Classification of Adulterated Particle Images in Coconut Oil Using Deep Learning Approaches

In the production of coconut oil for consumption, cleanliness and safety are the first priorities for meeting the standard in Thailand. The presence of color, sediment, or impurities is an important element that affects consumers’ or buyers’ decision to buy coconut oil. Coconut oil contains impurities that are revealed during the process of compressing the coconut pulp to extract the oil. Therefore, the oil must be filtered by centrifugation and passed through a fine filter. When the oil filtration process is finished, staff inspect the turbidity of coconut oil by examining the color with the naked eye and should detect only the color of the coconut oil. However, this method cannot detect small impurities, suspended particles that take time to settle and become sediment. Studies have shown that the turbidity of coconut oil can be measured by passing light through the oil and applying image processing techniques. This method makes it possible to detect impurities using a microscopic camera that photographs the coconut oil. This study proposes a method for detecting impurities that cause the turbidity in coconut oil using a deep learning approach called a convolutional neural network (CNN) to solve the problem of impurity identification and image analysis. In the experiments, this paper used two coconut oil impurity datasets, PiCO_V1 and PiCO_V2, containing 1000 and 6861 images, respectively. A total of 10 CNN architectures were tested on these two datasets to determine the accuracy of the best architecture. The experimental results indicated that the MobileNetV2 architecture had the best performance, with the highest training accuracy rate, 94.05%, and testing accuracy rate, 80.20%.

A SIFT Description Approach for Non-Uniform Illumination and Other Invariants

The new improvement keypoint description technique of image-based recognition for rotation, viewpoint and non-uniform illumination situations is presented. The technique is relatively simple based on two procedures, i.e., the keypoint detection and the keypoint description procedure. The keypoint detection procedure is based on the SIFT approach, Top-Hat filtering, morphological operations and average filtering approach. Where this keypoint detection procedure can segment the targets from uneven illumination particle images. While the keypoint description procedures are described and implemented using the Hu moment invariants. Where the central moments are being unchanged under image translations. The sensitivity, accuracy and precision rate of data sets were evaluated and compared. The data set are provided by color image database with variants uniform and non-uniform illumination, viewpoint and rotation changes. The evaluative results show that the approach is superior to the other SIFTs in terms of uniform illumination, non-uniform illumination and other situations. Additionally, the paper demonstrates the high sensitivity of 100%, high accuracy of 83.33% and high precision rate of 80.00%. Comparisons to other SIFT approaches are also included.

Particle/wall electroviscous effects at the micron scale: comparison between experiments, analytical and numerical models.

We report a experimental study of the motion of 1μm single particles interacting with functionalized walls at low and moderate ionic strengths conditions. The 3D particle’s trajectories were obtained by analyzing the diffracted particle images (point spread function). The studied particle/wall systems include negatively charged particles interacting with bare glass, glass covered with polyelectrolytes and glass covered with a lipid monolayer. In the low salt regime (pure water) we observed a retardation effect of the short-time diffusion coefficients when the particle interacts with a negatively charged wall; this effect is more severe in the perpendicular than in the lateral component. The decrease of the diffusion as a function of the particle-wall distance h was similar regardless the origin of the negative charge at the wall. When surface charge was screened or salt was added to the medium (10mM), the diffusivity curves recover the classical hydrodynamic behavior. Electroviscous theory based on the thin electrical double layer (EDL) approximation reproduces the experimental data except for small h. On the other hand, 2D numerical solutions of the electrokinetic equations showed good qualitative agreement with experiments. The numerical model also showed that the hydrodynamic and Maxwellian part of the electroviscous total drag tend to zero as h → 0 and how this is linked with the merging of both EDL’s at close proximity.

Research on 3D reconstruction method of wear particle dynamic image based on multi contour space mapping

Aiming at the problems of current image monitoring methods of lubrication oil wear particles, this paper designs and builds a dynamic monitoring system for oil wear particles based on microfluidic microscopic images. A contour-based 3D reconstruction method of debris particle images is proposed. The image sequences of rotating wear particles tracked by a single target are used as data, and the contour of the wear particle is extracted and the data is stored. The minimum area external rectangle method is used to correct the rotation of the particle images for the problem of deflection. And an algorithm based on cylindrical coordinate space conversion is used to convert the discrete contour point data into three-dimensional space. Complete the 3D model reconstruction of microfluidic wear particles. The ability to analyse wear particles in oil online monitoring technology is improved, which also shows new ideas for wear status monitoring and fault diagnosis technology.

On the determination of 3D position and orientation of spheroidal particles using defocusing and deep learning

Tracking the 3D position of tracer particles or small objects like cells or unicellular organisms in miniaturized lab-on-a-chip or biomedical devices is complicated since it is often not possible in these setups to use multi-camera approaches. Most successful single-camera approaches for these applications are based on holography or defocusing. Holographic methods have been used to track complex objects such has bacteria (Bianchi et al. (2019)) and even to estimate their orientation (Wang et al. (2016)). However, these methods require a complex and expensive experimental setup which is not always available in research laboratories. On the other hand, defocusing methods work with conventional microscopic optics, are easy to implement, and have shown excellent results in 3D PTV experiments (Qiu et al. (2019)). One main drawback is that they normally work only with spherical and mono-dispersed tracer particles. A defocusing method that has potential to measure non-spherical particles is the General Defocusing Particle Tracking (Barnkob and Rossi (2020)) which is based on pattern recognition and can be conceptually extended to more complex tasks by extending the reference library of particle images, including not only spherical particles at different depth positions, but also non-spherical particles at different orientations. However, whether this approach could work in practice is still unknown. First, is the information contained in simple defocused images sufficient to reconstruct depth and orientation of non-spherical particles, and eventually under which circumstances? Second, how to practically collect the labelled reference images?

Functional Analysis of the GPI Transamidase Complex by Screening for Amino Acid Mutations in Each Subunit

Glycosylphosphatidylinositol (GPI) anchor modification is a posttranslational modification of proteins that has been conserved in eukaryotes. The biosynthesis and transfer of GPI to proteins are carried out in the endoplasmic reticulum. Attachment of GPI to proteins is mediated by the GPI-transamidase (GPI-TA) complex, which recognizes and cleaves the C-terminal GPI attachment signal of precursor proteins. Then, GPI is transferred to the newly exposed C-terminus of the proteins. GPI-TA consists of five subunits: PIGK, GPAA1, PIGT, PIGS, and PIGU, and the absence of any subunit leads to the loss of activity. Here, we analyzed functionally important residues of the five subunits of GPI-TA by comparing conserved sequences among homologous proteins. In addition, we optimized the purification method for analyzing the structure of GPI-TA. Using purified GPI-TA, preliminary single particle images were obtained. Our results provide guidance for the structural and functional analysis of GPI-TA.

On the micro-PIV accuracy and reliability utilizing non-Gaussian particle images

Optical investigations of the dynamics of concentrated suspensions, such as in blood flows (Fitzgibbon et al. in Biophys J 108(10):2601–2608, 2015. http://doi/org/10.1016/j.bpj.2015.04.013) or slurry flows (Li et al. in Ocean Eng 163(October 2017):691–705, 2018. http://doi/org/10.1016/j.oceaneng.2018.06.046), are challenging due to reduced optical accessibility. Furthermore, the suspension particle image size can strongly deviate from the optimal particle image size for PIV measurements. Optical accessibility can be achieved by refractive index matching of surface labelled suspension particles. This results in particle images that are transparent in the particle image centre, but fluoresce at the particle image rim, resulting in ring-shaped particle images. In the present study, the influence of the particle image size on the cross-correlation result of such ring-shaped particle images is compared with Gaussian and plateau-shaped particle images. Particles of Gaussian image shape result from fully labelled particles with small image diameters and are commonly used in PIV measurements. Such particles are also utilized for the determination of the continuous phase velocities in the experimental part of the present study. With increasing image diameter, fully labelled particles are observed to assume plateau-shaped particle images. Monte Carlo simulations of synthetically generated images show that ring-shaped particle images have a superior behaviour, i.e. they assume a reduced displacement estimation error for noisy as well as for noise-free image data, compared to Gaussian and plateau-shaped particle images. This is also true for large particle image diameters when particle images are intersected at interrogation window borders or when different values of nonzero particle image displacements are considered. The detectability is similar for all three particle image shapes as long as particles do not intersect with the interrogation window border. Interestingly, for intersected particles of large image diameter, ring-shaped particle images show a slightly improved detectability compared to particle images of Gaussian and plateau shape. Furthermore, the detectability is insensitive against a nonzero particle image displacement. The usage of refractive index matched, ring-shaped particle images results in a good optical accessibility of the suspension. This allows to perform simultaneous cross-correlation evaluations on large ring-shaped particle images and fluid tracers with Gaussian particle images that are two orders of magnitude smaller compared to suspension particle images. Velocity measurements are taken on a suspension containing 5 vol% surface labelled, refractive index matched 60 $$\upmu \hbox {m}$$ μ m polymethylmethacrylate (PMMA) particles. Simultaneously, $$\upmu$$ μ PIV measurements of the carrier liquid flow are performed utilizing 1.19 $$\upmu \text {m}$$ μ m fluorescent polystyrene (PS) particles. Measurement results reveal a parabolic shape of the velocity profiles of both phases with a mean slip velocity of 7.4% at the position of maximum streamwise velocity in a 580 $$\upmu \text {m}$$ μ m high trapezoidal channel. An error analysis confirms the presence of these slip velocities within a 68.5% confidence interval. A measurement uncertainty in the order of magnitude of $${\mathcal {O}}(10^{-1}\ \mathrm{px})$$ O ( 10 - 1 px ) is reached for both fluid tracers and suspension particles. Overall, the present study demonstrates theoretically and experimentally that the usage of suspension particles with ring-shaped images is superior compared to Gaussian and plateau-shaped particle images of the same size. Additionally, the present study demonstrates that the usage of ring-shaped particle images allows to investigate suspension bulk dynamics by measuring velocity fields of both the suspended and the continuous phase simultaneously and with an overall uncertainty that is in the same order of magnitude as for standard $$\upmu$$ μ PIV measurements. Graphic abstract

Measurement of energy spectrum by using 100 eyes Tomographic PIV

PIV is one of the methods to measure velocity in a flow field, but its dynamic velocity range is narrower than other flow velocimeter. This disadvantage is particularly apparent in measurements of spectrum in turbulent boundary layers, where the higher wave number side of the spectrum cannot be measured with high accuracy. In this study, we captured images of the same particle in the flow field from many different direction simultaneously, and reduced the measurement error of the particle displacement by averaging the acquired particle positions, so called ‘Multiple Eye PIV’ [Maekawa, A., Sakakibara, J., 2018, Meas. Sci. Tech., 29, 064011]. We applied this method to obtain the energy spectrum in a turbulent pipe flow aiming for resolving higher wave number. Particle images were captured by a single high-speed CMOS camera (Fastcam Nova S6, 6000 fps, Photron) through a mirror array consists of 110 flat mirrors arranged in the shape of an axisymmetric ellipsoid (Fig.1), as shown in Fig.2. The images were evaluated by Tomographic PIV method to resolve three-dimensional velocity field. Fig.3 shows energy spectrum in a pipe measured by Tomographic-PIV with number of mirrors, N, up to 100 in addition to the 2D2C-PIV with a single mirror. Although the spectrum curve for the result of Tomographic-PIV begins to depart from the reference curve at wavenumber beyond 10-1 , such wavenumber grows as N increases, and consequently the plateau of the curve appeared at lower energy. Such a downward shift of the plateau is expected due to the improvement of the dynamic velocityrange, which is approximately one order in energy, i.e. three times in velocity, found between N=4 and 100. Note that the cases of N=4 and 40 loses the dynamic range against the 2C2D-PIV case. From the above, we can summarize that the advantage of Multiple Eye PIV over the 2C2D-PIV is effective when the number of mirrors is more than 40. In this experiment, the issue is that particles images flickered. In order to resolve this issue, we tried to use fluorescent particles, and obtained a clear particle images in the following experiment. We are now analyzing whether the energy spectrum can be measured with higher accuracy due to improved resolution of the particles.

Approximate Bayesian framework for 3D reconstruction in a volumetric PIV/PTV measurement

Non-invasive flow velocity measurement techniques like volumetric Particle Image Velocimetry (PIV) (Elsinga et al., 2006; Adrian and Westerweel, 2011) and Particle Tracking Velocimetry (PTV) (Maas, Gruen and Papantoniou, 1993) use multi-camera projections of tracer particle motion to resolve three-dimensional flow structures. A key step in the measurement chain involves reconstructing the 3D intensity field (PIV) or particle positions (PTV) given the projected images and known camera correspondence. Due to limited number of camera-views the projected particle images are non-unique making the inverse problem of volumetric reconstruction underdetermined. Moreover, higher particle concentration (>0.05 ppp) increases erroneous reconstructions or “ghost” particles and decreases reconstruction accuracy. Current reconstruction methods either use voxel-based representation for intensity reconstruction (e.g. MART (Elsinga et al., 2006)) or a particle-based approach (e.g. IPR (Wieneke, 2013)) for 3D position estimation. The former method is computationally intensive and has a lesser positional accuracy due to stretched shape of the reconstructed particle along the line of sight. The latter compromises triangulation accuracy (Maas, Gruen and Papantoniou, 1993) due to overlapping particle images for higher particle concentrations. Thus, each method has its own challenges and the error in 3D reconstruction significantly affects the accuracy of the velocity measurement. Though, other methods like maximum-a-posteriori (MAP) estimation have been previously developed (Levitan and Herman, 1987; Bouman and Sauer, 1996) for computed Tomography data, it has not been explored for PIV/ PTV 3D reconstruction. Here, we use a MAP estimation framework to model and solve the inverse problem. The cost function is optimized using a stochastic gradient ascent (SGA) algorithm. Such an optimization can converge to a better local maximum and also use smaller image patches for efficient iterations.

A novel streak velocimetry technique based on 2D fits of decaying phosphor particle images

A new 2D velocimetry technique based on streaks formed by individual phosphor particles, which are moving during their luminescence decay following pulsed excitation is proposed in this study. Tin-doped phosphor particles (Sr,Mg)3(PO4)2:Sn2+ are dispersed into flows and excited by a pulsed UV light sheet. During the phosphor decay time (~27 µs), the emission streaks due to particle motion are recorded. A 2D fitting is then applied on each particle streak against the analytical expression of intensity distribution, to obtain the velocity information for each particle. Unlike Particle Tracking Velocimetry (PTV) this technique does not rely on any particle image searching procedure.