基于LBP/C纹理的遥感影像居民地变化检测

Change Detection of Residential Area by Remote Sensing Image Based on LBP/C Texture

针对遥感影像上居民地纹理的特点,将基于方差的纹理分析方法、基于LAWS纹理能量测度的纹理分析方法、基于LBP/C(local binary pattern/contrast)的纹理分析方法分别对遥感影像上的居民地纹理特征进行分析描述;然后进行基于区域生长的扩张检测,得到居民地变化检测的结果。对不同地区、不同居民地分布情况的遥感影像进行了居民地扩张变化检测的试验。结果表明,基于LBP/C的变化检测方法可以高精度地检测出居民地的扩张变化,并提高了检测的自动化程度。

According to the texture characteristics of residential areas on the remote sensing images, the texture features of residential areas on the remote sensing images are analyzed and described by the methods of variance based, LAWS texture energy based and LBP/C （local binary pattern/contrast） based texture analysis. Then the extend detection based on region growing is performed, and the results of change detection of residential areas are acquired. Experiments are completed on three different remote sensing images with different regions and different distribution of residential areas. The experiment results indicate that the change detection method based on the LBP/C can detect the extend change of the residential areas with high accuracy, and improve detection automation.