Mid Sweden University

To ensure traffic safety during winter, chemical agents are typically used for de-icing and anti-icing. Smarter and more precise dispersion of chemicals, which considers local variations in concentration, has the potential to reduce the total amount applied. This paper presents a study of an optical chemometric sensor capable of measuring the NaCl concentration and the weight of the dispersed solution per square meter. The experiment was conducted in an indoor environment, where seven solutions of tap water and NaCl were poured onto a diffuse surface made of burned clay. Short-wave infrared light was illuminated onto the surface, and the light was diffusely reflected into a spectrometer after passing through the liquid layer twice. Absorption in the liquid layer alone can be extracted by subtracting the background and further modeled using Beer-Lambert's law. Both the concentration of NaCl and the amount of liquid can be computed by fitting an overdetermined equation system. Experimental results show a strong correlation between actual and computed concentrations, as well as between actual and computed liquid quantities. Suppression of ambient light, spectral variations of asphalt, harsh environments, dynamic range, and signal-to-noise ratio are among the challenges for outdoor chemometric sensing of asphalt pavements. 

Elemental mapping images can be achieved through step scanning imaging using pinholeopticsor microporeoptics(MPO),oralternativelybyfull-field X-ray fluorescenceimaging (FF-XRF). X-ray optics for FF-XRF canbe manufacturedwith different micro-channelgeometries such as square, hexagonal or circular channels. Each optic geometry creates different imaging artefacts. Square-channel MPOs generate a high intensity central spot due to two reflections via orthogonal channel walls inside a single channel, which is the desirable part for image formation, and two perpendicular lines forming a cross due to reflections in one plane only. Thus, we have studied the performance of a square-channel MPO in an FF-XRF imaging system. The setup consists of a commercially available MPO provided by Photonis and a Timepix3 readout chip with a silicon detector. Imaging of fluorescence from small metal particles has been used to obtain the point spreadfunction(PSF) characteristics. The transmissionthroughMPO channelsand variation of the critical reflection angle are characterized by measurements of fluorescence from copper and titanium metal fragments. Since the critical angle of reflection is energy dependent, the cross-arm artefacts will affect the resolution differently for different fluorescence energies. It is possible to identify metal fragments due to the form of the PSF function. The PSF function can be further characterized using a Fourier transform to suppress diffuse background signals in the image.

In this work, a multi-exposure method is proposed to increase the dynamic range (DR) of hyperspectral imaging using an InGaAs-based short-wave infrared (SWIR) hyperspectral line camera. Spectral signatures of materials were captured for scenarios in which the DR of a scene was greater than the DR of a line camera. To demonstrate the problem and test the proposed multi-exposure method, plastic detection in food waste and polymer sorting were chosen as the test application cases. The DR of the hyperspectral camera and the test samples were calculated experimentally. A multi-exposure method is proposed to create high-dynamic-range (HDR) images of food waste and plastic samples. Using the proposed method, the DR of SWIR imaging was increased from 43 dB to 73 dB, with the lowest allowable signal-to-noise ratio (SNR) set to 20 dB. Principal Component Analysis (PCA) was performed on both HDR and non-HDR image data from each test case to prepare the training and testing data sets. Finally, two support vector machine (SVM) classifiers were trained for each test case to compare the classification performance of the proposed multi-exposure HDR method against the single-exposure non-HDR method. The HDR method was found to outperform the non-HDR method in both test cases, with the classification accuracies of 98% and 90% respectively, for the food waste classification, and with 95% and 35% for the polymer classification. 

Hyperspectral remote sensing is known to suffer from wavelength bands blocked by atmospheric gases. Short-wave infrared hyperspectral imaging at in situ installations is shown to be affected by water vapour even if the pathlength of light through air is only hundreds of centimetres. This impact is especially noticeable with large variations of relative humidity, the coefficient of variation reaching 5 % in our test case. Using repeated calibrations of imaging system at the same relative humidity as in the measurement, we were able to reduce the coefficient of variation to 1 %. The measurement variations are also shown to induce significant error in material classification. Polymer type identification was selected as the test case for material classification. The measurement variations due to the change in relative humidity are shown to result in 20 % classification error at its minimum. With repeated calibrations or by eliminating the most affected wavelength bands from measurements, we were able to reduce the classification error to less than 1 %. Such improvement of measurement and classification precision may be important for industrial applications such as waste sorting, polymer classification etc. 

In this work, Sinewave Pulse Width Modulation (SPWM) of light sources is proposed to be used for spectral imaging and vibrational spectroscopy. Three Light Emitting Diodes (LEDs) transmit SPWM light at different sinewave frequencies and at 1050, 1300 and 1550 nm respectively. A video camera records the active illumination, reflected in the surface to be analyzed and simultaneously for all three LEDs. A similar demodulation method as for lock-in amplifiers is used to demodulate every single pixel into three spectral channels. Experimental results show that this material imaging analyzer, called MIA, can be used for vibrational spectroscopy such that snow, ice, water and plastic are classified from their corresponding spectral signatures at an overall success rate of 92%. Spectral signatures in this particular case are generated from sensing reflection of light at three wavelengths. MIA must thus be configured with LEDs having wavelengths suitable for the particular application. The response of the MIA light sources and the indium gallium arsenide InGaAs detector is close to perfectly linear. Peak signal to noise ratio PSNR was measured to be 54 dB. 

A three-dimensional (3D) reconstruction method and multi-camera calibration using multiple artificial reference markers have been used for precise volumetric surveillance of fast-flying objects. The method uses a two-layer 3D reconstruction that integrates two multi-camera stereo-nodes. The fields of view of stereo nodes are directed at an acute angles to each other to provide greater coverage with the given constraints and to determine the flight characteristics of objects in 3D. The object’s flight reconstruction includes a “rough” estimation of its positions relative to selected artificial reference points in both stereo nodes separately and subsequent “refinement” of calculated positions. In this paper, we describe the proposed method and calibration technique, using a multi-camera system to measure object characteristics in 3D. The proposed method applies to volumetric surveillance in situations where it is necessary to count, track, and analyze the activities of flying objects, especially birds, using high spatial resolution.

In this paper, we present a hyper-spectral imaging system and practical calibration procedure using a low-cost calibration reference made of polytetrafluoroethylene. The imaging system includes a hyperspectral camera and an active source of illumination with a variable spectral distribution of intensity. The calibration reference is used to measure the relative reflectance of any material surface independent of the spectral distribution of light and camera sensitivity. Winter road conditions are taken as a test application, and several spectral images of snow, icy asphalt, dry as-phalt, and wet asphalt were made at different exposure times using different illumination spectra. Graphs showing measured relative reflectance for different road conditions support the conclusion that measurements are independent of illumination. Principal component analysis of the acquired spectral data for road conditions shows well separated data clusters, demonstrating the system’s suitability for material classification. 

A multi-camera dome consists of number ofcameras arranged in layers to monitor a hemisphere aroundits center. In volumetric surveillance,a 3D space is required tobemonitoredwhich can be achievedby implementing numberof multi-camera domes. A monitoring height is consideredas a constraint to ensure full coverage of the space belowit. Accordingly, the multi-camera dome can be redesignedinto a cylinder such that each of its multiple layers hasdifferent coverage radius. Minimum monitoring constraintsshould be met at all layers. This work is presenting a costoptimized design for the multi-camera dome that maximizesits coverage. The cost per node and number of squaremetersper dollar of multiple configurations are calculated using asearch space of cameras and considering a set of monitoring and coverage constraints. The proposed design is costoptimized per node and provides more coverage as compared to the hemispherical multi-camera dome.

3D object recognition has been a cutting-edge research topic since the popularization of depth cameras. These cameras enhance the perception of the environment and so are particularly suitable for autonomous robot navigation applications. Advanced deep learning approaches for 3D object recognition are based on complex algorithms and demand powerful hardware resources. However, autonomous robots and powered wheelchairs have limited resources, which affects the implementation of these algorithms for real-time performance. We propose to use instead a 3D voxel-based extension of the 2D histogram of oriented gradients (3DVHOG) as a handcrafted object descriptor for 3D object recognition in combination with a pose normalization method for rotational invariance and a supervised object classifier. The experimental goal is to reduce the overall complexity and the system hardware requirements, and thus enable a feasible real-time hardware implementation. This article compares the 3DVHOG object recognition rates with those of other 3D recognition approaches, using the ModelNet10 object data set as a reference. We analyze the recognition accuracy for 3DVHOG using a variety of voxel grid selections, different numbers of neurons (N-h ) in the single hidden layer feedforward neural network, and feature dimensionality reduction using principal component analysis. The experimental results show that the 3DVHOG descriptor achieves a recognition accuracy of 84.91% with a total processing time of 21.4 ms. Despite the lower recognition accuracy, this is close to the current state-of-the-art approaches for deep learning while enabling real-time performance.