Individual tree is the basic unit of forest. Accurate detection of individual tree and efficient parameter acquisition are of great significance to improving management quality and production efficiency of forestry. In order to solve the problem of selecting the method and parameters of individual tree segmentation in subtropical coniferous forest, the key segmentation parameters of the watershed algorithm, point cloud-based cluster segmentation and layer stacking will be explored. At the same time, the individual tree clusters segmentation results of the three algorithms will be compared, and combine the recognition results with actual measurement data to evaluate the segmentation accuracy. The test results show that when the point cloud-based cluster segmentation and layer stacking segment individual tree clusters, there are over-segmentation or under-segmentation. But the watershed algorithm has the best segmentation results and the highest segmentation accuracy.