An unmanned aerial vehicle image processing system includes a digital camera mounted on an unmanned aerial vehicle, an autopilot controller installed in the unmanned aerial vehicle and pre-loaded with a base image, and a computer device. The autopilot controller programs a flight path as a series of mission lags to cover a target polygon on the base image in a lawnmower pattern for controlling the unmanned aerial vehicle to perform a flight mission in which the autopilot controller records plural GPS/INS data and triggers the digital camera to take plural photos which are further calibrated to generate plural calibrated photos corresponding to the plural GPS/INS data, respectively. The computer device processes the plural calibrated photos to generate plural projected photos by using the corresponding plural GPS/INS data, respectively, and stitches the plural projected photos to a mosaic image by a photo-stitching process.