Advanced Machine Vision Algorithms for Fruit Segmentation in Agricultural Environments
Machine vision systems are one of the most reliable systems in the agricultural and food products field, yet there are several challenges with their application. One of the main issues with machine vision systems is the segmentation of fruits in images. For example, cluttered environments such as greenhouses complicate the task of segmenting fruits, especially green tomatoes. This thesis focuses on segmenting fruits in images taken in agricultural environments in the form of two case studies. The first case study discusses the segmentation of tomatoes that are either single or clustered under different lighting and clutter conditions. In particular, this case study compares the segmentation accuracy of four different segmentation algorithms that are based on morphological operations and machine learning. Furthermore, the second case study discusses the segmentation of peaches for storage applications. The accuracy of the algorithm result is, then, compared with the ground truth data to evaluate the developed segmentation algorithm.