Efficient planar affine canonicalization

作者：

Highlights：

• A method based on the joint angular distribution of marginal skewness and kurtosis.

• The canonical shapes are computed in closed form from the raw image moments.

• Shape recognition and alignment is robust to the non-affine distortion found in natural images.

• Useful as a general preprocessing step to achieve affine invariance in pattern recognition.

摘要

•A method based on the joint angular distribution of marginal skewness and kurtosis.•The canonical shapes are computed in closed form from the raw image moments.•Shape recognition and alignment is robust to the non-affine distortion found in natural images.•Useful as a general preprocessing step to achieve affine invariance in pattern recognition.

论文关键词：Shape recognition,Affine canonicalization,ICA,Kurtosis

论文评审过程：Received 24 December 2016, Revised 26 May 2017, Accepted 17 July 2017, Available online 20 July 2017, Version of Record 29 July 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.07.017