Two novel methods for surface profile extraction based on multiple ultrasonic range measurements are described and compared. One of the methods employs morphological processing techniques, whereas the other employs a spatial voting scheme followed by simple thresholding. Morphological processing exploits neighbouring relationships between the pixels of the generated arc map. On the other hand, spatial voting relies on the number of votes accumulated in each pixel and ignores neighbouring relationships. Both approaches are extremely flexible and robust, in addition to being simple and straightforward. They can deal with arbitrary numbers and configurations of sensors as well as synthetic arrays. The methods have the intrinsic ability to suppress spurious readings, crosstalk and higher-order reflections, and process multiple reflections informatively. The performances of the two methods are compared on various examples involving both simulated and experimental data. The morphological processing method outperforms the spatial voting method in most cases with errors reduced by up to 80%. The effect of varying the measurement noise and surface roughness is also considered. Morphological processing is observed to be superior to spatial voting under these conditions as well.