中文摘要：

利用ALOS多光谱影像及CV模型对山东沿海三种不同类型海岸、人工海岸、砂质海岸、淤泥质海岸进行海岸线提取，为比较不同特征影像对海岸线提取效果的影响，对同一类型海岸又分为近红外影像、421彩色合成影像、NDWI影像、PCA影像并分别进行海岸线提取，最后利用缓冲区分析进行精度比较。分析得出：对于人工海岸、砂质海岸两种海岸，在一个像元缓冲区内，提取精度基本可以达到80％以上，两个像元内可达到95％以上；对于淤泥质海岸，四种特征影像的提取效果均比较差，只有NDWI影像在两个缓冲区内可达到87％；总体来说， ND-WI影像提取的效果最好。

英文摘要：

This paper uses ALOS multi -spectral images and CV models to extract coastlines on three different types coast of Shan -dong, which are artificial coast , sandy coast and muddy coast .To compare the extraction effect on different coast images , the same type coast image was divided into near -infrared image, 421 color composite image, NDWI image and PCA image, and finally t buffer analysis was used for comparing the accuracy .As analyzed for artificial coast and sandy coast in a pixel buffer , more than 80%extrac-tion accuracy can be achieved and within two pixels the result can reach to more than 95%;for muddy coast , four kinds of feature im-ages all work badly , only NDWI image can reach to 87%within two buffers;overall, NDWI image works best .