IDEG>Research Demo> 10>


An improved spectral analysis method with additional
attributes for 3D objects



 We present a method to generalize the spectral analysis for the 3D mesh. It can accommodate attributes in addition to the geometry such as curvatures and vertex colors. The proposed method uses Gram–Schmidt orthogonalization to extend the attributes associated with a geometry coordinate when the Laplacian matrix is calculated. We calculate geometric features using a feature detection method, and these features are used as attributes after they are transformed to a color space. In addition, we also provide additional attributes by user interaction. The spectral analysis with attributes can improve results in geometric filtering such as smoothing or enhancement. Our experimental results demonstrate that the proposed method can efficiently filter 3D models compared to existing methods. It can also be applied directly to mesh parameterization by considering the eigenvector corresponding to the second smallest eigenvalue as the texture coordinates using the matrix of least square conformal mapping. It also shows better results than the ones by the existing spectral conformal mapping method.

Han Kyun Choi, Seung Joo Lee, Min Ki Park and Kwan H.Lee, "An improved spectral analysis method with additional attributes for 3D objects", International Journal of Advanced Manufacturing Technology, Vol.59, No.5, pp.761-773, 2012-03 [PDF]