A method for codec-based recovery of a video, which includes recovering an image by combining image decoding information and non-overwritten image frames, is provided. The method includes: a cluster extraction step of extracting a pre-cluster region; a region extraction step of extracting, from the pre-cluster region extracted in the cluster extraction step, an M frame region encoded using an MPEG-4 visual codec and an M decoding region; a combining step of combining the M frame region and the M decoding region extracted in the region extraction step and listing the combination; and a recovering step of decoding the M frame region and the M decoding region, listed in the combining step, to be recovered as an image.