| The introduction of powerful compact wireless devices (such as handheld computers) raises new and unexplored aspects to optimizing complexity-distortion for video. Historically, the computational power of powerful platforms (e.g., computers) has exceeded the available communication bandwidth over emerging networks such as the Internet and wireless networks. Hence, improved performance was achieved by reducing the required transmission bitrate of video (through higher compression) and by increasing the computational complexity. Handheld devices do not follow the traditional balance of bandwidth and power; computational power is at a premium while bandwidth is comparatively more available. This thesis investigates the complexity-distortion options for a state-of-the-art video compression standard, H.264/JVT, which is an estimated three times more complex than the previous standards. The research focuses on analyzing the video standard's complexity for one of the most popular handheld devices, the iPAQ. Operational complexity-distortion and rate-complexity curves are presented for a number of Group of Picture (GoP) structures, entropy coding modes and compared against Pentium IV performance. The experiments include some of the earliest comparisons between JVT's entropy coding modes, context-adaptive binary arithmetic coding and content-adaptive variable-length coding. |
