Textural Image Reconstruction and Recognition Based on Improved 2D-PCA

被引：0

作者：

Wang, Yan ^{[1
]}

机构：

[1] Beijing Vocat Coll Finance & Commerce, Beijing 101101, Peoples R China

来源：

2012 THIRD INTERNATIONAL CONFERENCE ON THEORETICAL AND MATHEMATICAL FOUNDATIONS OF COMPUTER SCIENCE (ICTMF 2012) | 2013年 / 38卷

关键词：

Texture recognition; Image reconstruction; 2D-PCA;

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：

A novel improved 2D-PCA method is proposed to reconstruct and recognize textural images in this paper. The canonical 2D-PCA method is widely applied in image classification and reconstruction as an extension of the traditional PCA method. However, because of the characteristics of texture images: rotation-invariant, unordered and so on, canonical 2D-PCA is not suitable to process textual images. In this paper, the improved 2D-PCA method is proposed to better solve this problem: the texture images are first preprocessed by axis normalization. The experimental results show the efficiency of the proposed algorithm.

引用

页码：162 / 168

页数：7