MLB sports frames retrieval using color cipher similarities

We present a realistic and fully automatic content-based approach for retrieval MLB sports frames efficiently. Specifically, we present a rule-based method for detecting primary and recurrent scenes in MLB videos (e.g., MLB pitching scenes in baseball). The proposed approach transfers each MLB sports frame to a color cipher using only straightforward 8 rules. Subsequently, we utilize the color ciphers to compare the MLB sports frames, namely color ciphers comparison. We succeed in transferring the MLB sports frames retrieval problem to color ciphers comparison, so the computational complexity is decreased obviously. Our system keeps both advantages of the content based videos retrieval system (similarity-based retrieval) and a text based videos retrieval system (very rapid and mature).