Terrain Feature Line Extraction by Improved Gradient-Based Profile Analysis
Kun Zou, Hongzhang Weng · International Journal of Signal Processing Image Processing and Pattern Recognition · 2017
The traditional terrain feature line extraction method using gradient-based profile analysis may induce feature discontinuity along flat ridges or valleys, and suffers a lot from terrain noises.To conquer this problem, an improved gradient-based profile analysis algorithm is proposed, and applied to the extraction of ridges and valleys from digital elevation models (DEMs).Firstly, the improved Sobel filter is adopted to capture the gradient direction, based on which the optimal profile direction is determined.Secondly, a global profile analysis process is devised to identify both strong and weak feature points.Then, features are extended along their feature directions, and final feature candidates are constructed by strong feature points and nearby weak ones with coinciding feature directions.After that, feature thinning, loop breaking, pruning and smoothing are conducted in succession to obtain final feature lines.Experiments are made by testing the proposed method and the traditional method on some real DEM data.It is observed that the proposed method performs much better in preserving feature continuity and resisting noises, and it can produce decent feature lines conforming to the terrain reality.