CERES :: Browsing by Author "Yuen, Peter W. T."

Browsing by Author "Yuen, Peter W. T."

Now showing 1 - 20 of 38

Open Access
Automatic extraction of material defect size by infrared image sequence
(MDPI, 2020-11-20) Yuan, Lihua; Zhu, Xiao; Sun, Quanbin; Liu, Haibo; Yuen, Peter W. T.; Liu, Yonghuai
A typical pulsed thermography procedure results in a sequence of infrared images that reflects the evolution of temperature over time. Many features of defects, such as shape, position, and size, are derived from single image by image processing. Hence, determining the key frame from the sequence is an important problem to be solved first. A maximum standard deviation of the sensitive region method was proposed, which can identify a reasonable image frame automatically from an infrared image sequence; then, a stratagem of image composition was applied for enhancing the detection of deep defects in the key frame. Blob analysis had been adopted to acquire general information of defects such as their distributions and total number of defects. A region of interest of the defect was automatically located by its key frame combined with blob analysis. The defect information was obtained through image segmentation techniques. To obtain a robustness of results, a method of two steps of detection was proposed. The specimen of polyvinyl chloride with two artificial defects at different depths as an example was used to demonstrate how to operate the proposed method for an accurate result. At last, the proposed method was successfully adopted to examine the damage of carbon fiber-reinforced polymer. A comparative study between the proposed method and several state-of-the-art ones shows that the former is accurate and reliable and may provide a more useful and reliable tool for quality assurance in the industrial and manufacturing sectors.
Open Access
Characterisation of a cold atmospheric pressure plasma torch for medical applications: demonstration of device safety
(MDPI, 2021-12-14) Bennett, Adam; Urayama, Takuya; Papangelis, Konstantinos; Yuen, Peter W. T.; Yu, Nan
The safety and effectiveness of plasma devices are of crucial importance for medical applications. This study presents the novel design of an atmospheric plasma torch (SteriPlas) and its characterisation. The SteriPlas was characterised to ascertain whether it is safe for application on human skin. The emission spectrum discharged from the SteriPlas was shown to be the same as the emission from the MicroPlaSter Beta. The UV emitted from the SteriPlas was measured, and the effective irradiance was calculated. The effective irradiance enabled the determination of the maximum UV exposure limits, which were shown to be over two hours: significantly longer than the current two-minute treatment time. The use of an extraction system with a higher flow rate appears to reduce slightly the effective irradiance at the treatment area. The NOx and ozone emissions were recorded for both SteriPlas configurations. The NOx levels were shown to be orders of agnitude lower than their safety limits. The ozone emissions were shown to be safe 25 mm from the SteriPlas cage. A discussion of how safety standards differ from one regulatory body to another is given.
Open Access
CHIMES: An enhanced end-to-end Cranfield hyperspectral image modelling and evaluation system
(2020-02) Zahidi, Usman A.; Yuen, Peter W. T.; James, David B.
Hyperspectral remote sensing enables establishing semantics from an image by providing spectral details used for differentiating materials. The airborne/satellite setup for remote sensing are typically expensive in terms of time and cost-effectiveness. It is therefore important to predict performance of such systems as a precursor. Hyperspectral scene simulation is a technique which allows the detailed spatial and spectral information of a natural scene to be reconstructed without the need for expensive and time-consuming airborne/spaceborne image acquisition systems. It helps in predicting the potential performance of airborne/satellite systems, moreover, it enables varying atmospheric conditions, estimating degradation in sensor performance to provide better uncertainty analysis and traceability, performance analysis of data processing algorithms and counter-measures/camouflage assessment in military applications. Digital Imaging Remote Sensing Image Generation (DIRSIG) developed by Rochester Institute of Technology and Camoflauge Electro-Optic Simulator (CameoSim) by Lockheed Martin are the two earliest research and commercial products, respectively, that incorporate hyperspectral rendering for accurate physicsbased radiance estimation. Although CameoSim is a well-established Scene simulator, however it does not support volumetric scattering and localised adjacency model. DIRSIG has provided support form these features in newly developed version called DIRSIG5. Due to export control restriction it is typically not possible to access these simulators, hence motivates development of inhouse scene simulator. This thesis summarises the research which constitutes part of the deliverable under the DSTL R-Cloud project for the establishment of an in-house HSI scene simulator, which is known as the Cranfield Hyperspectral Image Modelling and Evaluation System (CHIMES). CHIMES is a physicsbased rendering enabled simulator and the main concept follows directly the radiative transfer (RT) big equation, with some components adopted from DIRSIG and CameoSim etc. The goal of the present research has been set and the work has been progressed in the following manner: • The establishment of CHIMES from scratch; • Validation of CHIMES through direct comparison with commercial-off-the-shelf (COTS) simulator such as CameoSim (CS); • Enhancement of CHIMES over the COTS simulator (e.g. CS) to include automatic in-scene atmospheric parametrisation, localised adjacency-effect model and volumetric scattering to achieve a more realistic scene simulation particularly for the rugged terrain; • To propose methods on how difficult issues such as shadows can be mitigated in scene simulation. This thesis summarises the work performed as according to the above 4 objectives with main results as follows: • CHIMES has been shown to reproduce the scene simulation performed by a COTS simulator (e.g. CameoSim) under various atmospheric conditions. • An automatic atmosphere parameterisation search algorithm has been shown to be effective to allow the simulation of the scene without the need of repeated trial and error atmospheric parameter adjustments. • Two adjacency models: the Background One-Spectra Adjacency Effect Model (BOAEM) and the Texture-Spectra Incorporated Adjacency Effect Model (TIAEM) have been developed under this work. The BOAEM is somewhat similar to that adopted in CS with a distinctive feature of volumetric scattering, however, the TIAEM is a terrain dependence adjacency which is much more sophisticated. It has been shown that at high altitude scene, TIAEM performs better than the BOAEM by 6.0% and by 10.0% better than CameoSim particularly in the 2D geometric simulation, in terms of `1-norm error. In the lower altitude scene, BOAEM performs better than both TIAEM and CameoSim by 22.0% and 16%. In a 3D scene (i.e. terrain with Digital Elevation Model (DEM)) with sensor at lower altitude CameoSim error raises by 5 times compared to GT. BOAEM still performs better than TIAEM by a similar 22% `1-norm error. • A means for assessing the shadowed pixels of the scene has been proposed and the validation of the model requires more comprehensive ground truth (GT) data which will be performed in the future research. Most of the above results have been published in three journal papers as part of the contributions towards the HSI research community
Open Access
Comparative noise performance of a coded aperture spectral imager
(SPIE, 2016-10) Piper, James; Yuen, Peter W. T.; Godfree, Peter; Ding, Mengjia; Soori, Umair; Selvagumar, Senthurran; James, David
Novel types of spectral sensors using coded apertures may offer various advantages over conventional designs, especially the possibility of compressive measurements that could exceed the expected spatial, temporal or spectral resolution of the system. However, the nature of the measurement process imposes certain limitations, especially on the noise performance of the sensor. This paper considers a particular type of coded-aperture spectral imager and uses analytical and numerical modelling to compare its expected noise performance with conventional hyperspectral sensors. It is shown that conventional sensors may have an advantage in conditions where signal levels are high, such as bright light or slow scanning, but that coded-aperture sensors may be advantageous in low-signal conditions
Open Access
Design and simulation of compressive snapshot multispectral imaging system
(2018-12) Ding, Mengjia; Yuen, Peter W. T.
Compressive Snapshot Spectral Imaging combines compressive sensing and snapshot spectral imaging (SSI) for restoring the image of the scene in both spatial and spectral contexts by using only a fewer number of sampling measurements of the captured image under the sparsity assumption. SSI is often realised through a coded aperture mask together with a single dispersive element as the main spatial modulator to implement compressive sampling. As one of the representative frameworks in this field, Coded Aperture Snapshot Spectral Imagers (CASSI) has prototyped a low-cost, compact platform to achieve compressive snapshot spectral imaging in the recent decade. Active research in the field includes advanced de-compressive recovery algorithms and also the employment of more sophisticated optical hardware for the design of more robust SSI system. This research addresses more of the latter direction and it focuses on how the CASSI framework can be further developed for various applications such as magnetic resonance imaging for medical diagnosis, enhancement of radar imaging system, facial expression detection and recognition, digital signal processing with sparse structure in terms of image denoising, image super-resolution and image classification. This thesis presents a summary of the research conducted over the past 4 years about the basic property of the CASSI system, which leads to the development of the spectral tuneable SSI design proposed during the course of the PhD study. This new design utilises a Dual-Prism assembly to embed the capability of wavelength-tuning without physically changing its optical elements. This Dual-Prism CASSI (DP-CASSI) adapts to dynamic environments far better than all the CASSI types of imagers published in the open domain which only function for a fixed set of wavelengths. This piece of work has been vii accepted by journal papers for publication. Other contributions of this research has been the enhancement of the Single-Prism (SP-CASSI) architecture and to produce a snapshot system with less aberration and better image quality than that published in the open domain. Moreover, the thesis also provides information about optical design of four different types of CASSI with slightly in-depth analysis about their optical system constructions, optical evaluations of system structure and their dispersive capabilities as the background of this research. Then a more detailed description of the proposed DP-CASSI with respected to its design and performance evaluation particularly its dispersion characteristics and the effects of system resolutions, are given. System verifications were conducted through ray-tracing simulation in three-dimension visualisation environments and the spectral characteristics of the targets are compared with that of the ground truth. The spectral tuning of the proposed DP-CASSI is achieved by adjusting the air gap displacement of dual-prism assembly. Typical spectral shifts of about 5 nm at 450 mm and 10 nm at 650 nm wavelength have been achieved in the present design when the air gap of the dual-prism is changed from 3.44 mm to 5.04 mm. The thesis summaries the optical designs, the performance and the pros and cons of the DP-CASSI system
Open Access
Design of a tunable snapshot multispectral imaging system through ray tracing simulation
(MDPI, 2019-01-05) Ding, Mengjia; Yuen, Peter W. T.; Piper, Jonathan; Godfree, Peter; Chatterjee, Ayan; Zahidi, Usman; Selvagumar, Senthurran; James, David; Richardson, Mark A.
Research on snapshot multispectral imaging has been popular in the remote sensing community due to the high demands of video-rate remote sensing system for various applications. Existing snapshot multispectral imaging techniques are mainly of a fixed wavelength type, which limits their practical usefulness. This paper describes a tunable multispectral snapshot system by using a dual prism assembly as the dispersion element of the coded aperture snapshot spectral imagers (CASSI). Spectral tuning is achieved by adjusting the air gap displacement of the dual prism assembly. Typical spectral shifts of about 1 nm at 400 nm and 12 nm at 700 nm wavelength have been achieved in the present design when the air-gap of the dual prism is changed from 4.24 mm to 5.04 mm. The paper outlines the optical designs, the performance, and the pros and cons of the dual-prism CASSI (DP-CASSI) system. The performance of the system is illustrated by TraceProTM ray tracing, to allow researchers in the field to repeat or to validate the results presented in this paper.
Open Access
Design of single prism coded aperture snapshot spectral imager using ray tracing simulation
(IEEE, 2019-03-07) Ding, Mengjia; Yuen, Peter W. T.; Richardson, Mark A.
Novel imaging systems published in the literature mostly concern with the performance of the final stage of the designed system which normally accompanies with a brief description of the system configuration only. Other information, such as how the system was optimized and the methodology adopted for improving them to their final stage are heavily lacking in the open domain. This paper addresses this issue by providing a guide for the modeling of compressive imaging based on Single Disperser Coded Aperture Snapshot Spectral Imaging (SD-CASSI), with focuses on the optimization of the dispersion capability, the reduction of spatial and chromatic aberrations for enhancing the performance of the SD-CASSI. As an example the system is designed for a numerical aperture of 0.125, 0.3% distortion at central wavelength 587.56 nm, and 32 spectral bands with a spatial resolution of 13 μm. The system was simulated by ray tracing program TracePro.
Open Access
Detection of psychological stress using a hyperspectral imaging technique
(IEEE, 2014-10-09) Chen, Tong; Yuen, Peter W. T.; Richardson, Mark A.; Liu, Guangyuan; She, Zhishun
The detection of stress at early stages is beneficial to both individuals and communities. However, traditional stress detection methods that use physiological signals are contact-based and require sensors to be in contact with test subjects for measurement. In this paper, we present a method to detect psychological stress in a non-contact manner using a human physiological response. In particular, we utilize a hyperspectral imaging (HSI) technique to extract the tissue oxygen saturation (StO2) value as a physiological feature for stress detection. Our experimental results indicate that this new feature may be independent from perspiration and ambient temperature. Trier Social Stress Tests (TSSTs) on 21 volunteers demonstrated a significant difference $p\< 0.005$ and a large practical discrimination (d 1/4 1.37) between normalized baseline and stress StO2 levels. The accuracy for stress recognition from baseline using a binary classifier was 76.19 and 88.1 percent for the automatic and manual selections of the classifier threshold, respectively. These results suggest that the StO2 level could serve as a new modality to recognize stress at standoff distances.
Open Access
An end-to-end hyperspectral scene simulator with alternate adjacency effect models and its comparison with cameoSim
(MDPI, 2019-12-24) Zahidi, Usman A.; Yuen, Peter W. T.; Piper, Jonathan; Godfree, Peter S.
In this research, we developed a new rendering-based end to end Hyperspectral scene simulator CHIMES (Cranfield Hyperspectral Image Modelling and Evaluation System), which generates nadir images of passively illuminated 3-D outdoor scenes in Visible, Near Infrared (NIR) and Short-Wave Infrared (SWIR) regions, ranging from 360 nm to 2520 nm. MODTRAN TM (MODerate resolution TRANsmission), is used to generate the sky-dome environment map which includes sun and sky radiance along with the polarisation effect of the sky due to Rayleigh scattering. Moreover, we perform path tracing and implement ray interaction with medium and volumetric backscattering at rendering time to model the adjacency effect. We propose two variants of adjacency models, the first one incorporates a single spectral albedo as the averaged background of the scene, this model is called the Background One-Spectra Adjacency Effect Model (BOAEM), which is a CameoSim like model created for performance comparison. The second model calculates background albedo from a pixel’s neighbourhood, whose size depends on the air volume between sensor and target, and differential air density up to sensor altitude. Average background reflectance of all neighbourhood pixel is computed at rendering time for estimating the total upwelled scattered radiance, by volumetric scattering. This model is termed the Texture-Spectra Incorporated Adjacency Effect Model (TIAEM). Moreover, for estimating the underlying atmospheric condition MODTRAN is run with varying aerosol optical thickness and its total ground reflected radiance (TGRR) is compared with TGRR of known in-scene material. The Goodness of fit is evaluated in each iteration, and MODTRAN’s output with the best fit is selected. We perform a tri-modal validation of simulators on a real hyperspectral scene by varying atmospheric condition, terrain surface models and proposed variants of adjacency models. We compared results of our model with Lockheed Martin’s well-established scene simulator CameoSim and acquired Ground Truth (GT) by Hyspex cameras. In clear-sky conditions, both models of CHIMES and CameoSim are in close agreement, however, in searched overcast conditions CHIMES BOAEM is shown to perform better than CameoSim in terms of ℓ1 -norm error of the whole scene with respect to GT. TIAEM produces better radiance shape and covariance of background statistics with respect to Ground Truth (GT), which is key to good target detection performance. We also report that the results of CameoSim have a many-fold higher error for the same scene when the flat surface terrain is replaced with a Digital Elevation Model (DEM) based rugged one.
Open Access
Endmember learning with k-means through SCD model in hyperspectral scene reconstructions
(MDPI, 2019-11-15) Chatterjee, Ayan; Yuen, Peter W. T.
This paper proposes a simple yet effective method for improving the efficiency of sparse coding dictionary learning (DL) with an implication of enhancing the ultimate usefulness of compressive sensing (CS) technology for practical applications, such as in hyperspectral imaging (HSI) scene reconstruction. CS is the technique which allows sparse signals to be decomposed into a sparse representation “a” of a dictionary Du" role="presentation" style="max-height: none; display: inline; line-height: normal; word-spacing: normal; overflow-wrap: normal; white-space: nowrap; float: none; direction: ltr; max-width: none; min-width: 0px; min-height: 0px; border-width: 0px; border-style: initial; position: relative;">Du . The goodness of the learnt dictionary has direct impacts on the quality of the end results, e.g., in the HSI scene reconstructions. This paper proposes the construction of a concise and comprehensive dictionary by using the cluster centres of the input dataset, and then a greedy approach is adopted to learn all elements within this dictionary. The proposed method consists of an unsupervised clustering algorithm (K-Means), and it is then coupled with an advanced sparse coding dictionary (SCD) method such as the basis pursuit algorithm (orthogonal matching pursuit, OMP) for the dictionary learning. The effectiveness of the proposed K-Means Sparse Coding Dictionary (KMSCD) is illustrated through the reconstructions of several publicly available HSI scenes. The results have shown that the proposed KMSCD achieves ~40% greater accuracy, 5 times faster convergence and is twice as robust as that of the classic Spare Coding Dictionary (C-SCD) method that adopts random sampling of data for the dictionary learning. Over the five data sets that have been employed in this study, it is seen that the proposed KMSCD is capable of reconstructing these scenes with mean accuracies of approximately 20–500% better than all competing algorithms adopted in this work. Furthermore, the reconstruction efficiency of trace materials in the scene has been assessed: it is shown that the KMSCD is capable of recovering ~12% better than that of the C-SCD. These results suggest that the proposed DL using a simple clustering method for the construction of the dictionary has been shown to enhance the scene reconstruction substantially. When the proposed KMSCD is incorporated with the Fast non-negative orthogonal matching pursuit (FNNOMP) to constrain the maximum number of materials to coexist in a pixel to four, experiments have shown that it achieves approximately ten times better than that constrained by using the widely employed TMM algorithm. This may suggest that the proposed DL method using KMSCD and together with the FNNOMP will be more suitable to be the material allocation module of HSI scene simulators like the CameoSim package
Open Access
Engaging students for the learning and assessment of the advanced computer graphics module using the latest technologies
(inScience Press, 2017-07) Liu, Yonghuai; Yang, Longzhi; Han, Jiwan; Lu, Bin; Yuen, Peter W. T.; Zhao, Yitian; Song, Ran
The advanced computer graphics has been one of the most basic and landmark modules in the field of computer science. It usually covers such topics as core mathematics, lighting and shading, texture mapping, colour and depth, and advanced modeling. All such topics involve mathematics for object modeling and transformation, and programming for object visualization and interaction. While some students are not as good in either mathematics or programming, it is usually a challenge to teach computer graphics to these students effectively. This is because it is difficult for students to link mathematics and programming with what they used to see in video games and the TV advertisements for example and thus they can easily be put off. In this paper, we investigate how the latest technologies can help alleviate the teaching and learning tasks. Instead of selecting the low level programming languages for demonstration and assignment such as Java, Java 3D, C++, or OpenGL, we selected Three.js, which is one of the latest and freely accessible 3D graphics libraries. It has a unique advantage that it provides a seamless interface between the main stream web browsers and 2D/3D graphics. The developed code can be run on a web browser such as Firefox, Chrome, or Safari for testing, debugging and visualization without code changing. The unique design patterns and objectives of Three.js can be very attractive to third party software houses to develop auxiliary functions, methods and tutorials and to make them freely available for the public. Such a unique property of Three.js and its widely available supporting resources are especially helpful to engage students, inspire their learning and facilitate teaching. To evaluate the effectiveness for using Three.js in teaching computer graphics we have set up an assignment for scene modeling in the last 4 years with focuses on the quality of the simulated scene (50%) and the quality of the assignment report (50%). We have evaluated different assessment forms of the module that we taught in the last four years: in 2013-2014 the module consisted of 20% assignment and 80% exam based on Java 3D; in 2014-2015 the same proportion of assignment/exam but based on WebGL, in 2015-2016 the module was 50-50% of assignment and exam but based on Three.js; and in this year the module is 100% assignment based on Three.js. The effectiveness of the module delivery has been evaluated both qualitatively and quantitatively from five aspects: a) average marks of students, b) moderator report, c) module evaluation questionnaire, d) external examiner’s comments and e) examination board recommendations. The results have shown that Three.js is indeed more successful in engaging students for learning and the 100% assignment assessment enables students to focus more on the design and development. This four year result is really encouraging to us as an educational institute to embrace the latest technologies for the delivery of such challenging modules as computer graphics and machine learning.
Open Access
Enhanced hyperspectral sharpening through improved relative spectral response characteristic (R-SRC) estimation for long-range surveillance applications
(MDPI, 2024-05-29) Yuen, Peter W. T.; Piper, Jonathan; Yuen, Catherine; Cakir, Mehmet
The fusion of low-spatial-resolution hyperspectral images (LRHSI) with high-spatial-resolution multispectral images (HRMSI) for super-resolution (SR), using coupled non-negative matrix factorization (CNMF), has been widely studied in the past few decades. However, the matching of spectral characteristics between the LRHSI and HRMSI, which is required before they are jointly factorized, has rarely been studied. One objective of this work is to study how the relative spectral response characteristics (R-SRC) of the LRHSI and HRMSI can be better estimated, particularly when the SRC of the latter is unknown. To this end, three variants of enhanced R-SRC algorithms were proposed, and their effectiveness was assessed by applying them for sharpening data using CNMF. The quality of the output was assessed using the L1-norm-error (L1NE) and receiver operating characteristics (ROC) of target detections performed using the adaptive coherent estimator (ACE) algorithm. Experimental results obtained from two subsets of a real scene revealed a two- to three-fold reduction in the reconstruction error when the scenes were sharpened by the proposed R-SRC algorithms, in comparison with Yokoya’s original algorithm. Experiments also revealed that a much higher proportion (by one order of magnitude) of small targets of 0.015 occupancy in the LRHSI scene could be detected by the proposed R-SRC methods compared with the baseline algorithm, for an equal false alarm rate. These results may suggest the possibility of SR to allow long-range surveillance using low-cost HSI hardware, particularly when the remaining issues of the occurrence of large reconstruction errors and comparatively higher false alarm rate for ‘rare’ species in the scene can be understood and resolved in future research.
Open Access
Extinction and permanence of the predator-prey system with general functional response and impulsive control
(Elsevier, 2020-06-22) Liu, Juan; Hu, Jie; Yuen, Peter W. T.
Traditional approach for modelling the evolution of populations in the predator-prey ecosystem has commonly been undertaken using specific impulsive response function, and this kind of modelling is applicable only for a specific ecosystem under certain environ- mental situations only. This paper attempts to fill the gap by modelling the predator-prey ecosystem using a ‘generalized’ impulsive response function for the first time. Different from previous research, the present work develops the modelling for an integrated pest management (IPM) especially when the stocking of predator (natural enemy) and the har- vesting of prey (pest) occur impulsively and at different instances of time. The paper firstly establishes the sufficient conditions for the local and the global stabilities of prey eradica- tion periodic solution by applying the Floquet theorem of the Impulsive different equation and small amplitude perturbation under a ‘generalized’ impulsive response function. Sub- sequently the sufficient condition for the permanence of the system is given through the comparison techniques. The corollaries of the theorems that are established by using the ‘general impulsive response function’ under the locally asymptotically stable condition are found to be in excellent agreement with those reported previously. Theoretical results that are obtained in this work is then validated by using a typical impulsive response func- tion (Holling type-II) as an example, and the outcome is shown to be consistent with the previously reported results. Finally, the implication of the developed theories for practical pest management is illustrated through numerical simulation. It is shown that the elim- ination of either the preys or the pest can be effectively deployed by making use of the theoretical model established in this work. The developed model is capable to predict the population evolutions of the predator-prey ecosystem to accommodate requirements such as: the combinations of the biological control, chemical control, any functional response function, the moderate impulsive period, the harvest rate for the prey and predator pa- rameter and the incremental stocking of the predator parameter
Open Access
Fast implementation of two-dimensional singular spectrum analysis for effective data classification in hyperspectral imaging
(Elsevier, 2017-05-12) Zabalza, Jaime; Qing, Chunmei; Yuen, Peter W. T.; Sun, Genyun; Zhao, Huimin; Ren, Jinchang
Although singular spectrum analysis (SSA) has been successfully applied for data classification in hyperspectral remote sensing, it suffers from extremely high computational cost, especially for 2D-SSA. As a result, a fast implementation of 2D-SSA namely F-2D-SSA is presented in this paper, where the computational complexity has been significantly reduced with a rate up to 60%. From comprehensive experiments undertaken, the effectiveness of F-2D-SSA is validated producing a similar high-level of accuracy in pixel classification using support vector machine (SVM) classifier, yet with a much reduced complexity in comparison to conventional 2D-SSA. Therefore, the introduction and evaluation of F-2D-SSA completes a series of studies focused on SSA, where in this particular research, the reduction in computational complexity leads to potential applications in mobile and embedded devices such as airborne or satellite platforms.
Open Access
Generative detect for occlusion object based on occlusion generation and feature completing
(Elsevier, 2021-06-17) Xu, Can; Yuen, Peter W. T.; Lang, Wenxi; Xin, Rui; Mao, Kaichen; Jiang, Haiyan
Detecting the object with external occlusion has always been a hot topic in computer version, while its accuracy is always limited due to the loss of original object information and increase of new occlusion noise. In this paper, we propose a occluded object detection algorithm named GC-FRCN (Generative feature completing Faster RCNN), which consists of the OSGM (Occlusion Sample Generation Module) and OSIM (Occlusion Sample Inpainting Module). Specifically, the OSGM mines and discards the feature points with high category response on the feature map to enhance the richness of occlusion scenes in the training data set. OSIM learns an implicit mapping relationship from occluded feature map to real feature map adversarially, which aims at improving feature quality by repair the noisy object feature. Extensive experiments and ablation studies have been conducted on four different datasets. All the experiments demonstrate the GC-FRCN can effectively detect objects with local external occlusion and has good robustness for occlusion at different scales.
Open Access
Hyperspectral image reconstruction using multi-colour and time-multiplexed LED illumination
(Elsevier, 2019-05-06) Tschannerl, Julius; Ren, Jinchang; Zhao, Huimin; Kao, Fu-Jen; Marshall, Stephen; Yuen, Peter W. T.
The rapidly rising industrial interest in hyperspectral imaging (HSI) has generated an increased demand for cost effective HSI devices. We are demonstrating a mobile and low-cost multispectral imaging system, enabled by time-multiplexed RGB Light Emitting Diodes (LED) illumination, which operates at video framerate. Critically, a deep Multi-Layer Perceptron (MLP) with HSI prior in the spectral range of 400–950 nm is trained to reconstruct HSI data. We incorporate regularisation and dropout to compensate for overfitting in the largely ill-posed problem of reconstructing the HSI data. The MLP is characterised by a relatively simple design and low computational burden. Experimental results on 51 objects of various references and naturally occurring materials show the effectiveness of this approach in terms of reconstruction error and classification accuracy. We were also able to show that utilising additional colour channels to the three R, G and B channels adds increased value to the reconstruction.
Open Access
Illumination invariance and shadow compensation via spectro-polarimetry technique
(Maney Publishing, 2013-06-28T00:00:00Z) Ibrahim, I.; Yuen, Peter W. T.; Hong, K.; Chen, T.; Soori, U.; Jackman, J.; Richardson, Mark A.
A major problem for obtaining target reflectance via hyperspectral imaging systems is the presence of illumination and shadow effects. These factors are common artefacts, especially when dealing with a hyperspectral imaging system that has sensors in the visible to near infrared region. This region is known to have highly scattered and diffuse radiance which can modify the energy recorded by the imaging system. Shadow effect will lower the target reflectance values due to the small radiant energy impinging on the target surface. Combined with illumination artefacts, such as diffuse scattering from the surrounding targets, background or environment, the shape of the shadowed target reflectance will be altered. In this study we propose a new method to compensate for illumination and shadow effects on hyperspectral imageries by using a polarization technique. This technique, called spectro-polarimetry, estimates the direct and diffuse irradiance based on two images, taken with and without a polarizer. The method is evaluated using a spectral similarity measure, angle and distance metric. The results of indoor and outdoor tests have shown that using the spectro-polarimetry technique can improve the spectral constancy between shadow and full illumination spectra.
Open Access
Image dehazing based on partitioning reconstruction and entropy-based alternating fast-weighted guided filters
(SPIE, 2017-05-26) Wang, Yuanyu; Yuen, Peter W. T.
A robust image dehazing algorithm based on the first-order scattering of the image degradation model is proposed. In this work, there are three contributions toward image dehazing: (i) a robust method for assessing the global irradiance from the most hazy-opaque regions of the imagery is proposed; (ii) more detailed depth information of the scene can be recovered through the enhancement of the transmission map using scene partitions and entropy-based alternating fast-weighted guided filters; and (iii) crucial model parameters are extracted from in-scene information. This paper briefly outlines the principle of the proposed technique and compares the dehazed results with four other dehazing algorithms using a variety of different types of imageries. The dehazed images have been assessed through a quality figure-of-merit, and experiments have shown that the proposed algorithm effectively removes haze and has achieved a much better quality of dehazed images than all other state-of-the-art dehazing methods employed in this work.
Open Access
An integrated sensor platform for food reassurance
(Institute of Food Technology and John Wiley, 2020-01-28) Povey, Malcolm J. W.; Blakemore, Ian; Al-Lashi, Raied; Tiede, Christian; Zou, Xiaobo; Huang, Xiaowei; Shi, Jiyong; Yuen, Peter W. T.
Given the increasing pressures on the global food system to produce more from the planet’s diminishing resources, the need to produce food that is safe to consume and reflects the information declared on pack is becoming ever more difficult to achieve as shown by the recent ‘horsemeat’ scandal. The salutary lesson that emerged from this latest food scandal was the vulnerability of the highly complex food supply networks that we rely on. For example, the wafers in the popular KitKat are ‘glued’ together with plaster-of-Paris which is sourced from India. These convoluted food networks increase the potential for product contamination through either physical, chemical or microbial means, with potentially life threatening consequences. Recent issues include the 2005 Sudan Red dye scandal, in which food products were contaminated with the potentially carcinogenic dye, and the 2013 worldwide recall of dairy products manufactured by Fonterra, because whey products suspected of being contaminated with botulism-causing bacteria were found during safety tests. It is important to remain vigilant in identifying and solving emerging challenges in global food networks to ensure a high level of consumer trust in the supply chain and food sold for consumption.
Open Access
Joint bilateral filtering and spectral similarity-based sparse representation: A generic framework for effective feature extraction and data classification in hyperspectral imaging
(Elsevier, 2017-10-10) Qiao, Tong; Yang, Zhijing; Ren, Jinchang; Yuen, Peter W. T.; Zhao, Huimin; Sun, Genyun; Marshall, Stephen; Beneditksson, Jon Atli
Classification of hyperspectral images (HSI) has been a challenging problem under active investigation for years especially due to the extremely high data dimensionality and limited number of samples available for training. It is found that hyperspectral image classification can be generally improved only if the feature extraction technique and the classifier are both addressed. In this paper, a novel classification framework for hyperspectral images based on the joint bilateral filter and sparse representation classification (SRC) is proposed. By employing the first principal component as the guidance image for the joint bilateral filter, spatial features can be extracted with minimum edge blurring thus improving the quality of the band-to-band images. For this reason, the performance of the joint bilateral filter has shown better than that of the conventional bilateral filter in this work. In addition, the spectral similarity-based joint SRC (SS-JSRC) is proposed to overcome the weakness of the traditional JSRC method. By combining the joint bilateral filtering and SS-JSRC together, the superiority of the proposed classification framework is demonstrated with respect to several state-of-the-art spectral-spatial classification approaches commonly employed in the HSI community, with better classification accuracy and Kappa coefficient achieved.