Toward real-time volumetric tomography for combustion diagnostics via dimension reduction

Tri-zone flame spatial structure imaging combined with endogenic polarized scattering

We propose a multi-mode optical imaging method to retrieve the 2D and 3D spatial structures of the preheating, reaction, and recombination zones of an axisymmetric steady flame. In the proposed method, an infrared camera, a visible light monochromatic camera, and a polarization camera are triggered synchronously to capture 2D flame images, and their corresponding 3D images are reconstructed by combining different projection position images. The results of the experiments conducted indicate that the infrared and visible light images represent the flame preheating and flame reaction zones, respectively. The polarized image can be obtained by computing the degree of linear polarization (DOLP) of raw images captured by the polarization camera. We discover that the highlighted regions in the DOLP images lie outside the infrared and visible light zones; they are insensitive to the flame reaction and have different spatial structures for different fuels. We deduce that the combustion product particles cause endogenic polarized scattering, and that the DOLP images represent the flame recombination zone. This study focuses on the combustion mechanisms, such as the formation of combustion products and quantitative flame composition and structure.

Gray wolf optimization-extreme learning machine approach for diabetic retinopathy detection

Many works have employed Machine Learning (ML) techniques in the detection of Diabetic Retinopathy (DR), a disease that affects the human eye. However, the accuracy of most DR detection methods still need improvement. Gray Wolf Optimization-Extreme Learning Machine (GWO-ELM) is one of the most popular ML algorithms, and can be considered as an accurate algorithm in the process of classification, but has not been used in solving DR detection. Therefore, this work aims to apply the GWO-ELM classifier and employ one of the most popular features extractions, Histogram of Oriented Gradients-Principal Component Analysis (HOG-PCA), to increase the accuracy of DR detection system. Although the HOG-PCA has been tested in many image processing domains including medical domains, it has not yet been tested in DR. The GWO-ELM can prevent overfitting, solve multi and binary classifications problems, and it performs like a kernel-based Support Vector Machine with a Neural Network structure, whilst the HOG-PCA has the ability to extract the most relevant features with low dimensionality. Therefore, the combination of the GWO-ELM classifier and HOG-PCA features might produce an effective technique for DR classification and features extraction. The proposed GWO-ELM is evaluated based on two different datasets, namely APTOS-2019 and Indian Diabetic Retinopathy Image Dataset (IDRiD), in both binary and multi-class classification. The experiment results have shown an excellent performance of the proposed GWO-ELM model where it achieved an accuracy of 96.21% for multi-class and 99.47% for binary using APTOS-2019 dataset as well as 96.15% for multi-class and 99.04% for binary using IDRiD dataset. This demonstrates that the combination of the GWO-ELM and HOG-PCA is an effective classifier for detecting DR and might be applicable in solving other image data types.

4D temperature measurements using tomographic two-color pyrometry

This work presents a new approach for high-speed four-dimensional (3D + t) thermometry using only two high-speed cameras which are equipped with different band pass filters to capture thermal radiation signals at two narrow wavelength bands. With the help of a customized fiber bundle and a beam splitter, a total number of nine projections at each band were recorded, and the temperature distribution was evaluated by tomographic two-color pyrometry. In order to validate the effectiveness of this method, the 3D temperature distribution of a premixed steady flat flame was evaluated. The determined temperatures were compared to those of other studies, as well as to the results from inverse Abel transform and line-of-sight data. Further, the 3D temperature evolution of a weakly turbulent diffusion flame was observed at a repetition rate of 7.5 kHz. Such 4D temperature measurements are expected to be valuable in understanding turbulent combustion mechanisms especially of practical devices.

Sparse regularization-based reconstruction for 3D flame chemiluminescence tomography

Flame chemiluminescence tomography (FCT) is a non-intrusive method that is based on using cameras to measure projections, and it plays a crucial role in combustion diagnostics and measurement. Mathematically, the inversion problem is ill-posed, and in the case of limited optical accessibility in practical applications, it is rank deficient. Therefore, the solution process should ideally be supported by prior information, which can be based on the known physics. In this work, the total variation (TV) regularization has been combined with the well-known algebraic reconstruction technique (ART) for practical FCT applications. The TV method endorses smoothness while also preserving typical flame features such as the flame front. Split Bregman iteration has been adopted for TV minimization. Five different noise conditions and the chosen regularization parameter have been tested in numerical studies. Additionally, for the 12 perspectives, an experimental FCT system is demonstrated, which is utilized to recover the three-dimensional (3D) chemiluminescence distribution of candle flames. Both the numerical and experimental studies show that the typical line artifacts that appear with the conventional ART algorithm when recovering the continuous chemiluminescence field of the flames are significantly reduced with the proposed algorithm.

Optimised genetic algorithm-extreme learning machine approach for automatic COVID-19 detection

The coronavirus disease (COVID-19), is an ongoing global pandemic caused by severe acute respiratory syndrome. Chest Computed Tomography (CT) is an effective method for detecting lung illnesses, including COVID-19. However, the CT scan is expensive and time-consuming. Therefore, this work focus on detecting COVID-19 using chest X-ray images because it is widely available, faster, and cheaper than CT scan. Many machine learning approaches such as Deep Learning, Neural Network, and Support Vector Machine; have used X-ray for detecting the COVID-19. Although the performance of those approaches is acceptable in terms of accuracy, however, they require high computational time and more memory space. Therefore, this work employs an Optimised Genetic Algorithm-Extreme Learning Machine (OGA-ELM) with three selection criteria (i.e., random, K-tournament, and roulette wheel) to detect COVID-19 using X-ray images. The most crucial strength factors of the Extreme Learning Machine (ELM) are: (i) high capability of the ELM in avoiding overfitting; (ii) its usability on binary and multi-type classifiers; and (iii) ELM could work as a kernel-based support vector machine with a structure of a neural network. These advantages make the ELM efficient in achieving an excellent learning performance. ELMs have successfully been applied in many domains, including medical domains such as breast cancer detection, pathological brain detection, and ductal carcinoma in situ detection, but not yet tested on detecting COVID-19. Hence, this work aims to identify the effectiveness of employing OGA-ELM in detecting COVID-19 using chest X-ray images. In order to reduce the dimensionality of a histogram oriented gradient features, we use principal component analysis. The performance of OGA-ELM is evaluated on a benchmark dataset containing 188 chest X-ray images with two classes: a healthy and a COVID-19 infected. The experimental result shows that the OGA-ELM achieves 100.00% accuracy with fast computation time. This demonstrates that OGA-ELM is an efficient method for COVID-19 detecting using chest X-ray images.

Three-dimensional rapid flame chemiluminescence tomography via deep learning

Flame chemiluminescence tomography (FCT) plays an important role in combustion monitoring and diagnostics due to the easy implementation and non-intrusion. However, on account of the high data throughput and the inefficiency of the conventional iteration methods, the 3D reconstructions in FCT are typically conducted off-line and time-consuming. In this work, we present a 3D rapid FCT reconstruction system based on convolutional neural networks (CNN) model for practical combustion measurement, which has the ability to reconstruct 3D flame distribution rapidly after training process. First, the numerical simulation has been performed by creating three cases of phantoms which are designed to mimic the 3D conical flame. Next, after the evaluation of loss function and training time, the optimal CNN architecture has been determined and certificated quantitatively. Finally, a real time FCT system consisting of 12 color CCD cameras is realized and multispectral separation algorithm is adopted to extract CH* and C2* components. Certificated by practical measurements testing, the proposed CNN model is able to reconstruct 3D flame structure from real time captured projections with credible accuracy and structure similarity. Furthermore, compared with conventional iteration reconstruction method, the proposed CNN model shows better performance on obviously improving reconstruction speed and it is expected to achieve 3D rapid monitoring of flames.

Simulating imaging-based tomographic systems using optical design software for resolving 3D structures of translucent media

Imaging-based tomography is emerging as the technique of choice for resolving 3D structures of translucent media, in particular for applications in external beam radiation therapy and combustion diagnostics. However, designing experimental prototypes is time-consuming and costly, and is carried out without the certainty of the imaging optics being optimal. In this paper, we present an optical-design-software-based method that enables end-to-end simulation imaging-based tomography systems. The method, developed using the real ray tracing features of Zemax OpticStudio, was validated in the context of 3D scintillation dosimetry, where multiple imaging systems are used to image the 3D light pattern emitted within an irradiated cubic plastic scintillator volume. The flexibility of the workflow enabled the assessment and comparison of the tomographic performance of standard and focused plenoptic cameras for the reconstruction of a clinical radiation dose distribution. The versatility of the proposed method offers the potential to ease the developmental and optimization process of imaging systems used in volumetric emission computed tomography applications.

Demonstration of a cost-effective single-pixel UV camera for flame chemiluminescence imaging

Combustion is the dominant form of energy conversion for a span of power systems such as engines and power plant boilers. It is an extremely complicated process which produces a huge number of intermediate products that usually feature non-uniform spatial distributions. Among those intermediate products, free radicals emitting in the UV band are of special interest because they contain abundant useful information of the target flame. Imaging methods such as planar laser-induced fluorescence or chemiluminescence imaging with UV cameras are of paramount importance to resolve the non-uniformities for a better understanding of combustion. However, a major limitation of UV cameras is that they are usually expensive, especially when multiple cameras are needed, such as in a tomographic system. In this work, we report the attempt of flame imaging with a cost-effective single-pixel UV camera which enables 2D spatial resolution of a single-pixel detector through light modulation to overcome this limitation. Meanwhile, numerical studies were conducted to systematically assess the performance of a few representative reconstruction algorithms. The impact of important factors such as sampling ratio and measurement matrix were also investigated. A validation as well as a demonstrative experiment was also conducted to reconstruct a laminar diffusion flame. The high similarity between the reconstruction and the image taken by a CMOS camera proves the feasibility of flame imaging with a single-pixel camera.

Assessment of plenoptic imaging for reconstruction of 3D discrete and continuous luminous fields

Volumetric tomography has become an indispensable tool for flow diagnostics. However, it usually suffers from high experimental costs as multiple cameras are required in a typical tomographic system. Plenoptic imaging (PI) is a promising alternative which can simultaneously record spatial and angular information using only one single camera. Although PI has been pioneered by a few groups for 3D flow imaging, this particular application is still at its early stage of development and there are some aspects that need further investigation. In this work, we will systematically assess three representative tomographic algorithms for PI via numerical studies. In addition, we show here how 3D PI inversion can be interpreted from a tomographic perspective and how to conveniently perform the calibration with an existing well-established method which can take into account the effect of lens distortion. A proof-of-concept experiment was also conducted, and the conclusions drawn were consistent with those from numerical studies. Although this work was discussed under the context of flow/flame imaging, the general conclusions are also applicable to other application fields, such as biomedical imaging.

Time-resolved measurements of a swirl flame at 4 kHz via computed tomography of chemiluminescence

Computed tomography of a chemiluminescence (CTC) system was implemented to provide time-resolved 3D measurements of an unconfined turbulent swirl flame. This system was designed in a cost-effective manner and employed three customized view registration assemblies to simultaneously capture eight projections of the target flame at a repetition rate of 4 kHz. Both time-resolved and time-averaged tomographic reconstructions were performed based on data acquired for a duration of 250 ms. Both qualitative and quantitative validations suggested the correctness of our implementation. The time-resolved instantaneous reconstructions successfully captured the evolution of the structural features of the swirl flame such as local extinctions and the helical mode. Based on the reconstructions, the centroids of chemiluminescence for all the layers were calculated. The trajectory of these centroids provided insights into the flow motion and suggested a rotating helical structure of the swirl flame. These results demonstrated the feasibility of resolving the dynamics of turbulent swirl flames with a kHz temporal resolution using the relatively inexpensive CTC system.

Rapid tomographic reconstruction based on machine learning for time-resolved combustion diagnostics

Optical tomography has attracted surged research efforts recently due to the progress in both the imaging concepts and the sensor and laser technologies. The high spatial and temporal resolutions achievable by these methods provide unprecedented opportunity for diagnosis of complicated turbulent combustion. However, due to the high data throughput and the inefficiency of the prevailing iterative methods, the tomographic reconstructions which are typically conducted off-line are computationally formidable. In this work, we propose an efficient inversion method based on a machine learning algorithm, which can extract useful information from the previous reconstructions and build efficient neural networks to serve as a surrogate model to rapidly predict the reconstructions. Extreme learning machine is cited here as an example for demonstrative purpose simply due to its ease of implementation, fast learning speed, and good generalization performance. Extensive numerical studies were performed, and the results show that the new method can dramatically reduce the computational time compared with the classical iterative methods. This technique is expected to be an alternative to existing methods when sufficient training data are available. Although this work is discussed under the context of tomographic absorption spectroscopy, we expect it to be useful also to other high speed tomographic modalities such as volumetric laser-induced fluorescence and tomographic laser-induced incandescence which have been demonstrated for combustion diagnostics.