基于对偶树复小波变换与PCA方法结合的图像变化检测算法研究

Image change detection based on dual-tree complex wavelet transform and principal component analysis

图像变化检测是遥感图像处理领域重要方向,大多数变化检测算法都存在算法复杂度高、抗噪性弱等缺陷,利用对偶树复小波变换的平移不变性与能提高方向分辨率的优点,把对偶树复小波变换运用于变化检测中,可以提高图像细节变化的检测和算法抗噪性。首先用对偶树复小波变换对图像进行尺度分解,把图像在每个尺度上分解成一个低通子图和六个方向的高通子图。然后运用PCA(主向量分析法)提取每个尺度与方向上的特征并降维,然后运用k均值算法将图像像素分成为变化与不变化两类,最后通过多尺度融合,得到变化检测图像。

Image change detection is a very important part of remote sensing image processing.Many algo rithms have defects,such as highly complex or weakly antinosie.Since the dual-tree complex wavelet trans form （DT-CWT） is shift invariant and has improved directional resolution,the DT-CWT is introduced in image change detection in order to provide accurate detection of small changes and attractive robustness against noise.Firstly,the DT-CWT is used to decompose the image into a low-pass subband and six directional high-pass sub bands at each scale.Secondly,principal component analysis （PCA） is used to create eigenvector and k-means is used to categorize pixels into two parts （change and unchanged）.Finally,both the intrascale fusion and the interscale fusion are used to detect the changed images.