Image splicing method

The disclosure provides an image splicing method, including: dividing a to-be-predicted image into first and second cropped images, wherein an overlap area exists between the first and second cropped images; and dividing the overlap area into first and second sub-overlap regions; finding a first image region that does not include the second sub-overlap region in the first cropped image, and finding a second image region that does not include the first sub-overlap region in the second cropped image; generating first and second prediction result images corresponding to the first and second cropped images respectively; finding a first prediction image region corresponding to the first image region in the first prediction result image, and finding a second predicted image region corresponding to the second image region in the second prediction result image; and splicing the first and second prediction image regions as a spliced image.