Abstract: Addressing the scientific issue that the deformation and landslide evolution process of high and large slopes in open-pit mines is difficult to quantitatively describe, image data has the advantage of multidimensional information. Based on the spatiotemporal UAV remote sensing image data for slope monitoring, this study takes the slope spatial data contained in image pixels as the starting point to drive the identification of landslide disaster evolution information in remote sensing images. Coupling OpenCV technology, the method of analyzing and processing by comparing the image similarity in each period is adopted, proposing the concept of Cumulative Difference Degree-Time (CDD-T), and constructing a landslide evolution description method based on cumulative difference degree. This eliminates the influence of truncation errors in the spatial data processing of the existing slope monitoring system. Taking the deformation area of the external dump of the Hezigouluan open-pit mine in Ximeng area, China, as a typical engineering case, the spatial-temporal evolution law of the deformation area is quantitatively described. The research results show that by using the nⁿ×nⁿ fine uniform segmentation method, drawing the CDD-T evolution curve of the segmented area, it is analyzed that the slope top range of +985～+970 m and the southwest side +970～+940 m horizontal range are relatively dangerous areas. Combining the CDD-T heat map of each time point in the fine zoning, the spatial-temporal evolution law of the fine area landslide is quantitatively characterized, and the potential landslide mechanical mechanism is determined to be a push-type. During the landslide evolution process, the CDD-T curve of the spatiotemporal UAV remote sensing image and the change law of the corresponding area's cumulative displacement-time curve are basically consistent, and the maximum relative difference of the slope angles of the two curves is 0.046, with an average relative difference of 0.006. 83% of the GNSS monitoring points have a CDD-T slope greater than the displacement curve slope of the monitoring points, thus proving that the image cumulative difference degree can identify landslide risks earlier than the cumulative displacement. The developed remote sensing image cumulative difference degree calculation method can quantitatively describe the landslide evolution law, opening up new avenues for research on engineering geological disaster prevention and control.