Rate-distortion modelling of scalable video sequences with different complexities

Rate-Distortion (R-D) models are used to estimate the minimum bit rate required for real time Scalable Video Coding (SVC) streams. This paper proposes R-D models for different types of video sequences complexity wise that can be used generally for wired and wireless video transmissions. Models of Base Layer (BL) rate and Enhancement Layer (EL) rate are provided in parallel with R-D models for different complex video sequences to achieve more accurate rate predictions. Models are developed with the function of Spatial Index (SI) and Temporal Index (TI) of each Group of Pictures (GOP) in video sequences. An algorithm is developed to estimate the rates required by each GOP in video sequences with different complexities if the distortion is given and vice versa. In our experimental results and comparisons, the graphs of different complex videos show that the calculated rates are estimated closer to the actual rates than the calculated rates of the existing R-D models.