Search

Design requirements for an 8000 frame/s dual-wavelength ratiometric chemical species tomography system, intended for hydrocarbon vapor imaging in one cylinder of a standard automobile engine, are examined. The design process is guided by spectroscopic measurements on iso-octane and by comprehensive results from laboratory phantoms and research engines, including results on temporal resolution performance. Novel image reconstruction techniques, necessary for this application, are presented. Recent progress toward implementation, including details of the optical access arrangement employed and signal-to-noise issues, is described. We present first cross-cylinder IR absorption measurements from a reduced channel-count (nontomographic) system and discuss the prospects for imaging. OCIS codes: 100.6950, 110.3080, 120.1740, 300.6260.

The recognition of gait pattern variation is of high importance to various industrial and commercial applications, including security, sport, virtual reality, gaming, robotics, medical rehabilitation, mental illness diagnosis, space exploration, and others. The purpose of this paper is to study the nature of gait variability in more detail, by identifying gait intervals responsible for gait pattern variations in individuals, as well as between individuals, using cognitive demanding tasks. This work uses deep learning methods for sensor fusion of 116 plastic optical fiber (POF) distributed sensors for gait recognition. The floor sensor system captures spatiotemporal samples due to varying ground reaction force (GRF) in multiples of up to 4 uninterrupted steps on a continuous $2\times 1$ m area. We demonstrate classifications of gait signatures, achieving up to 100% F1-score with Convolutional Neural Networks (CNN), in the context of gait recognition of 21 subjects, with imposters and clients. Classifications under cognitive load, induced by 4 different dual tasks, manifested lower F1-scores. Layer-Wise Relevance Propagation (LRP) methods are employed to decompose a trained neural network prediction to relevant standard events in the gait cycle, by generating a “heat map” over the input used for classification. This allows valuable insight into which parts of the gait spatiotemporal signal have the heaviest influence on the gait classification and consequently, which gait events, such as heel strike or toe-off, are mostly affected by cognitive load. [ABSTRACT FROM AUTHOR]

Floor Sensors (FS) are used to capture information from the force induced on the contact surface by feet during gait. On the other hand, the Ambulatory Inertial Sensors (AIS) are used to capture the velocity, acceleration and orientation of the body during different activities. In this paper, fusion of the stated modalities is performed to overcome the challenge of gait classification from wearable sensors on the lower portion of human body not in contact with ground as in FS. Deep learning models are utilized for the automatic feature extraction of the ground reaction force obtained from a set of 116 FS and body movements from AIS attached at 3 different locations of lower body, which is novel. Spatio-temporal information of disproportionate inputs obtained from the two modalities is balanced and fused within deep learning network layers whilst reserving the categorical content for each gait activity. Our approach of fusion compensates the degradation in spatio-temporal accuracies in individual modalities and makes the overall classification outcomes more accurate. Further assessment of multi-modality based results show significant improvements in f-scores using different deep learning models i.e., LSTM (99.90%), 2D-CNN (88.73%), 1D-CNN (94.97%) and ANN (89.33%) respectively. [ABSTRACT FROM AUTHOR]

Gait activity classifications from single-modality data, e.g. acquired by separate vision, pressure, sound and inertial measurements, can be improved by complementary multi-modality fusion to capture a larger set of distinctive gait activity features. We demonstrate a feature-level based sensor fusion of spatio-temporal data obtained from a set of 116 collaborative floor sensors for spatio-temporal sampling of the ground reaction force and ambulatory inertial sensors at 3 positions on the human body. Principle Component Analysis and Canonical Correlation Analysis are used for automatic feature extraction. Fusion at feature level elucidates the balance between otherwise disproportional number of inputs from the two modalities, while reducing the overall number of inputs for classification without degrading substantially the information content. Improvement in the classification is achieved using K-Nearest Neighbor and Kernel Support Vector Machine, manifesting f-scores of 0.95 and 0.94 respectively. [ABSTRACT FROM AUTHOR]

IEEE journals are to a degree anthropomorphic: we have a bright and vivacious family, with veterans in their 70s, coexisting with ambitious newborns. On that scale, the event we are celebrating is the IEEE Sensors Journal having matured from its teenage years. As a typical 20-year old, it has put behind the doubts, the rush to test the unknown, the first disappointments, and the first dizzying success, now emerging as a publication with a distinctive face and character. [ABSTRACT FROM AUTHOR]

The essential human gait parameters are briefly reviewed, followed by a detailed review of the state of the art in deep learning for the human gait analysis. The modalities for capturing the gait data are grouped according to the sensing technology: video sequences, wearable sensors, and floor sensors, as well as the publicly available datasets. The established artificial neural network architectures for deep learning are reviewed for each group, and their performance are compared with particular emphasis on the spatiotemporal character of gait data and the motivation for multi-sensor, multi-modality fusion. It is shown that by most of the essential metrics, deep learning convolutional neural networks typically outperform shallow learning models. In the light of the discussed character of gait data, this is attributed to the possibility to extract the gait features automatically in deep learning as opposed to the shallow learning from the handcrafted gait features. [ABSTRACT FROM AUTHOR]

Hyperspectral tomography in THz amplitude contrast is presented using a time-domain spectroscopy system in the spectral range 0.3–2.5 THz. The Fourier transformed signal data are used to reconstruct test objects’ cross-sections at multiple frequencies using standard filtered backprojection. The full hyperspectral set of images reconstructed at around 300 adjacent spectral points is used to trace the combined contribution of Beer–Lambert volume attenuation, Fresnel reflection losses, and Rayleigh roughness scattering losses, which is in good overall agreement with the experimental results. The image quality for Styrofoam (refractive index around 1.02, attenuation coefficient <1 mm−1) test objects is best in the range 0.8–2.0 THz depending on the porosity of the material. [ABSTRACT FROM AUTHOR]

Human footsteps can provide a unique behavioural pattern for robust biometric systems. We propose spatio-temporal footstep representations from floor-only sensor data in advanced computational models for automatic biometric verification. Our models deliver an artificial intelligence capable of effectively differentiating the fine-grained variability of footsteps between legitimate users (clients) and impostor users of the biometric system. The methodology is validated in the largest to date footstep database, containing nearly 20,000 footstep signals from more than 120 users. The database is organized by considering a large cohort of impostors and a small set of clients to verify the reliability of biometric systems. We provide experimental results in 3 critical data-driven security scenarios, according to the amount of footstep data made available for model training: at airports security checkpoints (smallest training set), workspace environments (medium training set) and home environments (largest training set). We report state-of-the-art footstep recognition rates with an optimal equal false acceptance and false rejection rate (equal error rate) of 0.7 percent an improvement ratio of 371 percent compared to previous state-of-the-art. We perform a feature analysis of deep residual neural networks showing effective clustering of client's footstep data and to provide insights of the feature learning process. [ABSTRACT FROM AUTHOR]

We demonstrate accurate spatio-temporal gait data classification from raw tomography sensor data without the need to reconstruct images. This is based on a simple yet efficient machine learning methodology based on a convolutional neural network architecture for learning spatio-temporal features, automatically end-to-end from raw sensor data. In a case study on a floor pressure tomography sensor, experimental results show an effective gait pattern classification F-score performance of 97.88 $\pm$ 1.70%. It is shown that the automatic extraction of classification features from raw data leads to a substantially better performance, compared to features derived by shallow machine learning models that use the reconstructed images as input, implying that for the purpose of automatic decision-making it is possible to eliminate the image reconstruction step. This approach is portable across a range of industrial tasks that involve tomography sensors. The proposed learning architecture is computationally efficient, has a low number of parameters and is able to achieve reliable classification F-score performance from a limited set of experimental samples. We also introduce a floor sensor dataset of 892 samples, encompassing experiments of 10 manners of walking and 3 cognitive-oriented tasks to yield a total of 13 types of gait patterns. [ABSTRACT FROM PUBLISHER]

In this paper, we assess the capability of a unique unobtrusive footprint imaging sensor system, based on plastic optical fiber technology, to allow efficient gait analysis from time domain sensor data by pattern recognition techniques. Trial gait classification experiments are executed as ten manners of walking, affecting the amplitude and frequency characteristics of the temporal signals. The data analysis involves the design of five temporal features, subsequently analyzed in 14 different machine learning models, representing linear, non-linear, ensemble, and deep learning models. The model performance is presented as cross-validated accuracy scores for the best model-feature combinations, along with the optimal hyper-parameters for each of them. The best classification performance was observed for a random forest model with the adjacent mean feature, yielding a mean validation score of 90.84% ± 2.46%. We conclude that the floor sensor system is capable of detecting changes in gait by means of pattern recognition techniques applied in the time domain. This suggests that the footprint imaging sensor system is suitable for gait analysis applications ranging from healthcare to security. [ABSTRACT FROM PUBLISHER]

We demonstrate an algorithm, relevant to tomography sensor systems, to obtain images from the parallel reconstruction of essentially localized elements at different scales. This is achieved by combining methodology to reconstruct images from limited and/or truncated data, with the time-frequency capabilities of the wavelet transform. Multiscale, as well as time-frequency, localization properties of the separable two-dimensional wavelet transform are exploited as an approach for faster reconstruction. The speedup is realized not only by reducing the computation load on a single processor, but also by achieving the parallel reconstruction of several tiled blocks. With tiled-block image reconstruction by wavelet-based, parallel filtered back-projection (FBP), we measure more than 36 times gain in speed, compared with standard FBP. [ABSTRACT FROM PUBLISHER]

A method to produce spatially resolved images of the distribution of absorbing particles in the exhaust plume of a modified helicopter gas turbine engine is presented. Over a small region of the plume, in situ sensing of soot particles by laser-induced incandescence (LII) is demonstrated using fiber lasers with higher power ( $\sim 10$ W), longer pulse duration (>100 ns), and higher pulse repetition rates (>10 kHz) than the conventional LII. The sensitivity of the method is illustrated by the detection of ambient absorbing particles in background conditions with engine at rest. With a running engine, single-beam images are obtained in 0.01 s. The feasibility of using long-pulsed fiber lasers for soot particle concentration measurement is investigated using a representative laboratory system. The time-resolved LII behavior and the measurement linearity are investigated, demonstrating the suitability of using fiber lasers for soot particle measurement for aero-engine emissions. Results for normalized soot concentration are compared with extractive measurements illustrating good correlation across a range of engine speeds. This paper is the first step toward the development of a non-intrusive system for the measurement of 2-D soot concentration in the cross section of an aero-engine exhaust plume. [ABSTRACT FROM PUBLISHER]

We present a numerical study demonstrating the simultaneous reconstruction of the temperature, species concentration, and pressure of molecular water vapor, by tomographic means, from near-infrared laser absorption measurements over beam paths transecting the gaseous region. The synthetic measurement data is predicted from a set of randomized distributions of the reconstructed variables using spectroscopic theory and the HITRAN parameter database. Design optimization geometric orientation of the beams is performed to maximize reconstruction accuracy. [ABSTRACT FROM PUBLISHER]

We report on the photonic variant of the previously introduced guided-path tomography (GPT), by demonstrating a system for footstep imaging using plastic optical fiber (POF) sensors. The 1 m × 2 m sensor head is manufactured by attaching 80 POF sensors on a standard commercial carpet underlay. The sensing principle relies on the sensitivity of POF to bending, quantified by measuring light transmission. The photonic GPT system, comprising the sensor head with processing hardware and software, covered by a mass-production general-purpose carpet top, successfully performs footstep imaging and correctly displays the position and footfall of a person walking on the carpet in real time. We also present the implementation of fast footprint center of mass calculations, suitable for recording gait and footfall. A split-screen movie, showing the frame-by-frame camera-captured action next to the reproduced footprints, can be downloaded at http://ieeexplore.ieee.org. [ABSTRACT FROM PUBLISHER]

Computer simulations are used to introduce a new approach to measure temperature fields of a gas using laser attenuation measurements, which exhibits good theoretical accuracy despite very high static pressures. Near-infrared laser light is used to target specific molecular absorption lines of water vapor, whose strength depends (nonlinearly) on the gas temperature. This temperature can be inferred when multiple laser paths coincide nearby by tomographic reconstruction of the attenuation coefficient and then spectral fitting to local temperature, species concentration, and gas pressure. Temperature phantoms (invented distributions for the purpose of numerical testing) are used to simulate experimental results, which are then contaminated with Gaussian noise and used to reconstruct the temperature field. The root-mean-square reconstruction temperature error varied from \sim 0.4\%, in no Gaussian noise and at 1 bar, to 2.2%, in 5% noise and at 50 bar. [ABSTRACT FROM PUBLISHER]

Physical limitations found in industrial environments often restrain imaging for process tomography. When information is collected from sparse sensors, the acquired data is limited in terms of radial and angular sampling of the imaged slice. To overcome this problem, we demonstrate an efficient solution based on the parallel implementation of the sinogram recovery algorithm (SRA) for limited views in its variant based on the calculation of the coordinates of the center of mass (CoMs) of the subject under test, rather than performing the complete sinogram restoration. By introducing a modification in the existing SRA, we achieve high parallelization of each stage, making it ideal for implementation in hardware accelerated systems, such as field programmable gate arrays. The potential to parallelize the SRA is first studied in MATLAB, by processing all data projections concurrently and verifying performance by matching the results from the parallel and sequential implementations. Furthermore, the algorithm is coded in very high speed integrated circuits hardware description language, which is implemented and tested on a Xilinx Virtex 6 board. We report speedups of between three and four orders of magnitude, whereas the errors in CoMs' coordinates are reduced. [ABSTRACT FROM PUBLISHER]

Analog balanced Photodetection has found extensive usage for sensing of a weak absorption signal buried in laser intensity noise. This paper proposes schemes for compact, affordable, and flexible digital implementation of the already established analog balanced detection, as part of a multichannel digital tomography system. Variants of digitally balanced detection (DBD) schemes, suitable for weak signals on a largely varying background or weakly varying envelopes of high frequency carrier waves, are introduced analytically and elaborated in terms of algorithmic and hardware flow. The DBD algorithms are implemented on a low-cost general purpose reconfigurable hardware (field-programmable gate array), utilizing less than half of its resources. The performance of the DBD schemes compare favorably with their analog counterpart: A common mode rejection ratio of 50 dB was observed over a bandwidth of 300 kHz, limited mainly by the host digital hardware. The close relationship between the DBD outputs and those of known analog balancing circuits is discussed in principle and shown experimentally in the example case of propane gas detection. [ABSTRACT FROM AUTHOR]

This article introduces the design and implementation of a hierarchical multi digital signal processor system aimed to perform parallel multichannel measurements and data processing of the type widely used in hard-field tomography. Details are presented of a complete tomography system with modular and expandable architecture, capable of accommodating a variety of data processing modalities, configured by software. The configuration of the acquisition and processing circuits and the management of the data flow allow a data frame rate of up to 250 kHz. Results of a case study, guided path tomography for temperature mapping, are shown as a direct demonstration of the system’s capabilities. Digital lock-in detection is employed for data processing to extract the information from ac measurements of the temperature-induced resistance changes in an array of 32 noninteracting transducers, which is further exported for visualization. [ABSTRACT FROM AUTHOR]

This article introduces the design and implementation of an affordable high-performance set of identical data acquisition channels with digital processing capabilities. Each channel incorporates a versatile 16-bit sigma-delta analog-to-digital converter (ADC) with reconfigurable filter characteristics. The main component of each channel, a low-cost field-programmable gate array (FPGA), controls the ADC, serves as a random access memory to store the ADCs user-defined filters, and performs digital processing. A special case is illustrated, with the FPGA software configured to perform lock-in detection, which is widely applied in a number of tomography modalities. The detection scheme, based on a quadrature demodulator, utilizes only a fraction of the FPGA resources and introduces errors orders of magnitude less than the quantization error of the ADC. Implementations other than a lock-in amplifier can be realized without additional hardware intervention. [ABSTRACT FROM AUTHOR]