军事目标红外图像的形状识别系统

A Shape Recognition System of Infrared Target Image

随着目标自动跟踪、识别技术的发展,对目标特征提取技术的要求变得愈来愈高,基于目标形状的跟踪、识别系统的发展非常迅速。本文针对目标红外图像的特点,提出了一种新的目标形状分类、识别系统,该系统包括目标图像预处理、目标分割提取、目标边界探测以及目标形状分类、识别。系统中,我们采用一种均值——中值混合滤波方法来消除噪音的影响,利用峰谷法分割提取目标,进而提取目标的边界,在目标边界提取的基础上,提出了边距离分布特征的概念用于目标的分类、识别。

The development of target automatic tracking and recognition needs more effective technology of target features extraction.In this paper,a new infrared target shape classification and recognition system is proposed.The system function includes target image pre-processing,target segmentation,contour extraction and target shape classification and recognition.The authors have developed a mean-median hybrid filter (MMHF) method to suppress noise,used peak-valley method to segment the target and developed a new target shape description method to classify and recognize the target.