Abstract: In order to improve the destroying effect of air-to-air missile based on infrared imaging guidance，a recognizing method for the airplane’s key attacking part is proposed，which combines the recognizing of flying gesture with geometry contour analysis.The infrared airplane image is separated into background，airframe and plume by the fuzzy-c means clustering.The composed invariant moment is computed，which consists of the Hu moment and the normalized moment of inertia.Then the flying gesture is judged by the composed invariant moment and the ratio of the different parts’ area for airplane in infrared image.The airplane’s head，shaft，wing and tail are abstracted and the key attacking part，for example，cockpit，is recognized by the geometrical proportional relation of these parts.In order to improve the discerning speed and precision，the processing for the subsequent image frames is guided by the steady judging parameters which are acquired by the prior frames’ processing.Experimental results show that the method given by this paper can identify the airplane’s key attacking part in infrared image effectively and provide essential support for the air-to-air missile’s precise strike.

Key words: infrared image, airplane’s flying gesture, geometry contour, key attacking part

摘要： 针对红外成像制导空空导弹增强攻击毁伤效果的要求，提出了一种基于红外图像的飞机飞行姿态识别和飞机目标几何轮廓分析的关键攻击部位识别方法。该方法利用模糊C均值聚类算法将经过降噪处理的红外飞机图像分割成背景、飞机机身和尾焰三部分;分别计算飞机目标的Hu矩和归一化转动惯量来构成组合不变矩，同时结合分割图像的面积比特征来实现飞机飞行姿态的识别;通过对飞机机头、机轴、机翼或尾翼的提取并依照相应的比例关系，来实现对飞机关键攻击部位如驾驶舱的识别。为了提高关键攻击部位的识别速度和精度，利用当前帧获得的稳定判别参数来指导后续帧图像的处理。试验结果表明，提出的方法能够准确有效地识别红外飞机目标的关键部位，可为空空导弹的精确打击提供必要的支持。

关键词: 红外图像, 飞机飞行姿态, 几何轮廓, 关键攻击部位