For large-scale image retrieval, hashing algorithms are one of the most widely used methods due to their computational and storage efficiency. Compared with most of the existing data-dependent pair/triplet-based hashing methods, the hashing method based on central similarity quantization can optimize the global similarity more efficiently and alleviate the problem of missing the global nature of the data distribution. However, there still exists a lack of expression of the feature capability. Because of the different objective functions, there is an incompatible conflict between the optimal clustering position and the ideal hash position, leading to serious ambiguity and erroneous hashing after binarization. Therefore, we employ a hinge embedding function to explicitly force the termination of the metric loss to prevent negative pairwise infinite discretization. In addition, the performance difference of the models used in deep hash retrieval can also limit the efficiency of retrieval. To solve this problem, we propose an integration learning framework for image retrieval, which can learn compact hash codes by hash center constraints. We introduce the integration strategy and integrate the retrieval results using the weighted average method. Comprehensive experiments on three benchmark datasets, MS COCO, VOC2012, and ImageNet, show that the present framework has superior average accuracy mean on different lengths of hash code retrieval.
This paper proposes a trace hydrogen sensing system based on distributed feedback fiber laser (DFB-FL). The sensor head was fabricated by inserting the DFB-FL into a glass capillary coated with hydrogen sensitive material. The wavelength of the DFB-FL changed under the exothermic reaction of Pt-WO3 with hydrogen and was demodulated by a Michelson interferometer. The experimental results at hydrogen concentration of 50 to 3800 ppm show that the hydrogen sensitivity of the system is 0.0386 pm/ppm and the linearity index R2 is >0.99. The repeatability of the system is 99.4% and the accuracy is 97.43% at hydrogen concentration of 50 ppm. The figure of merit of the system is 0.0385. The proposed DFB-FL sensor with good repeatability and anti-electronic interference has application prospects in many fields such as the detection of trace hydrogen in transformer oil.
Infrared (IR) imaging can highlight thermal radiation objects even under poor lighting or severe sheltering but suffers from low resolution, contrast, and signal-to-noise ratio. While visible (VIS) light imaging can guarantee abundant texture details of targets, it is invalid in low lighting or sheltering conditions. Therefore, IR and VIS image fusion has more extensive applications, but it is a still challenging work because conventional methods cannot balance dynamic range, edge enhancement, and lightness constancy during fusion. To overcome these drawbacks, we propose a self-supervised dataset-free method for adaptive IR and VIS image fusion named deep Retinex fusion (DRF). The key idea of DRF is first generating component priors that are disentangled from a physical model using generative networks; then combining these priors, which are captured by networks via adaptive fusion loss functions based on Retinex theory; and finally reconstructing the IR and VIS fusion results. Furthermore, to verify the effectiveness of our reported physics driven DRF, qualitative and quantitative experiments via comparing with other state-of-the-art methods are performed using public datasets and in practical applications. These results prove that DRF can provide distinctions between day and night scenes and preserve abundant texture details and high-contrast IR information. Additionally, DRF can adaptively balance IR and VIS information and has good noise immunity. Therefore, compared to large dataset trained methods, DRF, which works without any dataset, achieves the best fusion performance.
Most current approaches in the literature of scene text recognition train the language model via a text dataset far sparser than in natural language processing, resulting in inadequate training. Therefore, we propose a simple transformer encoder–decoder model called the multilingual semantic fusion network (MSFN) that can leverage prior linguistic knowledge to learn robust language features. First, we label the text dataset with forward, backward sequences, and subwords, which are extracted by tokenization with linguistic information. Then we introduce a multilingual model to the decoder corresponding to three different channels of the labeled dataset. The final output is fused by different channels to get more accurate results. In experiments, MSFN achieves cutting-edge performance across six benchmark datasets, and extensive ablative studies have proven the effectiveness of the proposed method. Code is available at https://github.com/lclee0577/MLViT.
Since target information can be extracted from the light scattering distributions, light scattering from randomly rough surface has been studied in details. As numerical light scattering computation methods can provide precise information avoiding complicated operations and expensive experimental setups, various numerical methods have been widely used. However, most of them require ensemble average computation to obtain the stable results, inevitably decreasing the calculation efficiency. In order to further accelerate the processing speed by avoiding the ensemble average calculation, a high speed method based on surface slope probability density function is designed in this paper, which using the statistical parameters of randomly rough surfaces for direct light scattering calculation. With numerical simulations proving its high accuracy and rapid speed in light scattering computation, the slope probability density function based method is a potential tool for light scattering computation and analysis.
As an important marker in disease diagnosis, red blood cell morphology measurement is necessary in biological and medical fields. However, traditional setups as microscopes and cytometers cannot provide enough quantitative information in morphology detections. In order to capture tiny variations of red blood cells affected by metal ions in external environment, quantitative interferometric microscopy is applied: combining with phase retrieval and cell recognition, cellular phases as well as additional quantitative cellular parameters can be acquired automatically and accurately. The research proves that quantitative interferometric microscopy can be potentially applied in cellular observations and measurements for both biological and medical applications.
KEYWORDS: Digital imaging, Live cell imaging, Digital image correlation, Numerical simulations, Real time imaging, Phase retrieval, Optical engineering, Image processing, Microscopy, Blood
Dual-view transport of intensity equation (TIE) method is an ideal way for quantitative live cell imaging as it has advantages such as real-time imaging, multimode observations, compact setup, and large field of view (FoV). However, due to the image recorder installation error, the inevitable FoV mismatch between the captured under- and over-focus intensities reduces the accuracy in both amplitude and phase retrievals. Here, to eliminate this undesired FoV mismatch, the phase correlation-based digital FoV correction is adopted to recognize and compensate the rotation, scale, and translation between the under- and over-focus images. Both the numerical simulations as well as the experiments in standard sample detection and quantitative live cell imaging prove that the digital FoV correction combined dual-view TIE method can maintain the consistence of the dual FoVs, thus guaranteeing the high-accurate amplitude and phase computations, proving the proposed method is a promising quantitative live cell imaging tool in various applications such as biological observations and medical diagnostics.
A variable aperture-based ptychographical iterative engine (vaPIE) is demonstrated both numerically and experimentally to reconstruct the sample phase and amplitude rapidly. By adjusting the size of a tiny aperture under the illumination of a parallel light beam to change the illumination on the sample step by step and recording the corresponding diffraction patterns sequentially, both the sample phase and amplitude can be faithfully reconstructed with a modified ptychographical iterative engine (PIE) algorithm. Since many fewer diffraction patterns are required than in common PIE and the shape, the size, and the position of the aperture need not to be known exactly, this proposed vaPIE method remarkably reduces the data acquisition time and makes PIE less dependent on the mechanical accuracy of the translation stage; therefore, the proposed technique can be potentially applied for various scientific researches.
As a lensfree imaging technique, ptychographic iterative engine (PIE) method can provide both quantitative sample amplitude and phase distributions avoiding aberration. However, it requires field of view (FoV) scanning often relying on mechanical translation, which not only slows down measuring speed, but also introduces mechanical errors decreasing both resolution and accuracy in retrieved information. In order to achieve high-accurate quantitative imaging with fast speed, digital micromirror device (DMD) is adopted in PIE for large FoV scanning controlled by on/off state coding by DMD. Measurements were implemented using biological samples as well as USAF resolution target, proving high resolution in quantitative imaging using the proposed system. Considering its fast and accurate imaging capability, it is believed the DMD based PIE technique provides a potential solution for medical observation and measurements.
To realize portable device with high contrast imaging capability, we designed a quantitative phase microscope using transport of intensity equation method based on a smartphone. The whole system employs an objective and an eyepiece as imaging system and a cost-effective LED as illumination source. A 3-D printed cradle is used to align these components. Images of different focal planes are captured by manual focusing, followed by calculation of sample phase via a self-developed Android application. To validate its accuracy, we first tested the device by measuring a random phase plate with known phases, and then red blood cell smear, Pap smear, broad bean epidermis sections and monocot root were also measured to show its performance. Owing to its advantages as accuracy, high-contrast, cost-effective and portability, the portable smartphone based quantitative phase microscope is a promising tool which can be future adopted in remote healthcare and medical diagnosis.
In order to quantitatively analyze scattering from two dimensional randomly rough Gaussian surfaces, Kirchhoff approximation method is adopted in numerical calculation for analyzing full angular Stokes vectors of light scattering. With studying both the p- and s-polarized scattering fields from various materials such as metals and dielectrics, it is found that V components of scattering light from metals and dielectrics are different. Via analytical calculation according to slope probability density, the V component difference is attributed to refractive index of materials. Both numerical and analytical calculations prove the V component difference in light scattering can act as a criterion for metal and dielectric identification.
Massive image acquisition is required along the optical axis in the classical image-analysis-based autofocus method, which significantly decreases autofocus efficiency. A wavefront-sensing-based autofocus technique is proposed to increase the speed of autofocusing and obtain high localization accuracy. Intensities at different planes along the optical axis can be computed numerically after extracting the wavefront at defocus position with the help of the transport-of-intensity equation method. According to the focus criterion, the focal plane can then be determined, and after sample shifting to this plane, the in-focus image can be recorded. The proposed approach allows for fast, precise focus detection with fewer image acquisitions compared to classical image-analysis-based autofocus techniques, and it can be applied in commercial microscopes only with an extra illumination filter.
In order to obtain quantitative phase distributions from interferograms, phase retrieval composed of phase extracting and unwrapping is adopted in quantitative interferometric microscopy. However, phase unwrapping often requires a long time, limiting applications such as high-speed phase observations and measurements. In order to accelerate the processing speed, a phase unwrapping free Hilbert transform (HT)-based phase retrieval method is proposed. Though another background interferogram without a sample is needed, phase unwrapping can be omitted, saving a large amount of time for phase recovery. Additionally, the proposed HT-based method can maintain more sample details, thus providing high-accurate quantitative phase imaging. Considering its fast speed and high accuracy in phase retrieval, it is believed that the unwrapping free HT-based phase retrieval method can be potentially applied in high throughput cellular observations and measurements.
In order to improve detection speed and accuracy of biological cells, a quantitative non-interference optical phase recovery method is proposed in commercial microscope, taking the red blood cells as the classical phase objects. Three bright field micrographs were collected in the experiment. Utilizing the transport intensity equation (TIE), the quantitative phase distributions of red blood cell are gained and agree well with the previous optical phase models. Analysis shows that the resolution of introduced system reaches sub-micron. This method not only quickly gives quantitative phase distribution of cells, but also measures a large number of cells simultaneously. So it is potential in the use of real-time observing and quantitative analyzing of cells in vivo.
Erythrocyte morphology is an important factor in disease diagnosis, however, traditional setups as microscopes and cytometers cannot provide enough quantitative information of cellular morphology for in-depth statistics and analysis. In order to capture variations of erythrocytes affected by metal ions, quantitative interferometric microscopy (QIM) is applied to monitor their morphology changes. Combined with phase retrieval and cell recognition, erythrocyte phase images, as well as phase area and volume, can be accurately and automatically obtained. The research proves that QIM is an effective tool in cellular observation and measurement.
Quantitative interferometric microscopy is used in biological and medical fields and a wealth of applications are proposed in order to detect different kinds of biological samples. Here, we develop a phase detecting cytometer based on quantitative interferometric microscopy with expanded principal component analysis phase retrieval method to obtain phase distributions of red blood cells with a spatial resolution ~1.5 μm. Since expanded principal component analysis method is a time-domain phase retrieval algorithm, it could avoid disadvantages of traditional frequency-domain algorithms. Additionally, the phase retrieval method realizes high-speed phase imaging from multiple microscopic interferograms captured by CCD camera when the biological cells are scanned in the field of view. We believe this method can be a powerful tool to quantitatively measure the phase distributions of different biological samples in biological and medical fields.
Quantitative interferometric microscopy is an important method for observing biological samples such as cells and tissues. In order to obtain continuous phase distribution of the sample from the interferogram, phase extracting and phase unwrapping are both needed in quantitative interferometric microscopy. Phase extracting includes fast Fourier transform method and Hilbert transform method, etc., almost all of them are rapid methods. However, traditional unwrapping methods such as least squares algorithm, minimum network flow method, etc. are time-consuming to locate the phase discontinuities which lead to low processing efficiency. Other proposed high-speed phase unwrapping methods always need at least two interferograms to recover final phase distributions which cannot realize real time processing. Therefore, high-speed phase unwrapping algorithm for single interferogram is required to improve the calculation efficiency. Here, we propose a fast phase unwrapping algorithm to realize high-speed quantitative interferometric microscopy, by shifting mod 2π wrapped phase map for one pixel, then multiplying the original phase map and the shifted one, then the phase discontinuities location can be easily determined. Both numerical simulation and experiments confirm that the algorithm features fast, precise and reliable.
The paper proposed a simple large scale bio-sample phase detecting equipment called gravity driven phase detecting cytometer, which is based on quantitative interferometric microscopy to realize flowing red blood cells phase distribution detection. The method has advantages on high throughput phase detecting and statistical analysis with high detecting speed and in real-time. The statistical characteristics of red blood cells are useful for biological analysis and disease detection. We believe this method is shedding more light on quantitatively measurement of the phase distribution of bio-samples.
The statistical distribution of natural phenomena is of great significance in studying the laws of nature. Here, in this paper, based on laser scattering particle counter, a simple random pulse signal generating and testing system is designed for studying the counting distributions of three typical objects including particles suspended in the air, standard particles, and background noises. Moreover, in order to have a deep understanding of the experimental results from laser scattering particle counter, a random process model is also proposed theoretically to study the random law of measured results. Both normal and lognormal distribution fittings are applied to analyze the experimental results, and we have proved that statistical amplitude and width distributions of particles suspended in the air, standard particles, and background noise match well with lognormal distribution when natural numbers are used as the variables. This study is an important reference for statistical data processing for laser scattering particle counter, moreover, it will also be a useful guide for designing laser scattering particle counter with high accuracy and processing speed.
Volume Moiré Tomography (VMT) is an important technique to diagnose the flow field. In this Letter, the characteristic of temporal phase-shifting is analyzed for VMT. When the distance between two cross gratings is not on the Talbot distance, the phase-shifting factors are existed between moiré patterns of different orders. Especially, when the distance conforms to the sub-Talbot distance, the phase-shifting factors are maximum. This characteristic of temporal phaseshifting could be used for real 3-D flow fields reconstruction in the future.
Single-shot quantitative interferometric microscopy (QIM) needs a high-accuracy and rapid phase retrieval algorithm. Retrieved phase distributions are often influenced by phase aberration background caused by both imaging system and phase retrieval algorithms. Here, we propose an improved phase aberration compensation (PAC) approach in order to eliminate the phase aberrations inherent in the data. With this method, sample-free parts are identified and used to calculate the background phase, reducing phase errors induced in samples and providing high-quality phase images. We now demonstrate that QIM based on this PAC approach realizes high-quality phase imaging from a single interferogram. This is of great potential for a real-time speedy diagnosis.
Mueller matrix is a useful tool for analyzing polarization characteristics in a wealth of research fields. With Mueller
matrix, the modulation effects of samples on polarization could be quantitatively analyzed and discussed. In this paper,
all elements in the Mueller matrix are calculated when the lights scatter from one dimensional randomly rough surfaces
at different conditions with Kirchhoff approximation method which owns high accuracy and fast calculation speed.
Besides, theoretical analysis of the light scattering from randomly rough dielectrics and metal surfaces is also proposed
in this paper. Moreover, with both theoretical analysis and numerical simulations, we have explained the variations of all
elements in Mueller matrix, more importantly, m34 is highly focused which is quite a significant mark in both randomly
rough dielectric and metal surfaces. To our best knowledge, it is the first time this obvious difference is both analyzed
and discussed via both theoretical analysis and numerical calculation, and is successfully explained via phase difference
between incident and reflective waves. According to the analysis, more information of the target could be obtained in
order to determine the characteristics of the target. The paper will be an important reference for polarization imaging in
laser radar and remote sensing, etc.
Quantitative phase imaging of cells with high accuracy in a completely noninvasive manner is a challenging task. To provide a proper solution to this important need, interferometric phase microscopy is described which relies on the off-axis interferometry, confocal microscopy and high-speed image capture technology. Phase retrieval from the single interferogram is done by algorithms based on the fast Fourier transform, traditional Hilbert transform and two-step Hilbert transform, respectively. Furthermore, a phase aberrations compensation approach is applied to correct the phase distribution of the red blood cells obtained via the three methods mentioned before without the pre-known knowledge for removing the wave front curvature introduced by the microscope objectives, off-axis imaging, etc., which otherwise hinders the phase reconstruction. The improved results reveal the better inner structures of the red blood cells. The development of quantitative phase imaging technique is shedding light on their future directions and applications for basic and clinical research.
Phase distribution detection of cells and tissues is concerned since it is an important auxiliary method for observing biological samples. Here, in this paper, we have proposed phase
retrieval algorithms dealing with microscopic interferograms in order to solve two-dimensional phase
distribution. Based on phase distributions solved by phase retrieval algorithms, three-dimensional
refractive index distribution of biological sample is reconstructed which could reflect inner structure of
the cell. We believe these methods could be powerful tools in biological and medical fields.
Microscopic interferometry is a noncontact technique for quantitative phase imaging of live cells. The method combines
the principles of single-shot slightly-off-axis interferometry and confocal microscopy and is characterized by real-time
acquisition capabilities and optimized spatial resolution. However, slightly-off-axis interferometry requires less detector
bandwidth than traditional off-axis interferometry and fewer phase-shifted steps than on-axis interferometry. Meanwhile,
confocal microscopy allows microstructure magnification imaging. To validate the utility of this technique, experimental
and theoretical comparisons are given. The potential of the technique for phase microcopy is demonstrated by
experiments on red blood cells. This study will set the basis for interferometric phase measurements of dynamic
processes with fine spatial details, especially for observing live biological cell dynamics.
Tomographic interference microscopy is a method which can obtain the three-dimensional refractive index of live cells
and tissues. In this paper, the cone-shaped light with the transverse scanning is adopted, which offers non-contact, highresolution
and real-time cell magnifying imaging facilities relatively to the traditional parallel beam projecting on the
tissue sample. However, the index-induced focal shift is the common disadvantage for this tomographic interference
microscopy, which leads to the decline in image quality and impacts later reconstruction. Then the image sharpness
metric based upon the histogram of image to determine the index-induced focal shift is introduced. The experimental
results show that the variances of histograms are compatible with a Gaussian function. The peak value of the Gaussian
curve corresponds to the optimal imaging position's histogram variance.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.