基于多特征融合的多视点飞机目标识别算法

Aircraft Target Recognition Based on Multi-features Fusion Under Multi-views

飞机作为三维目标,在不同视点观察下,会呈现出不同的姿态。针对多视点下,飞机目标姿态多变而导致识别率较低的问题,提出了一种基于DSm T理论和SVM分类器相结合的多特征融合的多视点飞机目标识别算法。首先对图像进行二值化处理,提取多个特征量;然后针对DSm T理论中信度赋值构造困难问题,利用SVM分类器对证据源进行信度赋值;最后利用DSm T组合规则在决策级进行融合,从而完成对飞机目标的识别。实验验证该算法在飞机姿态发生较大变化时,依然能够获得较高的识别率。

The aircraft as a 3D object,the position of the plane will change in the different angle of view. Aiming at the low recognition rate of aircraft under multiple viewpoints and multiple positions,an aircraft recognition method based on DSm T( Dezert-Smarandache theory) and SVM( Support Vector Machine) classifier on multiple features under multi-views is proposed. Firstly,the image is preprocessed with binarization and then multiple features are extracted. Moreover,for the problem of the construction of the basic belief assignment in DSm T is difficult,the basic belief assignment is proposed to assign the evidence sources by SVM classifier.Finally,based on DSm T combination rules on decision level fusion the aircraft target recognition are completed. According to the experiment results,this algorithm has a high recognition rate even though aircraft under multiple viewpoints and multiple positions.