Non-closed surface reconstruction based on wavelet trans-form and maximum blurred entropy

摘要

Though Poisson surface reconstruction (PSR) method can reconstruct watertight surface effectively, even from very noisy cloud point, it cannot be used to reconstruct non-closed surface as yet. In this paper, a method based on wavelet transform and maximum blurred entropy is proposed, which is from the idea of image threshold to reconstruct non-closed surface with PSR. Based on enclosed surface reconstructed by PSR, and the perimeter of triangles formed the spurious closed surface are longer than the triangles formed the model surface, there is obvious boundary between them. Thus, an optimum threshold can be calculated to segment the spurious enclosed surface from reconstructed surface. Firstly, the threshold candidates are attained by using wavelet transform. And then, the maximum blurred entropy algorithm is utilized for obtaining the optimum threshold. The experimental results validate the method proposed, which is of lower complexity, efficiency, and robustness compared with the state-of-the-art methods.