The research on real-time automatic identification technology for landslides is of great significance for protecting people’s lives, property, and ecological safety. It can solve the problem of poor timeliness in landslide risk investigation and prevention due to the lack of timely identification of landslides at present. Considering the changes of vegetation coverage index (NDVI) as one of the important criteria for landslide detection, the article combines NDVI change detection technology, automatic threshold selection algorithm, and morphological technology to achieve real-time and automatic recognition of landslides. Compared with existing research algorithms, Compared with existing research algorithms, it adds some important parameters in the automatic landslide recognition process (such as NDVI, mountain shadows, etc.). The adaptive automatic threshold selection algorithm reduces manual involvement and significantly enhances its timeliness while ensuring high recognition accuracy. This article is based on two optical images and takes a certain area in Mentougou District, Beijing as the research area. Real time and automatic recognition of landslides in this area from September 7, 2021 to September 7, 2022 is carried out, using the results of manual visual interpretation as the correct standard. The recognition results of the article are compared with their accuracy, and the landslide detection rate reached 92.31%, proving the accuracy and high efficiency of this method for detecting landslides. Finally, the method is applied to the central part of Dujiangyan City, which further proves the effectiveness and generalization ability of the method.