可见光图像中飞机目标的特征选择及提取

Selection and extraction of features of aircraft in optical image

摘要:

针对目前大部分飞机目标识别算法都依赖建立的样本库,导致对样本库外的飞机不能很好识别的问题,分析了可见光图像中飞机目标在模式识别中的特征选择及提取方法.比较几种不同特征选择及提取方法,依据人工判读识别飞机目标的思维方法以及可见光图像的特点,提出了可见光图像中飞机目标的5个典型识别特征的提取方法,最后利用模糊模式识别方法完成对飞机目标的识别.实验结果表明:该算法能够对可见光图像中的飞机目标进行较好识别,准确率较高,平均识别率达到了90%以上.

Abstract:

At present most of the aircraft recognitions rely on sample database,which leads to poor accuracy for aircraft outside the database.The problems of feature selection and extraction for aircraft in optical image were studied through analyzing the principle of the methods commonly used in pattern recognition.Extracting method for five recognition features based on human way of thinking and the characteristic of aircraft in optical image was proposed.Fuzzy pattern recognition was applied in the aircraft recognition.Experimental results show that the method can accomplish aircraft recognition with high accuracy.