为快速准确识别机场遥感图像飞机目标, 提出了一种深度卷积神经网络与边缘轮廓特征提取技术结合的识别算法。利用深度卷积神经网络对机场遥感图像中飞机目标进行深度特征提取, 针对飞机停机位置存在阴影的问题, 结合优化后的Canny算子得到目标轮廓, 经由支持向量机给飞机分类。算法主要有两个阶段。第一阶段为训练阶段, 主要对深度卷积神经网络进行训练, 将获得的特征归一化；利用Canny算子得到边缘特征, 通过主成分分析法得到飞机主轴, 求解主轴两侧边缘点欧氏距离作为特征向量；接着完成支持向量机分类器训练。第二阶段为测试阶段, 主要对算法进行验证并测试准确性。实验结果证明, 算法识别的正确率高达94. 39%, 能够较好地识别飞机目标。

To recognize aircraft targets in airport remote sensing images quickly and accurately, a recognition algorithm combining the deep convolutional neural network with the edge contour feature extraction technique is proposed. The depth features of the aircrafts in the airport remote sensing image are extracted by using the deep convolutional neural network. To solve the shadow problem in aircraft parking positions, the target contour is obtained by using the optimized Canny operator, and then the aircrafts are classified by using Support Vector Machine (SVM). The the algorithm consists of the following two stages. The first stage is the training phase, which mainly trains the deep convolutional neural network and normalizes the obtained features. Then the edge features are obtained by using Canny operator and the major axis is obtained by using Principal Component Analysis (PCA) method. The Euclidean distance between the edge points along the two sides of the spindle is extracted as the eigenvector, and finally SVM classifier training is implemented. The second stage is the testing phase, in which the algorithm is verified and its accuracy is tested. Experimental results show that the recognition rate of the method can reach 94. 39%, which can effectively recognize the aircraft targets.