Forest pest monitoring image analysis based on fuzzy pixel clustering and watershed approaches

Abstract

The orthophotos of Pinus tabulaeformis and seabuckthorn were collected by UAV, these images were used as test images, and the performance of six image segmentation algorithms were qualitatively analyzed and quantitatively compared such as fuzzy pixel clustering and watershed algorithms. The error rate, relative final measurement accuracy, and running time are used as evaluation criteria. The experimental results show that the segmentation algorithms’ performance of the affected forest image is closely related to the image-capturing height and noise. Finally, the guiding suggestions for the application of the orthophoto segmentation algorithm are given from unmanned aerial vehicles in the affected forest area.