摘要
特征提取在模式识别和分类中起着关键的作用,本文针对红外目标的准确分类识别问题,围绕红外图像特征提取和特征选择进行研究,提出了基于PCA的区域特征选择方法。该方法首先采用基于数学形态学的滤波技术对红外图像进行预处理,有效地增强了目标区域,便于目标特征的提取;其次,本文研究了区域形状特征提取及基于PCA的特征选择方法,通过对区域特征进行优化选择,构造准确描述目标特性且维数较低的特征。实验结果表明,本文提出的方法有效地提取红外目标的特征,可用于红外目标的分类且有利于提高算法的效率。
Feature selection plays an important role in pattern recognition and classification. To deal with the infrared target recognition problems, the feature extraction and the feature selection of infrared images are studied, and a method of region feature selection based on PCA is presented. Firstly, an infrared image pre-processing method based on mathematical morphology is researched to suppress the effect of noises and to enhance the target area. Secondly, the method of region feature extraction and optimal feature selection are studied based on PCA, and the effective features are constructed for the target by the analysis of the PCA. The experiment results show that the method can effectively extract the typical region features of infrared image and reduce the dimension, it is beneficial to improve the efficiency of the algorithm of infrared target classification and recognition.
出处
《航空兵器》
2010年第2期17-20,27,共5页
Aero Weaponry
基金
航空科学基金项目(20070151003)
