基于自适应神经网络的遥感图像飞机目标检测方法

Aircraft target detection method in remote sensing image based on the adaptive neural network

应用在卫星遥感图像的识别技术研究是当今世界上的热点方向,针对现在基于遥感图像进行飞机目标检测的精度低、通用性差,易受到地面房屋、天气等因素的干扰,文中提出应用自适应神经网络模型来进行基于遥感图像的飞机目标检测,采用大量的不同机型、不同姿态的飞机遥感图像对自适应神经网络模型进行训练,建立相对完善的神经网络模型。经试验证明,文中提出的基于自适应神经网络的遥感图像飞机目标检测方法精度较普通检测方法有显著提高,且通用性有所提高。因此经过训练后的神经网络模型不易受到房屋、天气等的因素干扰,文中提出的方法在军事和民用领域有较高的实用价值和工程实践价值。

The research on the recognition technology of satellite remote sensing image is a hotspot in the world.The accuracy of aircraft target detection based on remote sensing image is low,the versatility is not strong,and it is easy to be interfered by ground house,weather and other factors.The neural network model is applied to detect the aircraft target based on remote sensing image.The adaptive neural network model is trained by a large number of aircraft remote sensing images with different models and different attitudes to establish a relatively complete neural network model.The experimental results show that the accuracy of the aircraft target detection method based on adaptive neural network is higher than that of the common detection method,and the versatility is improved.The trained neural network model is not easily interfered by the housing,weather and other factors.The method proposed in this paper has strong practical value and engineering practice value in military and civilian fields.