Study on object recognition based on independent component analysis

被引：0

作者：

Huang, XM ^{[1
]}

Liu, CM ^{[1
]}

Zhang, LM ^{[1
]}

机构：

[1] Fudan Univ, Dept EE, Shanghai 200433, Peoples R China

来源：

ADVANCES IN NEURAL NETWORKS - ISNN 2004, PT 1 | 2004年 / 3173卷

关键词：

D O I：

暂无

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

This paper proposes a new scheme based on Independent Component Analysis(ICA) for object recognition with affine transformation. For different skewed shapes of recognized object, an invariant descriptor can be extracted by ICA, and it can solve some skewed object recognition problems. Simulation results show that the proposed method can recognize not only skewed objects but also misshapen objects, and it has a better performance than other traditional methods, such as Fourier method in object recognition.

引用

页码：720 / 725

页数：6

共 7 条

[1] Affine-similar shape retrieval: Application to multiview 3-D object recognition [J].