Dynamic search range control algorithm for inter-frame coding in scalable video coding

Scalable video coding (SVC) is the ongoing standard as an extension of H.264/AVC to provide stable multimedia services to various service environments and end systems. An exhaustive block motion estimation has been used to determine the best coding mode for each block type in this standard. This achieves a high coding efficiency; however it causes very high computational complexity in the encoding system. To reduce this complexity, we propose a novel algorithm for inter-frame coding based on a dynamic search range control (DSRC) mechanism to provide coarse grain signal-to-noise ratio (CGS) and spatial scalability. An efficient learning method is applied to control the search ranges of the motion estimation process according to block modes from the base layer where the full search is used. We use an adaptive learning scheme based on the steepest decent adjustment with mode information from the base layer. This scheme is also extended to the training search range in the base layer. We verify that the overall encoding time can be reduced approximately 84% based on a comparative analysis of experimental results using JSVM 9.12 reference software.