphology. First Order statistical method from the image histogram used for leaf color extraction is Mode. Single Layer Perceptron used as a Neural Network Classifier. The classification will determine best weight value in the network for 3 categories of freshness: fresh, slightly withered and very withered. From the experiments result, Classification accuracy for identifying the freshness of the Mustard greens using Single Layer Perceptron is 90%. Conducted classification accuracy results indicate that the methods good enough for classify freshness level of Mustard Greens using color and stem shape feature.