During the last years, the amount of multimedia content has greatlyincreased. This has multiplied the need of efficient compression ofthe content but also the ability to search, retrieve, browse, or filterit. Generally, video compression and indexing have been investigatedseparately. However, as the amount of multimedia contentgrows, it will be very interesting to study representations that, at thesame time, provide good compression and indexing functionalities.Moreover, even if the indexing metadata is created for functionalitiessuch as search, retrieval, browsing, etc., it can also be employedto increase the efficiency of current video codecs. Here, we use itto improve the long term prediction step of the H.264/AVC videocodec. This paper focuses on the comparison between four differentMPEG-7 descriptors when used in the proposed scheme.