基于局部不变特征的空中红外面目标的模糊识别

Fuzzy Recognition of Air borne Infrared Planar Targets Based on Local Invariant Features

红外面目标识别是红外成象精确制导的关键，由于目标与象之间存在着透视投影变换关系，则必然导致成象的形变；另外，红外成象比较模糊，加上背景、干扰弹及目标自身的影响，在象平面经常出现遮挡及不确定性．若用基于全局的特征进行识别，极易造成误识．针对空中红外面目标的成象特点，用交比构造目标的局部不变特征，提出用局部特征的相关性结合目标的对称性进行模糊决策，抑制由于局部畸变造成的误识．对成象有缺陷的目标进行识别结果表明，算法有较强的抗遮挡能力．

It is important to recognize infrared planar targets for IR imaging accuracy guiding. The imaging distortion will be faced, because there exists imaging deformation caused by perspective projection transformation from target to image and there often exist uncertainty and occluding caused by image blurring and background, deceit, etc. in the image plane. The misclassification is too often made by using global features for recognizing. A fuzzy recognition algorithm is presented by using relativity of local invariant features combined with target's symmetry. Local invariant features are made up from target's cross ratio with respect of air borne IR planar objects. The proposed algorithm can supress misclassification caused by local distortion successfully. A number of targets with some imaging default are tested, as a result, the algorithm is of strong anti occluding.