Filtering Green Vegetation Out from Colored Point Clouds of Rocky Terrains Based on Various Vegetation Indices: Comparison of Simple Statistical Methods, Support Vector Machine, and Neural Network

Filtering out vegetation from a point cloud based on color is only rarely used, largely due to the lack of knowledge suitability input information (color, indices) and thresholding methods. We have evaluated multiple indices (ExG, ExR, ExB, ExGr, GRVI, MGRVI, RGBVI, IKAW, VARI, CIVE, GLI, VEG) combined them with 10 methods threshold determination training set selection (including machine learning methods) renowned Otsu’s method. All these combinations were applied four clouds representing vegetated rocky terrain, results compared. The ExG GLI generally most suitable for this purpose, best F-scores 97.7 95.4, respectively, best-balanced accuracies same combination method/vegetation index 98.9 98.3%, respectively. Surprisingly, achieved using simplest method determination, considering single class (vegetation) normal distribution. This algorithm outperformed all other methods, including those support vector deep neural network. Thanks its simplicity ease use (only several patches must be manually selected as set), can recommended removal anthropogenic surfaces.