Two-dimensional local moment, surface fitting and their fast computation

作者:

Highlights:

摘要

It is proved that local moments are related to their surface fitting coefficients by a linear transform, and by this the surface-fitting-based-feature extraction is converted into the computation of 2D local moments. A NPT recursive method is proposed to compute 2D local moments quickly and then to efficiently implement surface fitting. The new method has a novel advantage that its computational complexity for both 2D local moment computation and surface fitting is independent of the mask size, which much efficiently speed up the feature extraction.

论文关键词:Local moment,Surface fitting,Polynomial,Monomial,Negative Pascal transform (NPT),Range image segmentation,Edge detection,Computational complexity

论文评审过程:Received 2 February 1993, Revised 21 September 1993, Accepted 1 November 1993, Available online 19 May 2003.

论文官网地址:https://doi.org/10.1016/0031-3203(94)90163-5