Proposing Improved FCM based Method for Segmenting Irregular Shaped Fruit Image Captured in Natural Light

Abstract

The latest improvements in computer vision have made it possible to use this technology in almost every aspect of human life. This has made a lot of new things possible. Since putting fruits and vegetables into groups has been shown to be a difficult task that needs more research and development, this application sector has become very important. Because of how similar things are within the same class and how different things are within the same class, trying to classify fruits and vegetables presents a number of problems that must be solved before success can be reached. Because there are so many different uses, it is important to choose the right sensors for gathering data and a way to represent features. Even though methods have been made to classify fruits and vegetables so that their quality can be judged and they can be picked by robots, the current state of the art can only be used with small datasets and a small number of classes. One of the biggest problem that modern machine learning algorithms have to deal with is that the problem is often multidimensional and there is often a lot of input that is very high dimensional. A lot of time and effort has been put into making and analysing classifiers for hyperdimensional characteristics, which need a lot of computing power to be optimised. In real-world applications over the past few years, many different feature description methods and machine learning techniques, such as Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Decision Trees, Artificial Neural Networks (ANN), and the model being proposed here, have been used to classify fruits and vegetables. In this paper, we look at all of the new computer vision approaches to classify fruits and vegetables that have been proposed by the scientific community. People have said that these ideas are on the cutting edge.