A geotagged video spatial indexing method for video retrieval based on a two-dimensional (2D) temporal grid is disclosed and includes:



generating the 2D temporal grid by using the earliest start time, the latest end time of all geotagged video clips and a temporal resolution; calculating row and column number information of each geotagged video clip based on its start time and end time; generating a spatial point for each geotagged video clip based on the row and column number information and obtaining a spatial point set; generating a R-tree spatial index structure corresponding to the spatial point set using a R-tree spatial index method; and locating the corresponding cell in a temporal grid based on retrieval conditions, generating a spatial point based on row and column number information of the grid cell, and finding the spatial point in the R-tree spatial index structure to get the geotagged video corresponding thereto.