基于相位一致特征的CBERS-02B遥感图像自动配准

An Automatic Registration Method of CBERS-02B Remote Sensing Image Based on Phase Congruency Feature

以CBERS-02B卫星HR高分辨率和CCD多光谱遥感图像数据为基础,针对基于特征的自动配准方法中的特征检测与特征匹配两个关键步骤,通过引入性能优良的相位一致特征检测方法和特征相似与空间关系相结合的特征匹配策略,实现了一种基于相位一致特征的遥感图像高精度自动配准方法。实验结果表明,该方法对遥感图像亮度和对比度具有不变性,能稳定可靠提取HR高分辨率和CCD多光谱遥感图像显著的点特征,精确匹配相位一致特征点,实现了CBERS-02B卫星不同谱段,不同传感器和不同时相遥感图像间高精度自动配准,所进行实验的自动配准精度均到达了优于0.3像元的系统配准精度。因此,该自动配准方法适合应用于有高配准精度要求的遥感图像间自动配准。

This paper presents an automated image registration method based on the phase congruency feature to address the problems associated with the registration of CBERS-02B HR high resolution and CCD multi-spectral remote sensing image data. Phase congruency feature detector of high-performance and a matching strategy combining feature similarity and spatial relations are introduced into the proposed high accuracy remote sensing image automatic registration method as new feature detection and feature matching method.The experimental results indicate that the significant features can be detected and matched reliably with this method of being invariant to intensity and contrast variations,and a high registration accuracy of less than systematic mismatch error 0.3 pixels has been achieved in all the tests of CBERS-02B remote sensing images at different spectrum,different sensor,and/or different time.Therefore,the proposed automatic registration method is more adequate to the situation of high accuracy registration requirements.