Enhancing 3D human pose estimation with NIR single-pixel imaging and time-of-flight technology: a deep learning approach.

The extraction of 3D human pose and body shape details from a single monocular image is a significant challenge in computer vision. Traditional methods use RGB images, but these are constrained by varying lighting and occlusions. However, cutting-edge developments in imaging technologies have introduced new techniques such as single-pixel imaging (SPI) that can surmount these hurdles. In the near-infrared spectrum, SPI demonstrates impressive capabilities in capturing a 3D human pose. This wavelength can penetrate clothing and is less influenced by lighting variations than visible light, thus providing a reliable means to accurately capture body shape and pose data, even in difficult settings. In this work, we explore the use of an SPI camera operating in the NIR with time-of-flight (TOF) at bands 850-1550 nm as a solution to detect humans in nighttime environments. The proposed system uses the vision transformers (ViT) model to detect and extract the characteristic features of humans for integration over a 3D body model SMPL-X through 3D body shape regression using deep learning. To evaluate the efficacy of NIR-SPI 3D image reconstruction, we constructed a laboratory scenario that simulates nighttime conditions, enabling us to test the feasibility of employing NIR-SPI as a vision sensor in outdoor environments. By assessing the results obtained from this setup, we aim to demonstrate the potential of NIR-SPI as an effective tool to detect humans in nighttime scenarios and capture their accurate 3D body pose and shape.

A High-Performance and Cost-Effective Field Programmable Gate Array-Based Motor Drive Emulator.

This work presents a hardware-based digital emulator capable of digitally driving a permanent magnet synchronous machine electronic setup. The aim of this work is to present a high-performance, cost-effective, and portable complementary solution when new paradigms of electronic drive design are generated, such as machine early failure detection, fault-tolerant drive, and high-performance control strategy implementations. In order to achieve the high performance required by the digital emulator, the electronic drive models (permanent-magnet synchronous machine, voltage-source inverter, motor-control strategy) are digitally described in Verilog hardware description language and implemented on a field programmable gate array (FPGA) digital platform using two approaches: parallel and sequential methods. The results obtained show the effectiveness of the digital emulator design, and the resources used by the solution presented can be implemented on a low-cost digital platform that reveals a cost-effective operation of the solution presented.

Deep-learning blurring correction of images obtained from NIR single-pixel imaging.

In challenging scenarios characterized by low-photon conditions or the presence of scattering effects caused by rain, fog, or smoke, conventional silicon-based cameras face limitations in capturing visible images. This often leads to reduced visibility and image contrast. However, using near-infrared (NIR) light within the range of 850-1550 nm offers the advantage of reduced scattering by microparticles, making it an attractive option for imaging in such conditions. Despite NIR's advantages, NIR cameras can be prohibitively expensive. To address this issue, we propose a vision system that leverages NIR active illumination single-pixel imaging (SPI) operating at 1550 nm combined with time of flight operating at 850 nm for 2D image reconstruction, specifically targeting rainy conditions. We incorporate diffusion models into the proposed system to enhance the quality of NIR-SPI images. By simulating various conditions of background illumination and droplet size in an outdoor laboratory scenario, we assess the feasibility of utilizing NIR-SPI as a vision sensor in challenging outdoor environments.

Modeling of Particulate Pollutants Using a Memory-Based Recurrent Neural Network Implemented on an FPGA.

The present work describes the training and subsequent implementation on an FPGA board of an LSTM neural network for the modeling and prediction of the exceedances of criteria pollutants such as nitrogen dioxide (NO2), carbon monoxide (CO), and particulate matter (PM10 and PM2.5). Understanding the behavior of pollutants and assessing air quality in specific geographical regions is crucial. Overexposure to these pollutants can cause harm to both natural ecosystems and living organisms, including humans. Therefore, it is essential to develop a solution that can accurately evaluate pollution levels. One potential approach is to implement a modified LSTM neural network on an FPGA board. This implementation obtained an 11% improvement compared to the original LSTM network, demonstrating that the proposed architecture is able to maintain its functionality despite reducing the number of neurons in its initial layers. It shows the feasibility of integrating a prediction network into a limited system such as an FPGA board, but easily coupled to a different system. Importantly, this implementation does not compromise the prediction accuracy for both 24 h and 72 h time frames, highlighting an opportunity for further enhancement and refinement.

Broken Bar Fault Detection Using Taylor-Fourier Filters and Statistical Analysis.

Broken rotor bars in induction motors make up one of the typical fault types that are challenging to detect. This type of damage can provoke adverse effects on the motors, such as mechanical and electrical stresses, together with an increase in electricity consumption, causing higher operative costs and losses related to the maintenance times or even the motor replacement if the damage has led to a complete failure. To prevent such situations, diverse signal processing algorithms have been applied to incipient fault detection, using different variables to analyze, such as vibrations, current, or flux. To counteract the broken rotor bar damage, this paper focuses on a motor current signal analysis for early broken bar detection and classification by using the digital Taylor-Fourier transform (DTFT), whose implementation allows fine filtering and amplitude estimation with the final purpose of achieving an incipient fault detection. The detection is based on an analysis of variance followed by a Tukey test of the estimated amplitude. The proposed methodology is implemented in Matlab using the O-splines of the DTFT to reduce the computational load compared with other methods. The analysis is focused on groups of 50-test of current signals corresponding to different damage levels for a motor operating at 50% and 75% of its full load.

Single-pixel imaging: An overview of different methods to be used for 3D space reconstruction in harsh environments.

Different imaging solutions have been proposed over the last few decades, aimed at three-dimensional (3D) space reconstruction and obstacle detection, either based on stereo-vision principles using active pixel sensors operating in the visible part of the spectra or based on active Near Infra-Red (NIR) illumination applying the time-of-flight principle, to mention just a few. If extremely low quantum efficiencies for NIR active illumination yielded by silicon-based detector solutions are considered together with the huge photon noise levels produced by the background illumination accompanied by Rayleigh scattering effects taking place in outdoor applications, the operating limitations of these systems under harsh weather conditions, especially if relatively low-power active illumination is used, are evident. If longer wavelengths for active illumination are applied to overcome these issues, indium gallium arsenide (InGaAs)-based photodetectors become the technology of choice, and for low-cost solutions, using a single InGaAs photodetector or an InGaAs line-sensor becomes a promising choice. In this case, the principles of Single-Pixel Imaging (SPI) and compressive sensing acquire a paramount importance. Thus, in this paper, we review and compare the different SPI developments reported. We cover a variety of SPI system architectures, modulation methods, pattern generation and reconstruction algorithms, embedded system approaches, and 2D/3D image reconstruction methods. In addition, we introduce a Near Infra-Red Single-Pixel Imaging (NIR-SPI) sensor aimed at detecting static and dynamic objects under outdoor conditions for unmanned aerial vehicle applications.

ECG-Based Identification of Sudden Cardiac Death through Sparse Representations.

Sudden Cardiac Death (SCD) is an unexpected sudden death due to a loss of heart function and represents more than 50% of the deaths from cardiovascular diseases. Since cardiovascular problems change the features in the electrical signal of the heart, if significant changes are found with respect to a reference signal (healthy), then it is possible to indicate in advance a possible SCD occurrence. This work proposes SCD identification using Electrocardiogram (ECG) signals and a sparse representation technique. Moreover, the use of fixed feature ranking is avoided by considering a dictionary as a flexible set of features where each sparse representation could be seen as a dynamic feature extraction process. In this way, the involved features may differ within the dictionary's margin of similarity, which is better-suited to the large number of variations that an ECG signal contains. The experiments were carried out using the ECG signals from the MIT/BIH-SCDH and the MIT/BIH-NSR databases. The results show that it is possible to achieve a detection 30 min before the SCD event occurs, reaching an an accuracy of 95.3% under the common scheme, and 80.5% under the proposed multi-class scheme, thus being suitable for detecting a SCD episode in advance.

Fabrication of Microbolometer Arrays Based on Polymorphous Silicon-Germanium.

This work reports the development of arrays of infrared sensors (microbolometers) using a hydrogenated polymorphous silicon-germanium alloy (pm-SixGe1-x:H). Basically, polymorphous semiconductors consist of an amorphous semiconductor matrix with embedded nanocrystals of about 2-3 nm. The pm-SixGe1-x:H alloy studied has a high temperature coefficient of resistance (TCR) of 4.08%/K and conductivity of 1.5 × 10-5 S∙cm-1. Deposition of thermosensing film was made by plasma-enhanced chemical vapor deposition (PECVD) at 200 °C, while the area of the devices is 50 × 50 µm2 with a fill factor of 81%. Finally, an array of 19 × 20 microbolometers was packaged for electrical characterization. Voltage responsivity values were obtained in the range of 4 × 104 V/W and detectivity around 2 × 107 cm∙Hz1/2/W with a polarization current of 70 µA at a chopper frequency of 30 Hz. A minimum value of 2 × 10-10 W/Hz1/2 noise equivalent power was obtained at room temperature. In addition, it was found that all the tested devices responded to incident infrared radiation, proving that the structure and mechanical stability are excellent.

Deep Learning Classification for Diabetic Foot Thermograms.

According to the World Health Organization (WHO), Diabetes Mellitus (DM) is one of the most prevalent diseases in the world. It is also associated with a high mortality index. Diabetic foot is one of its main complications, and it comprises the development of plantar ulcers that could result in an amputation. Several works report that thermography is useful to detect changes in the plantar temperature, which could give rise to a higher risk of ulceration. However, the plantar temperature distribution does not follow a particular pattern in diabetic patients, thereby making it difficult to measure the changes. Thus, there is an interest in improving the success of the analysis and classification methods that help to detect abnormal changes in the plantar temperature. All this leads to the use of computer-aided systems, such as those involved in artificial intelligence (AI), which operate with highly complex data structures. This paper compares machine learning-based techniques with Deep Learning (DL) structures. We tested common structures in the mode of transfer learning, including AlexNet and GoogleNet. Moreover, we designed a new DL-structure, which is trained from scratch and is able to reach higher values in terms of accuracy and other quality measures. The main goal of this work is to analyze the use of AI and DL for the classification of diabetic foot thermograms, highlighting their advantages and limitations. To the best of our knowledge, this is the first proposal of DL networks applied to the classification of diabetic foot thermograms. The experiments are conducted over thermograms of DM and control groups. After that, a multi-level classification is performed based on a previously reported thermal change index. The high accuracy obtained shows the usefulness of AI and DL as auxiliary tools to aid during the medical diagnosis.

On the synchronization techniques of chaotic oscillators and their FPGA-based implementation for secure image transmission.

Synchronizing chaotic oscillators has been a challenge to guarantee successful applications in secure communications. That way, three synchronization techniques are applied herein to twenty two chaotic oscillators, three of them based on piecewise-linear functions and nineteen proposed by Julien C. Sprott. These chaotic oscillators are simulated to generate chaotic time series that are used to evaluate their Lyapunov exponents and Kaplan-Yorke dimension to rank their unpredictability. The oscillators with the high positive Lyapunov exponent are implemented into a field-programmable gate array (FPGA), and afterwards they are synchronized in a master-slave topology applying three techniques: the seminal work introduced by Pecora-Carroll, Hamiltonian forms and observer approach, and open-plus-closed-loop (OPCL). These techniques are compared with respect to their synchronization error and latency that is associated to the FPGA implementation. Finally, the chaotic oscillators providing the high positive Lyapunov exponent are synchronized and applied to a communication system with chaotic masking to perform a secure image transmission. Correlation analysis is performed among the original image, the chaotic channel and the recovered image for the three synchronization schemes. The experimental results show that both Hamiltonian forms and OPCL can recover the original image and its correlation with the chaotic channel is as low as 0.00002, demonstrating the advantage of synchronizing chaotic oscillators with high positive Lyapunov exponent to guarantee high security in data transmission.

Use of kurtosis for locating deep blood vessels in raw speckle imaging using a homogeneity representation.

Visualization of deep blood vessels in speckle images is an important task as it is used to analyze the dynamics of the blood flow and the health status of biological tissue. Laser speckle imaging is a wide-field optical technique to measure relative blood flow speed based on the local speckle contrast analysis. However, it has been reported that this technique is limited to certain deep blood vessels (about ? = 300 ?? ? m ) because of the high scattering of the sample; beyond this depth, the quality of the vessel's image decreases. The use of a representation based on homogeneity values, computed from the co-occurrence matrix, is proposed as it provides an improved vessel definition and its corresponding diameter. Moreover, a methodology is proposed for automatic blood vessel location based on the kurtosis analysis. Results were obtained from the different skin phantoms, showing that it is possible to identify the vessel region for different morphologies, even up to 900 ?? ? m in depth.

VHDL Descriptions for the FPGA Implementation of PWL-Function-Based Multi-Scroll Chaotic Oscillators.

Nowadays, chaos generators are an attractive field for research and the challenge is their realization for the development of engineering applications. From more than three decades ago, chaotic oscillators have been designed using discrete electronic devices, very few with integrated circuit technology, and in this work we propose the use of field-programmable gate arrays (FPGAs) for fast prototyping. FPGA-based applications require that one be expert on programming with very-high-speed integrated circuits hardware description language (VHDL). In this manner, we detail the VHDL descriptions of chaos generators for fast prototyping from high-level programming using Python. The cases of study are three kinds of chaos generators based on piecewise-linear (PWL) functions that can be systematically augmented to generate even and odd number of scrolls. We introduce new algorithms for the VHDL description of PWL functions like saturated functions series, negative slopes and sawtooth. The generated VHDL-code is portable, reusable and open source to be synthesized in an FPGA. Finally, we show experimental results for observing 2, 10 and 30-scroll attractors.

Gyroscope-driven mouse pointer with an EMOTIV® EEG headset and data analysis based on Empirical Mode Decomposition.

This paper presents a project on the development of a cursor control emulating the typical operations of a computer-mouse, using gyroscope and eye-blinking electromyographic signals which are obtained through a commercial 16-electrode wireless headset, recently released by Emotiv. The cursor position is controlled using information from a gyroscope included in the headset. The clicks are generated through the user's blinking with an adequate detection procedure based on the spectral-like technique called Empirical Mode Decomposition (EMD). EMD is proposed as a simple and quick computational tool, yet effective, aimed to artifact reduction from head movements as well as a method to detect blinking signals for mouse control. Kalman filter is used as state estimator for mouse position control and jitter removal. The detection rate obtained in average was 94.9%. Experimental setup and some obtained results are presented.

Novel Oversampling Technique for Improving Signal-to-Quantization Noise Ratio on Accelerometer-Based Smart Jerk Sensors in CNC Applications.

Jerk monitoring, defined as the first derivative of acceleration, has become a major issue in computerized numeric controlled (CNC) machines. Several works highlight the necessity of measuring jerk in a reliable way for improving production processes. Nowadays, the computation of jerk is done by finite differences of the acceleration signal, computed at the Nyquist rate, which leads to low signal-to-quantization noise ratio (SQNR) during the estimation. The novelty of this work is the development of a smart sensor for jerk monitoring from a standard accelerometer, which has improved SQNR. The proposal is based on oversampling techniques that give a better estimation of jerk than that produced by a Nyquist-rate differentiator. Simulations and experimental results are presented to show the overall methodology performance.