A method for processing image data includes obtaining at least two captured images, where focal points of the obtained captured images are different; obtaining a focus-capture parameter, and fusing the captured images according to the focus-capture parameter to obtain a fused image; and performing joint coding on the fused image, and outputting an image; where when registration, mapping, cutting, and fusion are performed on the captured images according to the focus-capture parameter, a mask image is corrected through analysis of motion information of the captured images, and images are selected; and the corrected mask image includes a sequence number of a captured image selected at each pixel in the connected motion object region after the captured images are marked with sequence numbers. The method is used to implement high-efficiency and high-quality image processing.