基于BP神经网络模型的遥感图像道路分割处理方法研究

图像分割是军事目标识别的主要处理方法.由于神经网络对于解决目标识别问题具有适合用于高速并行处理系统、可以实现特征空间较为复杂的划分等优势,采用神经网络模型在复杂的遥感图像背景中,识别出宽度很小的道路是较为理想的.在简单介绍BP神经网络模型的基础上,论述了BP神经网络模型在目标识别中关于道路分割问题的处理方法,并以实例证明了采用BP神经网络模型对遥感图像进行分割,得到的结果图像能够从复杂的背景图像中分割出道路,并能清楚地反映道路的方向和分叉,对于遥感图像的目标识别有重要的实用价值.

The study of road segmentation methods of remote image based on BP NN model

Image segmentation is the main method to identify a military target. In regard to identifying the target object,the neural network has many advantages, for example, it is capable of implementing more complex partitioning of feature space and it is amenable to high-performance parallel-processing implementations. For the advantages given above, it is ideal to apply the neural network model to identifying a relatively narrow road in a very complex background of remote images. The paper briefly introduces the BP NN model, and discusses how to deal with road segmentation by using the BP NN model to identify the target. In conclusion, through examples,the BP NN model can identify a road from a complicated background image and can clearly reflect the direction and bifurcation of the road, which proves it is pragmatically valuable in identifying the remote-sensing image.