一种新的遥感影像变化检测算法

A NOVEL REMOTE SENSING IMAGE CHANGE DETECTION ALGORITHM

针对传统的遥感影像变化检测算法不能同时确保准确率高、抗噪能力强和时间成本低的现状,设计一种新的基于非下采样剪切波变换和自适应脉冲耦合神经网络相结合的遥感影像变化检测算法。充分采用邻域信息有效降低虚警数量;借助非下采样剪切波变换的多方向、多尺度分解并利用全局、局部滤波,有效降低漏警数量;非下采样剪切波变换具有平移、旋转和尺度不变性,有效提升抗噪能力;自适应脉冲耦合神经网络分类准确率高并且时间成本很低。实验结果表明,与其他算法相比,该算法具有更高的检测准确度、更强的抗噪能力和更低的时间成本。实验结果验证了该算法的优越性和可行性。

Traditional remote sensing image change detection algorithm cannot simultaneously ensure high accuracy,strong anti-noise ability and low time cost.To solve this question,this paper designed a novel remote sensing image change detection algorithm based on nonsubsampled shearlet transform and adaptive pulse coupled neural network.Neighborhood information was applied to effectively reduce the number of false alarms,and we used multi-directional and multi-scale decomposition of nonsubsampled shearlet transform and global and local filtering to effectively reduce the number of missed alarms.The nonsubsampled shearlet transform had the invariance of translation,rotation and scale,which effectively improved the anti-noise ability.The classification accuracy of the adaptive pulse coupled neural network was high and the time cost was low.The experimental results show that compared with the other algorithm,the algorithm has higher detection accuracy for OE,PCC,KC,Recall and F1 performance indicators.It has stronger anti-noise ability and lower time cost.The experimental results fully prove the superiority and feasibility of the algorithm.