The Image Segmentation of Osmanthus fragrans Based on Optimization Algorithms

Image segmentation is of great importance in many areas including artificial intelligence, image processing, computer vision, etc. The optimization of image segmentation can be converted to seek the best threshold values to divide an image into specific objects and background. In this study, we have used the newly proposed AO, G-NHGWO, GBO, LPB and LA optimization algorithms to identify the optimal threshold values for image segmentation of Osmanthus fragrans flowers. The widely used Kapur, Shanbhag, weighted Otsu, Kittler, Yen and Li methods were utilized as objective functions for thresholding selection. The image segmentation results have exhibited the ability and superiority of employed optimization algorithms and thresholding strategies. The research of efficient image segmentation of Osmanthus fragrans in this study is helpful to the machine recognition and classification of the widely cultivated plants, which will improve the automatic flower picking and industrialization of Osmanthus fragrans.