3D Hand Pose Estimation by Finding Appearance-Based Matches in a Large Database of Training Views

Ongoing work towards appearance-based 3D hand pose estimation from a single image is presented. A large database of synthetic hand views is generated using a 3D hand model and computer graphics. The views display different hand shapes as seen from arbitrary viewpoints. Each synthetic view is automatically labeled with parameters describing its hand shape and viewing parameters. Given an input image, the system retrieves the most similar database views, and uses the shape and viewing parameters of those views as candidate estimates for the parameters of the input image. Preliminary results are presented, in which appearance-based similarity is defined in terms of the chamfer distance between edge images.

Database Indexing Methods for 3D Hand Pose Estimation

Estimation of 3D hand pose is useful in many gesture recognition applications, ranging from human-computer interaction to automated recognition of sign languages. In this paper, 3D hand pose estimation is treated as a database indexing problem. Given an input image of a hand, the most similar images in a large database of hand images are retrieved. The hand pose parameters of the retrieved images are used as estimates for the hand pose in the input image. Lipschitz embeddings of edge images into a Euclidean space are used to improve the efficiency of database retrieval. In order to achieve interactive retrieval times, similarity queries are initially performed in this Euclidean space. The paper describes ongoing work that focuses on how to best choose reference images, in order to improve retrieval accuracy.

FPGA-based minutia matching for biometric fingerprint image database retrieval

In this paper, a parallel-matching processor architecture with early jump-out (EJO) control is proposed to carry out high-speed biometric fingerprint database retrieval. The processor performs the fingerprint retrieval by using minutia point matching. An EJO method is applied to the proposed architecture to speed up the large database retrieval. The processor is implemented on a Xilinx Virtex-E, and occupies 6,825 slices and runs at up to 65 MHz. The software/hardware co-simulation benchmark with a database of 10,000 fingerprints verifies that the matching speed can achieve the rate of up to 1.22 million fingerprints per second. EJO results in about a 22% gain in computing efficiency.

Characterisation of the frequent exacerbator phenotype in COPD patients in a large UK primary care population

BACKGROUND: The 'frequent exacerbator' is recognised as an important phenotype in COPD. Current understanding about this phenotype comes from prospective longitudinal clinical trials in secondary/tertiary care with little information reported in primary care populations.

AIMS: To characterize the frequent-exacerbator phenotype and identify associated risk factors in a large UK primary care COPD population.

METHODS: Using a large database of primary care patients from 80 UK general practices, patients were categorised using GOLD 2014 criteria into high and low risk groups based on exacerbation history. A multivariate logistic regression model was used to investigate covariates associated with the frequent-exacerbator phenotype and risk of experiencing a severe exacerbation (leading to hospitalisation).

RESULTS: Of the total study population (n = 9219), 2612 (28%) fulfilled the criteria for high risk frequent-exacerbators. Independent risk factors (adjusted odds ratio [95% CI]) for ≥2 exacerbations were: most severely impaired modified Medical Research Council (mMRC) dyspnoea score (mMRC grade 4: 4.37 [2.64-7.23]), lower FEV1 percent predicted (FEV1 <30%: 2.42 [1.61-3.65]), co-morbid cardiovascular disease (1.42 [1.19-1.68]), depression (1.56 [1.22-1.99]) or osteoporosis (1.54 [1.19-2.01]), and female gender (1.20 [1.01-1.43]). Older patients (≥75 years), those with most severe lung impairment (FEV1 <30%), those with highest mMRC score and those with co-morbid osteoporosis were identified as most at risk of experiencing exacerbations requiring hospitalisation.

CONCLUSIONS: Although COPD exacerbations occur across all grades of disease severity, female patients with high dyspnoea scores, more severely impaired lung function and co-morbidities are at greatest risk. Elderly patients, with severely impaired lung function, high mMRC scores and osteoporosis are associated with experience of severe exacerbations requiring hospitalisation.

Multi-relational algorithm for mining association rules in large databases

Multi-relational data mining enables pattern mining from multiple tables. The existing multi-relational mining association rules algorithms are not able to process large volumes of data, because the amount of memory required exceeds the amount available. The proposed algorithm MRRadix presents a framework that promotes the optimization of memory usage. It also uses the concept of partitioning to handle large volumes of data. The original contribution of this proposal is enable a superior performance when compared to other related algorithms and moreover successfully concludes the task of mining association rules in large databases, bypass the problem of available memory. One of the tests showed that the MR-Radix presents fourteen times less memory usage than the GFP-growth. © 2011 IEEE.

Uric acid biorhythm, a feature of long-term variation in a clinical laboratory database

Background: The biorhythm of serum uric acid was evaluated in a large sample of a clinical laboratory database by spectral analysis and the influence of the gender and age on uric acid variability. Methods: Serum uric acid values were extracted from a large database of a clinical laboratory from May 2000 to August 2006. Outlier values were excluded from the analysis and the remaining data (n = 73,925) were grouped by gender and age ranges. Rhythm components were obtained by the Lomb Scargle method and Cosinor analysis. Results: Serum uric acid was higher in men than in women older than 13 years (p<0.05). Compared with 0-12 year group, uric acid increased in men but not in women older than 13 years (p<0.05). Circannual (12 months) and transyear (17 months) rhythm components were detected, but they were significant only in adult individuals (>26 years, p<0.05). Cosinor analysis showed that midline estimating statistic of rhythm (MESOR) values were higher in men (range: 353-368 mu mol/L) than in women (range: 240-278 mu mol/L; p<0.05), independent of the age and rhythm component. The extent of predictable change within a cycle, approximated by the double amplitude, represented up to 20% of the corresponding MESOR. Conclusions: Serum uric acid biorhythm is dependent on gender and age and it may have relevant influence on preanalytical variability of clinical laboratory results.

A forensic investigation of single human hair fibres using FTIR-ATR spectroscopy and chemometrics

Human hair fibres are ubiquitous in nature and are found frequently at crime scenes often as a result of exchange between the perpetrator, victim and/or the surroundings according to Locard's Principle. Therefore, hair fibre evidence can provide important information for crime investigation. For human hair evidence, the current forensic methods of analysis rely on comparisons of either hair morphology by microscopic examination or nuclear and mitochondrial DNA analyses. Unfortunately in some instances the utilisation of microscopy and DNA analyses are difficult and often not feasible. This dissertation is arguably the first comprehensive investigation aimed to compare, classify and identify the single human scalp hair fibres with the aid of FTIR-ATR spectroscopy in a forensic context. Spectra were collected from the hair of 66 subjects of Asian, Caucasian and African (i.e. African-type). The fibres ranged from untreated to variously mildly and heavily cosmetically treated hairs. The collected spectra reflected the physical and chemical nature of a hair from the near-surface particularly, the cuticle layer. In total, 550 spectra were acquired and processed to construct a relatively large database. To assist with the interpretation of the complex spectra from various types of human hair, Derivative Spectroscopy and Chemometric methods such as Principal Component Analysis (PCA), Fuzzy Clustering (FC) and Multi-Criteria Decision Making (MCDM) program; Preference Ranking Organisation Method for Enrichment Evaluation (PROMETHEE) and Geometrical Analysis for Interactive Aid (GAIA); were utilised. FTIR-ATR spectroscopy had two important advantages over to previous methods: (i) sample throughput and spectral collection were significantly improved (no physical flattening or microscope manipulations), and (ii) given the recent advances in FTIR-ATR instrument portability, there is real potential to transfer this work.s findings seamlessly to on-field applications. The "raw" spectra, spectral subtractions and second derivative spectra were compared to demonstrate the subtle differences in human hair. SEM images were used as corroborative evidence to demonstrate the surface topography of hair. It indicated that the condition of the cuticle surface could be of three types: untreated, mildly treated and treated hair. Extensive studies of potential spectral band regions responsible for matching and discrimination of various types of hair samples suggested the 1690-1500 cm^-1 IR spectral region was to be preferred in comparison with the commonly used 1750-800 cm^-1. The principal reason was the presence of the highly variable spectral profiles of cystine oxidation products (1200-1000 cm^-1), which contributed significantly to spectral scatter and hence, poor hair sample matching. In the preferred 1690-1500 cm^-1 region, conformational changes in the keratin protein attributed to the α-helical to β-sheet transitions in the Amide I and Amide II vibrations and played a significant role in matching and discrimination of the spectra and hence, the hair fibre samples. For gender comparison, the Amide II band is significant for differentiation. The results illustrated that the male hair spectra exhibit a more intense β-sheet vibration in the Amide II band at approximately 1511 cm^-1 whilst the female hair spectra displayed more intense α-helical vibration at 1520-1515cm^-1. In terms of chemical composition, female hair spectra exhibit greater intensity of the amino acid tryptophan (1554 cm^-1), aspartic and glutamic acid (1577 cm^-1). It was also observed that for the separation of samples based on racial differences, untreated Caucasian hair was discriminated from Asian hair as a result of having higher levels of the amino acid cystine and cysteic acid. However, when mildly or chemically treated, Asian and Caucasian hair fibres are similar, whereas African-type hair fibres are different. In terms of the investigation's novel contribution to the field of forensic science, it has allowed for the development of a novel, multifaceted, methodical protocol where previously none had existed. The protocol is a systematic method to rapidly investigate unknown or questioned single human hair FTIR-ATR spectra from different genders and racial origin, including fibres of different cosmetic treatments. Unknown or questioned spectra are first separated on the basis of chemical treatment i.e. untreated, mildly treated or chemically treated, genders, and racial origin i.e. Asian, Caucasian and African-type. The methodology has the potential to complement the current forensic analysis methods of fibre evidence (i.e. Microscopy and DNA), providing information on the morphological, genetic and structural levels.

'It ain't over till it's over' : Yogi Berra bias on prediction markets

The recent expansion of prediction markets provides a great opportunity to test the market efficiency hypothesis and the calibration of trader judgements. Using a large database of observed prices, this article studies the calibration of prediction markets prices on sporting events using both nonparametric and parametric methods. While only minor bias can be observed during most of the lifetime of the contracts, the calibration of prices deteriorates very significantly in the last moments of the contracts’ lives. Traders tend to overestimate the probability of the losing team to reverse the situation in the last minutes of the game.

Estimating health related costs and savings from balcony acoustic design for road traffic noise

A multi-faceted study is conducted with the objective of estimating the potential fiscal savings in annoyance and sleep disturbance related health costs due to providing improved building acoustic design standards. This study uses balcony acoustic treatments in response to road traffic noise as an example. The study area is the State of Queensland in Australia, where regional road traffic noise mapping data is used in conjunction with standard dose–response curves to estimate the population exposure levels. The background and the importance of using the selected road traffic noise indicators are discussed. In order to achieve the objective, correlations between the mapping indicator (LA10 (18 hour)) and the dose response curve indicators (Lden and Lnight) are established via analysis on a large database of road traffic noise measurement data. The existing noise exposure of the study area is used to estimate the fiscal reductions in health related costs through the application of simple estimations of costs per person per year per degree of annoyance or sleep disturbance. The results demonstrate that balcony acoustic treatments may provide a significant benefit towards reducing the health related costs of road traffic noise in a community.

Robust image hash functions using higher order spectra

Robust hashing is an emerging field that can be used to hash certain data types in applications unsuitable for traditional cryptographic hashing methods. Traditional hashing functions have been used extensively for data/message integrity, data/message authentication, efficient file identification and password verification. These applications are possible because the hashing process is compressive, allowing for efficient comparisons in the hash domain but non-invertible meaning hashes can be used without revealing the original data. These techniques were developed with deterministic (non-changing) inputs such as files and passwords. For such data types a 1-bit or one character change can be significant, as a result the hashing process is sensitive to any change in the input. Unfortunately, there are certain applications where input data are not perfectly deterministic and minor changes cannot be avoided. Digital images and biometric features are two types of data where such changes exist but do not alter the meaning or appearance of the input. For such data types cryptographic hash functions cannot be usefully applied. In light of this, robust hashing has been developed as an alternative to cryptographic hashing and is designed to be robust to minor changes in the input. Although similar in name, robust hashing is fundamentally different from cryptographic hashing. Current robust hashing techniques are not based on cryptographic methods, but instead on pattern recognition techniques. Modern robust hashing algorithms consist of feature extraction followed by a randomization stage that introduces non-invertibility and compression, followed by quantization and binary encoding to produce a binary hash output. In order to preserve robustness of the extracted features, most randomization methods are linear and this is detrimental to the security aspects required of hash functions. Furthermore, the quantization and encoding stages used to binarize real-valued features requires the learning of appropriate quantization thresholds. How these thresholds are learnt has an important effect on hashing accuracy and the mere presence of such thresholds are a source of information leakage that can reduce hashing security. This dissertation outlines a systematic investigation of the quantization and encoding stages of robust hash functions. While existing literature has focused on the importance of quantization scheme, this research is the first to emphasise the importance of the quantizer training on both hashing accuracy and hashing security. The quantizer training process is presented in a statistical framework which allows a theoretical analysis of the effects of quantizer training on hashing performance. This is experimentally verified using a number of baseline robust image hashing algorithms over a large database of real world images. This dissertation also proposes a new randomization method for robust image hashing based on Higher Order Spectra (HOS) and Radon projections. The method is non-linear and this is an essential requirement for non-invertibility. The method is also designed to produce features more suited for quantization and encoding. The system can operate without the need for quantizer training, is more easily encoded and displays improved hashing performance when compared to existing robust image hashing algorithms. The dissertation also shows how the HOS method can be adapted to work with biometric features obtained from 2D and 3D face images.

Threshold privacy preserving keyword searches

We consider the following problem: users of an organization wish to outsource the storage of sensitive data to a large database server. It is assumed that the server storing the data is untrusted so the data stored have to be encrypted. We further suppose that the manager of the organization has the right to access all data, but a member of the organization can not access any data alone. The member must collaborate with other members to search for the desired data. In this paper, we investigate the notion of threshold privacy preserving keyword search (TPPKS) and define its security requirements. We construct a TPPKS scheme and show the proof of security under the assumptions of intractability of discrete logarithm, decisional Diffie-Hellman and computational Diffie-Hellman problems.

Emerging methods in using GIS to analyse Barmah Forest virus disease in Queensland, Australia

Barmah Forest virus (BFV) disease is an emerging mosquito-borne disease in Australia. We aimed to outline some recent methods in using GIS for the analysis of BFV disease in Queensland, Australia. A large database of geocoded BFV cases has been established in conjunction with population data. The database has been used in recently published studies conducted by the authors to determine spatio-temporal BFV disease hotspots and spatial patterns using spatial autocorrelation and semi-variogram analysis in conjunction with the development of interpolated BFV disease standardised incidence maps. This paper briefly outlines spatial analysis methodologies using GIS tools used in those studies. This paper summarises methods and results from previous studies by the authors, and presents a GIS methodology to be used in future spatial analytical studies in attempt to enhance the understanding of BFV disease in Queensland. The methodology developed is useful in improving the analysis of BFV disease data and will enhance the understanding of the BFV disease distribution in Queensland, Australia.

Genomic approach to organ-specific growth control

Growth is a fundamental aspect of life cycle of all organisms. Body size varies highly in most animal groups, such as mammals. Moreover, growth of a multicellular organism is not uniform enlargement of size, but different body parts and organs grow to their characteristic sizes at different times. Currently very little is known about the molecular mechanisms governing this organ-specific growth. The genome sequencing projects have provided complete genomic DNA sequences of several species over the past decade. The amount of genomic sequence information, including sequence variants within species, is constantly increasing. Based on the universal genetic code, we can make sense of this sequence information as far as it codes proteins. However, less is known about the molecular mechanisms that control expression of genes, and about the variations in gene expression that underlie many pathological states in humans. This is caused in part by lack of information about the second genetic code that consists of the binding specificities of transcription factors and the combinatorial code by which transcription factor binding sites are assembled to form tissue-specific and/or ligand-regulated enhancer elements. This thesis presents a high-throughput assay for identification of transcription factor binding specificities, which were then used to measure the DNA binding profiles of transcription factors involved in growth control. We developed ‘enhancer element locator’, a computational tool, which can be used to predict functional enhancer elements. A genome-wide prediction of human and mouse enhancer elements generated a large database of enhancer elements. This database can be used to identify target genes of signaling pathways, and to predict activated transcription factors based on changes in gene expression. Predictions validated in transgenic mouse embryos revealed the presence of multiple tissue-specific enhancers in mouse c- and N-Myc genes, which has implications to organ specific growth control and tumor type specificity of oncogenes. Furthermore, we were able to locate a variation in a single nucleotide, which carries a susceptibility to colorectal cancer, to an enhancer element and propose a mechanism by which this SNP might be involved in generation of colorectal cancer.

The role of lakes in carbon cycling in boreal catchments

Lakes are an important component of ecosystem carbon cycle through both organic carbon sequestration and carbon dioxide and methane emissions, although they cover only a small fraction of the Earth's surface area. Lake sediments are considered to be one of rather perma-nent sinks of carbon in boreal regions and furthermore, freshwater ecosystems process large amounts of carbon originating from terrestrial sources. These carbon fluxes are highly uncer-tain especially in the changing climate. -- The present study provides a large-scale view on carbon sources and fluxes in boreal lakes situated in different landscapes. We present carbon concentrations in water, pools in lake se-diments, and carbon gas (CO2 and CH4) fluxes from lakes. The study is based on spatially extensive and randomly selected Nordic Lake Survey (NLS) database with 874 lakes. The large database allows the identification of the various factors (lake size, climate, and catchment land use) determining lake water carbon concentrations, pools and gas fluxes in different types of lakes along a latitudinal gradient from 60oN to 69oN. Lakes in different landscapes vary in their carbon quantity and quality. Carbon (C) content (total organic and inorganic carbon) in lakes is highest in agriculture and peatland dominated areas. In peatland rich areas organic carbon dominated in lakes but in agricultural areas both organic and inorganic C concentrations were high. Total inorganic carbon in the lake water was strongly dependent on the bedrock and soil quality in the catchment, especially in areas where human influence in the catchment is low. In inhabited areas both agriculture and habitation in the catchment increase lake TIC concentrations, since in the disturbed soils both weathering and leaching are presumably more efficient than in pristine areas. TOC concentrations in lakes were related to either catchment sources, mainly peatlands, or to retention in the upper watercourses. Retention as a regulator of the TOC concentrations dominated in southern Finland, whereas the peatland sources were important in northern Finland. The homogeneous land use in the north and the restricted catchment sources of TOC contribute to the close relationship between peatlands and the TOC concentrations in the northern lakes. In southern Finland the more favorable climate for degradation and the multiple sources of TOC in the mixed land use highlight the importance of retention. Carbon processing was intensive in the small lakes. Both CO2 emission and the Holocene C pool in sediments per square meter of the lake area were highest in the smallest lakes. How-ever, because the total area of the small lakes on the areal level is limited, the large lakes are important units in C processing in the landscape. Both CO2 and CH4 concentrations and emissions were high in eutrophic lakes. High availability of nutrients and the fresh organic matter enhance degradation in these lakes. Eutrophic lakes are often small and shallow, enabling high contact between the water column and the sediment. At the landscape level, the lakes in agricultural areas are often eutrophic due to fertile soils and fertilization of the catchments, and therefore they also showed the highest CO2 and CH4 concentrations. Export from the catchments and in-lake degradation were suggested to be equally important sources of CO2 and CH4 in fall when the lake water column was intensively mixed and the transport of sub-stances from the catchment was high due to the rainy season. In the stagnant periods, especially in the winter, in-lake degradation as a gas source was highlighted due to minimal mixing and limited transport of C from the catchment. The strong relationship between the annual CO2 level of lakes and the annual precipitation suggests that climate change can have a major impact on C cycling in the catchments. Increase in precipitation enhances DOC export from the catchments and leads to increasing greenhouse gas emissions from lakes. The total annual CO2 emission from Finnish lakes was estimated to be 1400 Gg C a-1. The total lake sediment C pool in Finland was estimated to be 0.62 Pg, giving an annual sink in Finnish lakes of 65 Gg C a-1.