<p>In the proposed study, a new region-growing-based fast hybrid segmentation (FHS) technique has been developed for fast segmentation of unorganized point clouds. This algorithm is robust and segments the point clouds based on three parameters simultaneously, which are normal vector, curvature and Euclidian distance. This makes the segmentation of point clouds with planar and curved regions even faster. The algorithm uses a point-wise approach instead of the cluster-wise approach and thus avoids passing over every point in each nested loop which significantly reduces time and memory consumption. Moreover, during the segmentation of planar regions, it also avoids Euclidean distance calculation for every point giving even faster results. In this study, a new elbow technique for the automatic selection of an adequate number of clusters and a dedicated algorithm for the proper selection of the initial seed points have been developed which improves both quality and computational speed. The proposed method gives better results in terms of accuracy and smoothness at low computational cost and it is tested on many point cloud data at various parameter settings. The proposed algorithm is very efficient and effective for instant segmentation. It segments the planar areas about 6–8 times faster than the classical segmentation methods and for non-planar areas, it works about 4–6 times faster while generating high-quality results.</p>