论文部分内容阅读
为了实现遥感图像中对机场等感兴趣区域(ROI)准确的变化检测,提出了一种基于脉冲耦合神经网络(PCNN)的遥感图像变化检测方法。其基本原理是将变化前后的图像作为网络输入,经过多次迭代后用一维点火时间序列图表征输入图像信息,然后计算两幅时间序列图的相关系数确定是否有变化发生。最后对机场区域完全点火映射图进行异或运算,得到变化检测结果。实验结果表明,该方法对不同类型图像均有满意的检测结果。对比分析进一步证实了该方法具有较强的普适性和较高的检测精度。
In order to realize accurate change detection of ROI in airports and other remote sensing images, a method based on pulse coupled neural network (PCNN) is proposed to detect the change of remote sensing images. The basic principle is to input the image before and after the change as a network, and after a number of iterations, the input image information is characterized by a one-dimensional ignition time series graph, and then the correlation coefficient between the two time series graphs is calculated to determine if any change has taken place. Finally, the airport area complete ignition map XOR operation, change detection results. Experimental results show that this method has satisfactory results for different types of images. Comparative analysis further confirmed the method has strong universality and high detection accuracy.