摘要
无人机影像在小空间尺度上具有更好的探测地表物体能力。笔者针对无人机影像中车辆提取精度不高的问题,采用深度卷积神经网络(CNN)识别影像中车辆目标。首先生成大量的训练样本,构造层卷积神经网络并进行训练,然后根据训练模型对影像中待识别区域的车辆目标进行识别,并将识别结果与面向对象检测方法相比较。实验结果表明,该方法能够有效识别影像中的车辆目标,具有更高的识别率,优于其他方法。
Drone images have better ability to detect surface objects on a small spatial scale. In this paper, for the problem of low vehicle extraction accuracy in UAV images, deep convolutional neural network(CNN) is used to identify vehicle targets in the image. First, a large number of training samples are generated, a layer convolutional neural network is constructed and trained, and then the vehicle target in the image to be identified is identified according to the training model. Compare the recognition results with the object-oriented detection method. The experimental results show that this method can effectively identify the vehicle targets in the image and has a higher recognition rate than other methods.
引文
[1]李德仁,李清泉,杨必胜,等.3S技术与智能交通[J].武汉大学学报(信息科学版),2008(4):331-336.
[2]赵书玲,陈德海.基于交通环境因子的城市生态交通规划的理论框架[J].江西理工大学学报,2009,30(3):28-32.