Temporal scalable video coding with hierarchical reference picture selection method

Temporal scalable coding includes a function of extracting coded data with various frame rates from the original coded data. By using this function, the time resolution can be adjusted according to the processing power of the terminal and the transmission rate, so that best-effort video quality matched to the receiving environment is provided. In conventional temporal scalable coding, the frame in the same layer or the next lower layer closest in time to the frame to be encoded is selected As the reference picture. The problem then arises that the prediction efficiency is low. In the reference picture selection method, on the other hand, a reference picture memory for multiple frames is prepared, and a reference picture is selected for the encoding for each block. Thus, the Prediction efficiency is improved. However, unless the layer structure is introduced into the reference picture selection method, the problem of reference picture mismatch arises due to the decimation of the encoded data, which prevents the realization of a temporal scalable function. This paper proposes the hierarchical reference picture selection (HRPS) method, in which the reference picture selection method is applied while solving the mismatch problem. In HRPS, the reference picture memory is managed as a hierarchical structure and the reference picture selection method can be applied to each layer. This improves the coding efficiency of each layer. A procedure for assigning the reference picture memory capacity for each layer in HRPS is also proposed. (c) 2005 Wiley Periodicals, Inc.