论文部分内容阅读
提出了一种基于主分量分析(PCA)和上下截集模糊Kohonen聚类网络(UDSFKCN)的、无监督的、不同时相的和卫星影像的像素级变化检测新算法。将PCA和UDSFKCN两种方法结合,并将它应用于不同时相的卫星影像的变化检测。该方法结合每个像素的邻域信息,利用PCA,产生每个像素对应的基于邻域信息的特征向量;又将变化区域检测问题转化为两类间的分类问题;然后利用UDSFKCN对每个像素所对应的特征向量进行变化类与未变化类的聚类,得到像素级的变化区域的检测图。实验结果表明,与传统方法相比,对于高斯和斑点噪声,本文算法具有更高的检测准确性和抗噪性能。
A new algorithm of pixel level change detection based on principal component analysis (PCA) and upper and lower intercept fuzzy Kohonen clustering network (UDSFKCN) is proposed for unsupervised, time-varying and satellite images. The PCA and UDSFKCN two methods combined, and it is applied to different phases of satellite image change detection. The method combines the neighborhood information of each pixel, and uses PCA to generate the eigenvector based on the neighborhood information of each pixel. In turn, the detection problem of the change region is transformed into the classification problem between two categories. Then, The corresponding eigenvectors are clustered between the changed and the un-changed classes to obtain the detection map of the pixel-level changing region. The experimental results show that compared with the traditional method, the proposed algorithm has higher detection accuracy and anti-noise performance for Gaussian and speckle noise.