Abstract: There are two problems of smoothing of a pointcloud, which is the dilemma between moving noise points and keeping the characteristic boundary, the expanding of noise points, during the reverse engineering and free-form surface measure- ment. The principle of anisotropic heat diffusion is adopted to develop a mathematic model of pointcloud’s smoothing. A pointcloud can be considered as a field of temperature field, boundary as adiabatic. So the smoothing is only on the local area and makes a good balance between moving noise points and keeping the characteristic boundary. Experimental results indicate that the method is robust. It gives a reliable guarantee for next modeling and free-from surface measurement.

Key words: Heat diffusion, Pointcloud denoise Reverse engineering, Smoothing, Dynamic Simulation, Heat Exchanger, response time, Simulink, Organic Rankine Cycle