The invention relates to the field of remote-sensing image processing techniques. Provided is a sea-land segmentation method for a large-size remote-sensing image, comprising: step A. performing a filling process on sea and land regions of a remote-sensing image to obtain resulting images of a rough sea-land segmentation; step B. combining the resulting images of the rough sea-land segmentation to perform extraction on segments of the remote-sensing image to obtain resulting images of a precise sea-land segmentation; and step C. fusing the resulting images of the rough sea-land segmentation and the resulting images of the precise sea-land segmentation to obtain a resulting image of the sea-land segmentation. The invention realizes an automatic sea-land segmentation method for a large-size remote-sensing image, thereby solving problems in which existing sea-land segmentation techniques have low sea-land segmentation precision and an almost ineffective segmentation effect when it comes to images having complica
