Hierarchical multi-scale segmentation of LiDAR images in forest areas

A two-level hierarchical method for LiDAR image segmentation in forest areas is proposed. This method represents a multi-scale analysis of LiDAR images by an attention operator at different scale ranges and for all pixels to detect feature points. As a result, the feature points as optimal seed locations for regiong-rowing segmentation are extracted and scale-adaptive region growing is applied at the seeds. At the second level, the final segmentation by the scale-adaptive region growing provides individual tree crowns. The conducted experiments confirmed the reliability of the proposed segmentation method and have shown its high potential in LiDAR image analysis for object detection.