No description available.

Software implementation of block-based video coding standards has been used in a wide range of applications. In many cases, such as real-time multimedia systems or power-constrained systems, the coding performance of software-only video encoders and decoders is limited by computational complexity. This thesis presents research work to develop techniques to manage computational complexity of video encoders. These techniques aim to provide significant complexity saving as well as adaptively controlling the computational complexity. This thesis first investigates experimentally the most computationally intensive functions in a video encoder. Based on the results of profile tests, several functions are selected as candidates, on which complexity reduction algorithms will be performed. These functions include discrete cosine transform and related functions as well as motion estimation. Adaptive complexity-reduction algorithms are proposed for computationally expensive functions: discrete cosine transform and motion estimation functions respectively. It is shown that these algorithms can flexibly control the computational complexity of each function with negligible loss of video quality. The inherent characteristics of coded macroblocks are investigated through experimental tests and they are categorized into "skipped" and" unskipped" macroblocks based on two parameters. An innovative algorithm is developed to reduce the computational complexity by predicting "skipped" macroblock prior to encoding and not carrying out the coding process on these macroblocks. The approaches described in this thesis can not only achieve adaptive control of the computational complexity of a video encoder, but also can manage the trade-off between complexity and distortion. These proposed algorithms are evaluated in terms of complexity reduction performance, rate-distortion performance and subjective and objective visual quality by experimental testing.The advantages and disadvantages of each algorithm are discussed.

No description available.

No description available.

No description available.

Lossy compression is widely applied for coding visual information in applications such as entertainment in order to achieve a high compression ratio. In this case, the video quality worsens as the compression ratio increases. Rate control tries to use the bit budget properly so the visual distortion is minimized. Rate control for H.264, the state-of-the-art hybrid video coder, is investigated. Based on the Rate-Distortion (R-D) slope analysis, an operational rate distortion optimization scheme for H.264 using Lagrangian multiplier method is proposed. The scheme tries to find the best path of quantization parameter (OP) options at each macroblock. The proposed scheme provides a smoother rate control that is able to cover a wider range of bit rates and for many sequences it outperforms the H.264 (JM92 version) rate control scheme in the sense of PSNR. The Bath University Matching Pursuit (BUMP) project develops a new matching pursuit (MP) technique as an alternative to transform video coders. By combining MP with precision limited quantization (PLO) and multi-pass embedded residual group encoder (MERGE), a very efficient coder is built that is able to produce an embedded bit stream, which is highly desirable for rate control. The problem of optimal bit allocation with a BUMP based video coder is investigated. An ad hoc scheme of simply limiting the maximum atom number shows an obvious performance improvement, which indicates a potential of efficiency improvement. An in depth study on the bit Rate-Atom character has been carried out and a rate estimation model has been proposed. The model gives a theoretical description of how the oit number changes. An adaptive rate estimation algorithm has been proposed. Experiments show that the algorithm provides extremely high estimation accuracy. The proposed R-D source model is then applied to bit allocation in the BUMP based video coder. An R-D slope unifying scheme was applied to optimize the performance of the coder'. It adopts the R-D model and fits well within the BUMP coder. The optimization can be performed in a straightforward way. Experiments show that the proposed method greatly improved performance of BUMP video coder, and outperforms H.264 in low and medium bit rates by up to 2 dB.

No description available.

The aim of this thesis is to accelerate the process of motion estimation (ME) for the implementation of real time, portable video encoding. To this end a number of different techniques have been considered and these have been investigated in detail. Data Level Parallelism (DLP) is exploited first, through the use of vector instruction extensions using configurable/re-configurable processors to form a fast System-On-Chip (SoC) video encoder capable of embedding both full search and fast ME methods. Further parallelism is then exploited in the form of Thread Level Parallelism (TLP), introduced into the ME process through the use of multiple processors incorporated onto a single Soc. A theoretical explanation of the results, obtained with these methodologies, is then developed for algorithmic optimisations. This is followed with the investigation of an efficient, orthogonal technique based on the use of a reduced number of bits (RBSAD) for the purposes of image comparison. This technique, which provides savings of both power and time, is investigated along with a number of criteria for its improvement to full resolution. Finally a VLSI layout of a low-power ME engine, capable of using this technique, is presented. The combination of DLP, TLP and RBSAD is found to reduce the clock frequency requirement by around an order of magnitude.

It is well known that stereoscopic images and video can be used to simulate the natural process of stereopsis within the Human Visual System (HVS), by providing the two stereo images/video-streams separately to the two eyes. However as compared to presenting traditional two-dimensional images/video to the HVS, providing stereoscopic information requires double the resources, in the form of transmission bandwidth and/or storage space. Thus to handle this excess data effectively, data compression techniques are required, which is the main focus of the research presented in this thesis. The thesis proposes two novel stereoscopic video CODECs, based on the latest video coding standard, H.264.

In this thesis, we introduce Irregular Variable Length Coding (IrVLC) and investigate its applications, characteristics and performance in the context of digital multimedia broadcast telecommunications. During IrVLC encoding, the multimedia signal is represented using a sequence of concatenated binary codewords. These are selected from a codebook, comprising a number of codewords, which, in turn, comprise various numbers of bits. However, during IrVLC encoding, the multimedia signal is decomposed into particular fractions, each of which is represented using a different codebook. This is in contrast to regular Variable Length Coding (VLC), in which the entire multimedia signal is encoded using the same codebook. The application of IrVLCs to joint source and channel coding is investigated in the context of a video transmission scheme. Our novel video codec represents the video signal using tessellations of Variable-Dimension Vector Quantisation (VDVQ) tiles. These are selected from a codebook, comprising a number of tiles having various dimensions. The selected tessellation of VDVQ tiles is signalled using a corresponding sequence of concatenated codewords from a Variable Length Error Correction (VLEC) codebook. This VLEC codebook represents a specific joint source and channel coding case of VLCs, which facilitates both compression and error correction. However, during video encoding, only particular combinations of the VDVQ tiles will perfectly tessellate, owing to their various dimensions. As a result, only particular sub-sets of the VDVQ codebook and, hence, of the VLEC codebook may be employed to convey particular fractions of the video signal. Therefore, our novel video codec can be said to employ IrVLCs. The employment of IrVLCs to facilitate Unequal Error Protection (UEP) is also demonstrated. This may be applied when various fractions of the source signal have different error sensitivities, as is typical in audio, speech, image and video signals, for example. Here, different VLEC codebooks having appropriately selected error correction capabilities may be employed to encode the particular fractions of the source signal. This approach may be expected to yield a higher reconstruction quality than equal protection in cases where the various fractions of the source signal have different error sensitivities. Finally, this thesis investigates the application of IrVLCs to near-capacity operation using EXtrinsic Information Transfer (EXIT) chart analysis. Here, a number of component VLEC codebooks having different inverted EXIT functions are employed to encode particular fractions of the source symbol frame. We show that the composite inverted IrVLC EXIT function may be obtained as a weighted average of the inverted component VLC EXIT functions. Additionally, EXIT chart matching is employed to shape the inverted IrVLC EXIT function to match the EXIT function of a serially concatenated inner channel code, creating a narrow but still open EXIT chart tunnel. In this way, iterative decoding convergence to an infinitesimally low probability of error is facilitated at near-capacity channel SNRs.