Signatures from MPEG or Motion JPEG encoded video clips are extracted based on the luminance, chrominance, and motion vector values. The signatures are stored in a database, along with corresponding location, size, and time length information. The signature of a query video clip, also encoded in the MPEG or Motion JPEG format, is extracted. The signature of the query video clip is compared with the signatures stored in the meta database, with video clips having signatures similar to the signature of the query video clip identified. The video clips are then retrieved and displayed by selection of a user.