一种坦克装甲车识别方法

坦克装甲车识别在军事图像处理方面具有重要的意义,对于实现对地面目标进行精确打击具有重要的实际用途.本文主要围绕复杂背景中目标图像区域识别方法进行了深入研究,重点阐述了图像二值化迭代选择阈值法、图像特征提取的目标区域Zernike矩计算法、图像分类识别的神经网络技术等重点技术,介绍了一种行之有效的识别方法,实现了在包含地面军事战术目标初始图像中对坦克装甲车区域的较准确识别.通过对近百幅坦克、装甲车等图像的试验测试,识别率达到了95%.

Method of tank and armored car recognition

Select the threshold iterative,this threshold is used to create the binary image then the area of the tank or the armored car is extracted.The Zernike moment of the area is used as the feature for recognition.The object can be classified by the neural network.With more than 100 experiences of tank and armored car images,the result of the experiment shows that the method is efficient,the rate of correct arrives 95%.