Pipelined parallelisation of automatic face inspection

A parallel pipeline is shown to be a natural method of speeding up a typical computer vision application, face inspection using `eigenfaces'. Faces within a stream of video images are continuously surveyed in a manner akin to a `conveyor belt' inspection process. The parallelisation is a new exemplar of a scheme for the rapid prototyping of large-scale, multi-algorithm applications suitable for transfer to a message-passing multicomputer. A general solution, pipelined processor farms, is preferred to a customised solution. This paper gives details of the software tools and software engineering methods employed to tackle this class of problem.     

Prediction and adaptive scanning in block-set wavelet image coder

Conditional probability to predict sub-block significance improves the peak signal-to-noise ratio (PSNR) performance of a block-set wavelet image coder. Adaptive scanning of sub-blocks results in a further gain in PSNR performance. Together, the related algorithmic changes introduced by the authors result in competitive PSNR performance without the need for an arithmetic coding stage. As a consequence, timing tests reveal that the critical decoder computational complexity is reduced.     

Generalised approach to parallelising image sequence codingalgorithms

The author describes the parallelisation of three different versions of the CCITT H.261 encoder algorithm using a generalised parallel design methodology based upon pipelines of processor farms (PPFs). For each algorithm, a theoretical upper-bound scaling model was derived by analysing the execution time profile of the algorithm and its feedback structure. The performance predicted by the model was, in each case, in good agreement with that achieved by the corresponding practical implementation. Practical throughput scaling up to a factor of 11 was achieved, using PPFs containing up to 16 processors. The design examples illustrate the impact which feedback has on potential speedup for image coding algorithms, and the diagnostic role of the model in identifying those algorithm components which restrict scaling performance. It is believed that the techniques presented may be useful both in developing embedded image coders based upon multiple DSP devices, and for simulation work with large image sequences in application areas such as image coding for HDTV and SHDTV     

User-configurable OCR enhancement for online natural history archives

The creation of structured digital libraries from paper-based archives is an area of growing demand in many scientific and cultural fields, and is not satisfied either by off-the-shelf OCR or commercial form- processing systems. This paper describes and evaluates a configurable archive construction system, which integrates document image pre-processing and analysis with text post-processing tools and a standard OCR package to meet digital archiving requirements. The prototype system is currently being used in conjunction with the UK Natural History Museum to help convert more than 500,000 cards of Lepidoptera (Butterflies and Moths) and Coleoptera (Beetles) to searchable digital archives. Evaluation results covering different aspects of the system from card scanning to overall word recognition rates for different database fields are summarised for two datasets comprising over 5,000 cards selected from different parts of these archives. First-pass end-to-end word recognition rates of 70–90% are reported for key data fields, subject to availability of suitable electronic dictionaries. Further validation and correction is supported through web-editing of the online digital archive.     

Using activity theory to understand intergenerational play: The case of Family Quest

We implemented a five-week family program called Family Quest where parents and children ages 9 to 13 played Quest Atlantis, a multiuser 3D educational computer game, at a local after-school club for 90-minute sessions. We used activity theory as a conceptual and an analytical framework to study the nature of intergenerational play, the collaborative activity between parents and children in the context of role-playing virtual game environment, and the opportunities and challenges of bringing parents and children together around an educational video game. Our analyses of five parent-child dyads revealed that the nature of intergenerational play is different for different parent-child dyads, but has positive outcomes. Implications of the study for supporting family learning and bonding through video games are discussed.     

Confidence testing optical-flow estimates

Localised estimates of the optical-flow field in a video-frame sequence are noise prone. By observing a similarity between first differential filtering used for two differing purposes, confidence testing of estimates is shown to be reliant on the curvature of an intensity surface. A successful confidence test moves towards hardware implementation     

Dose rate dependency of electronic personal dosemeters measuring X- and gamma-ray radiation

Three models of electronic personal dosemeters (EPDs)-Siemens Mk 2.3, Rados RAD-60S and Vertec Bleeper Sv-were irradiated with seven photon beam qualities: 60Co, 137Cs and the ISO narrow spectrum series X-ray qualities N-250, N-200, N-150, N-60 and N-20. The personal dose equivalent rates delivered to the devices varied between 0.002 and 0.25 mSv s(-1). Measurements were made with the EPDs mounted free-in-air as well as against Lucite and water phantoms. Results for all models of EPDs showed differences in personal dose equivalent energy response for different energies covered by this range of radiation qualities, with different models showing variations from 15 to 65%. In some cases, the personal dose equivalent rate response of these devices varied by a factor of 3 between irradiations at typical calibration dose rates and those normally encountered by nuclear energy workers.     

Parallel pipeline implementation of wavelet transforms

Wavelet transforms have been one of the important signal processing developments in the last decade, especially for applications such as time-frequency analysis, data compression, segmentation and vision. Although several efficient implementations of wavelet transforms have been derived, their computational burden is still considerable. The paper describes two generic parallel implementations of wavelet transforms, based on the pipeline processor farming methodology, which have the potential to achieve real-time performance. Results show that the parallel implementation of the oversampled wavelet transform achieves virtually linear speedup, while the parallel implementation of the discrete wavelet transform (DWT) also outperforms the sequential version, provided that the filter order is large. The DWT parallelisation performance improves with increasing data length and filter order, while the frequency-domain implementation performance is independent of wavelet filter order. Parallel pipeline implementations are currently suitable for processing multidimensional images with data length at least 512 pixels     

Performance metrics for embedded parallel pipelines

A statistical approach to performance prediction is applied to a system development methodology for pipelines comprised of independent parallel stages. The methodology is aimed at distributed memory machines employing medium-grained parallelization. The target applications are continuous-flow embedded systems. The use of order statistics on this type of system is compared to previous practical usage which appears largely confined to traditional Non-Uniform Memory Access (NUMA) machines for loop parallelization. A range of suitable performance metrics which give upper bounds or estimates for task durations are discussed. The metrics have a practical role when included in prediction equations in checking fidelity to an application performance specification. An empirical study applies the mathematical findings to the performance of a multicomputer for a synchronous pipeline stage. The results of a simulation are given for larger numbers of processors. In a further simulation, the results are extended to take account of waiting-time distributions while data are buffered between stages of an asynchronous pipeline. Order statistics are also employed to estimate the degradation due to an output ordering constraint. Practical illustrations in the image communication and vision application domains are included.