An image compression method, performed by a computer, includes: acquiring image data of an image captured by an imaging device mounted to a vehicle; setting a segmented region in a reference image corresponding to an image to be compressed and encoded; enlarging the reference image so that a magnification rate of a region including the periphery of the reference image is greater than a magnification rate of a region including a center point of the reference image in the segmented region, the center point being a contact point of an optical axis of the image incident to the imaging device and an imaging element in the imaging device; generating a predictive image in accordance with the enlarged reference image; and compressing and encoding image data of the image to be encoded in accordance with the generated predictive image.