| This dissertation addresses algorithmic optimizations for video encoding and augmented reality applications on mobile devices. First, a fast or computationally efficient mode prediction and selection approach is introduced which has the following attributes: (a) both the spatial and temporal information are used to achieve early termination using adaptive thresholds, (b) inclusion of a modulator allowing one to trade off computational efficiency and accuracy and (c) a homogenous region detection procedure for 8x8 blocks based on adaptive thresholds. Second, a fast adaptive termination algorithm for mode selection is introduced to increase its computation speed while attempting to maintain its coding efficiency for scalable video coding. The developed algorithm consists of the following three main steps which are applied not only to the enhancement layer but also to the base layer: a prediction step based on neighboring macroblocks, a first round check step, and a second round check step or refinement if failure occurs during the first round check. Third, a fast feature-assisted adaptive early termination approach is introduced in order to reduce the computational complexity while maintaining more or less the same video quality. This feature-assisted approach consists of three parts: (1) reduction of the number of available reference frames using predicted motion activity, extracted texture information, and skip mode from neighboring macroblocks, (2) a reference frame prediction based on neighboring macroblocks, and (3) an adaptive early termination threshold derived from a theoretical analysis based on all zero block detection. Extensive experimental results are performed to demonstrate the computational gain of the introduced approaches over the standard approaches. Finally, a head pose estimation algorithm is developed for real-time operation on mobile devices. The main attributes of the developed head pose estimation algorithm include: (1) Viola-Jones face detection, (2) color tracking based on an online calibration procedure, and (3) a robust and computationally efficient head pose estimation algorithm. |
