We use an ultra-spectral resolution spectrometer for mesospheric OH radicals to obtain a high resolution direct solar spectrum in the wavelength of 308 nm based on spatial heterodyne spectroscopy techniques and the sun direction equipment. By removing the dark current and baseline, phase correction and Fourier transform, we get spectra from interferogram data. The results are in good agreement with radiative transfer model simulation. These results provide experimental data for orbital high resolution solar spectra and scattering signal detection.

Four-dimensional atmospheric propagation codes of high-power laser are used to simulate turbulence and thermal blooming effects of a focused solid laser beam propagating through non-Kolmogorov turbulent atmosphere. The variations of 63.2% encircled energy beam radius and beam quality factor with non-Kolmogorov turbulent spectral index α and propagation fluctuated strength D/r0 are analyzed. Deviations of simulation results of laser propagation through non-Kolmogorov turbulence and Kolmogorov turbulence are also compared. Results show that the smaller the spectral index α, the larger the beam spreading and the lower the energy concentration for turbulence effect and combined effect of turbulence and thermal blooming. The existed scaling relation for describing the beam spreading of focused Gaussian beam propagation through atmosphere is no longer valid for non-Kolmogorov turbulent atmosphere. Comparisons show that when considering the turbulence effect only, the largest relative deviation of beam spreading induced by Kolmogorov turbulence and non-Kolmogorov turbulence reaches to 87.7%, however, when considering the thermal blooming effect, the largest relative deviation reduces to 43.7%. It means that thermal blooming reduces the relative deviation of beam spreading of laser propagation in Kolmogorov and non-Kolmogorov turbulence.

Taking the oxygen absorption band A as an example, we consider some constraints of the system such as detection distance, target and background signal-to-noise ratio(SNR), system operating altitude on the channel bandwidth based on the selection rules of number and position of ranging spectrum channels, and simulate and analyze the range of the channel bandwidth upper and lower limits to meet the real-time performance of system, target SNR and different detection distance conditions by using MODTRAN and MATLAB softwares. The results show that the higher the platform altitude is, the lower the lower limit of bandwidth is, the larger the range is, the more flexible the selection is, the longer the system detection distance is and the larger the effective detection range is under the same conditions; the higher target SNR requirement is, the larger the lower limit of bandwidth is, the smaller the detection range is, the less flexible the selection is, the smaller the effective detection range and scope are under the same conditions. Therefore, in the design of passive ranging system, we should firstly determine the location and number of spectrum channels based on the task design requirements, and then calculate the upper and lower limits curves of each channel under multiple constraints of the real-time system, the signal to noise ratio, and the maximum and minimum working distance of the system. Secondly it will be determined that the minimum intersection based on the upper and lower limit curves that satisfy both the minimum and maximum working distance. Finally, the best channel bandwidth values that meet the design requirements could be selected in the intersection. It can provide an effective theory support and calculation method for the parameter design and engineering of single-lens multi-spectrum passive ranging system with arbitrary precision and arbitrary distance.

Based on the power spectrum inversion method to produce oceanic turbulence phase screen, the long-exposure beam radius, spot centroid wander features and irradiance scintillation features of different beam arrays (rectangle, radial, single beams) in different oceanic turbulence conditions are simulated and analyzed. Results show that long-exposure beam radius, standard deviation of spot centroid wander and on-axis scintillation index of array beams increase with increasing of turbulent effects. Simultaneously, the long-exposure beam radius of the beam array is tending to be equivalent with that of the single Gaussian beam. However, after a longer distance propagation, the radius of single beam is slightly larger than the array beam radius. Compared to a single beam, beam arrays have smaller wander standard deviation under the same turbulence conditions but larger scintillation index. The scintillation in oceanic turbulence is stronger than in the atmospheric turbulence.

The variation relationship of ion yield for xenon atom under the single ionization in the mid-infrared band versus laser intensity is measured systematically by using the wavelength-tunable strong femtosecond laser pulses. The theoretical calculations based on the PPT and ADK models are done and the theoretical calculation results are compared with the experimental results. It is found that all the calculation results based on the ADK model are consistent with the experimental results in the whole measurement region, while the calculation results based on the ADK model do only when the wavelength is long enough and the intensity is strong enough. The unification for the two models in the deep tunneling ionization regime and the validity for the two models in describing the single ionization of atoms are directly verified in experiments, and thus these two models can be used as a precise tool to calibrate the absolute intensity of long wavelength laser pulses.

Extremely straight interference fringes can be formed at the position far away from the waist of Gaussian beams and based on this principle, a novel far-field-interference-based scanning beam interference lithography (SBIL) optical system is proposed. The analytic expression of the nonlinear error with respect to waist radius of Gaussian beams, the incident angle and the waist-to-substrate distance is established. By the numerical simulation, the relationships between the nonlinear error of fringes and the above parameters are analyzed in detail. The research results show that this optical system can effectively limit the nonlinear error of fringe phase to the nanometer scale, and possesses the advantages of simple optical path and high tolerance of assembly errors. The problem that the nonlinear error of fringes at the boundary of the exposure spot is deteriorated can be effectively solved if the waist-to-substrate distance is suitably shortened.

In this work, Ge15Sb20Se65 glass is synthesized by the melt-quenching method and then drawn into a bare glass fiber with a diameter of 500 μm. The minimum transmission loss of this fiber is about 1.68 dB/m at the wavelength of 6 μm. A homemade tapering platform allows to taper the chalcogenide fibers into different waist diameters of 20, 100 and 250 μm, respectively. The spectroscopic analysis of the tapered fibers in ethanol solution with different concentrations is presented. Based on the theory of fiber evanescent wave, the sensing characteristics of the tapered fiber in the ethanol solution are simulated by COMSOL Multiphysics software. The results of simulation and experiment are compared and discussed.

Aiming at the optimization design of Raman fiber amplifier with multiple backward pumps, we combine the artificial bee colony algorithm with the average power analysis to optimize the configuration of backward pump of Raman fiber amplifier. On the basis of the given gain, the optimized model is established in order to minimize the ripple of Raman gain. The average power analysis is utilized to solve Raman scattering equations, and the artificial bee colony algorithm is employed to find the optimal pump wavelength and pump power. The proposed method is used to design the Raman fiber amplifier at C band and C+L band. The results show that the gain ripple can be controlled below ±0.4 dB under different net gain levels. Compared with the existing methods, the combination of artificial bee colony algorithm and average power analysis has a good performance and can obtain the flatter optimization results, which has certain practical value.

Based on the principle of light intensity modulation and micro-electromechanical system (MEMS) technology, we design a kind of optical fiber acceleration sensor. Two silicon rectangular beams are used as the acceleration sensing elements. According to the relation between the shade displacement and the light power difference of the receiving optical fiber, as well as the relation between the deflection of the silicon rectangular beam at center position and the system acceleration, an expression of acceleration with the light power difference is given. The research result shows that, while the deflection of the silicon rectangular beam at center position is 0~10 μm, the corresponding acceleration value is 0~120 m·s -2, and the maximum acceleration detective value can be as much as 12g (g is the acceleration of gravity).

We study the underwater propagation characteristics of orbital angular momentum (OAM) states by experimental method. We build an underwater optical communication system of OAM states, and then discuss the underwater transmission capacity of OAM states by the spread spectra and the probability curves of the original and adjacent OAM states. The experimental results show that the probability of original OAM state decreases significantly with the increasing of salinity, and the propagation probability decreases with the increasing of topological charge interval.

In this paper, we design and optimize the ship-to-airship capture scheme of laser communication system according to the motion characteristics of an airship platform. A composite scanning strategy, consisting of coarse scanning and fine scanning, is proposed to solve the problem of the long capture time under conditions of small beam divergence angle and far distance. The working principle of the composite scanning strategy and the influencing factors in the capture process are analyzed emphatically, meanwhile, the important parameters in the coarse-fine composite scanning capture process, such as overlap area, capture field, scanning speed and motion compensation are optimized in detail. The simulation results show that the acquisition time reaches 62.7 s when the divergence angle is 400 μrad. Finally, we also validate the results by performing the outdoors airship experiment.

A new kind of refractive index sensor based on a hollow optical fiber with a metal-dielectric-metal multilayered films structure is proposed and its transmission spectrum is calculated by the establishment of an optical model. The performances are analyzed for the designed sensors with SiO2, cycloolefine polymer and AgI as the dielectric materials, respectively. When the refractive index of the detected liquid in the hollow optical fiber is in different ranges, the principles of the guided mode resonance, surface plasmon resonance (SPR) and waveguide-coupled SPR are employed for sensing by the designed sensor, respectively. Compared with the traditional hollow fiber sensors, the proposed sensor not only has a greatly extended detection range which is 1.3-1.64 and almost covers all the refractive indexes of liquids, but also its figure of merit is increased up to twice.

In view of the nonlinear relationship between displacement and bending loss in traditional fiber bending loss displacement sensor, and the problem that high sensitivity and large scale cannot be obtained at the same time, we design a novel displacement sensor based on the bending loss of U-type winding fiber. The fiber consists of a U-type turning and a spiral winding around the shaft, which of them are called U-type winding. It is proved theoretically that there is a linear relationship between the measured displacement and the bending loss. We deduce the expression of them, discuss the effect of spiral winding on the bending radius of fiber and the effect of U-type turning on the accuracy of the sensor, and carry out a series of experimental studies and performance tests. Experimental results demonstrate that displacement sensitivity of the sensor is 0.14 dB/mm in the range from 0 to 120 mm, and the linear correlation coefficient is greater than 0.99. There is no stress relaxation of the optical fiber in the winding process. It is suggested that the radius of curvature of the U-type turning fiber should be larger than 6 mm. The displacement sensor based on the bending loss of U-type winding fiber has good measurement precision and large scale. It can realize the monitoring of concrete structure cracks and continuous monitoring of the displacement of large structure.

A orthogonal frequency division multiplexing(OFDM)/mode division multiplexing multimode fiber transmission system based on photonic lanterns(PL) is proposed. Two mode selective PLs are used as multiplexer(MUX) and demultiplexer(DEMUX). LP01 mode and LP11b mode are selected as the transmission channels. An adaptive bit loading OFDM modulation is used to achieve the transmission of 7.2 Gb/s on a 50 m OM4 multimode fiber. The experimental results show that when the incident light power of LP01 mode is about 4 dB lower than that of LP11b mode, we can achieve the same received optical power by adjusting the polarization state of the two signals. When the received power is -13 dBm, the bit error rates(BERs) of two signals are 1.3×10-3 and 3.2×10-3, respectively, which are lower than the threshold of hard decision-forward error correction (HD-FEC). The system provides a solution for the large capacity data transmission with low cost at short distance.

Concerning to long-range, wide-field and multi-targets electro-optical tracking system, it is difficult to take account of the requirements of large-aperture, wide field-of-view and high angular resolution at the same time in the form of existing optical systems. By mimicking the visual physiological mechanism of the roving fovea of human eye, a new common aperture coaxial five mirrors medium wave infrared (MWIR) imaging and laser radar optical system with light-weighted object orientation scanning corrector optics is designed. The primary mirror aperture of this optical system is 1000 mm, the instantaneous scanning effective aperture is 600 mm, the maximum field-of-view angle is 20.7°, the angular resolution of the MWIR imaging sensor is higher than 13 mrad, and the operating distance of the laser radar reaches 20 km. The new optical system has wide field-of-view, which can track more than six long-range fast-moving targets at the same time.

An optical design is proposed for the Michelson interferometer based line field-swept source optical coherence tomography (OCT) system. An aspheric lens group for beam shaping is utilized in the illumination system and a cylindrical lens group is used for the line focus of beams, thus the scanning dimension is reduced and the goal of high-speed imaging is realized. The shaping of Gaussian beam is obtained via adjustment of the parameters of the aspheric lens group. In the projection system, the telecentric structure can reduce the system errors, the combination of different materials can be selected to correct chromatic aberration in the near infrared wave band, and the design of two systems is completed. In the Zemax non-sequential mode, the illumination and projection channels are combined by a beam splitter and the ray tracing is done. The interference performance of this OCT system is confirmed. The results show that the designed OCT system can be used for the realization of high-speed and high-quality imaging.

The enhancement of infrared images containing sea and sky requires not only to remove effects by heavy wind and waves, but also to enhance target region’s contrast. Exploring several typical image enhancing algorithms, and combining the characteristics of infrared image’s histogram of gradient magnitude, an algorithm based on the infrared image’s gradient domain is proposed. Low magnitude is set to zero to reduce wave disturbance, gradient magnitude range is adjusted to control brightness, and valid gradient magnitude range is equalized to enhance target region’s details. Experimental results show that the proposed algorithm can obviously reduce the interference information of the image and improve the contrast and information entropy of target region.

Three-dimensional point cloud data can be obtained from fringe projection method. However, these point cloud data may have holes for complex shaped objects, and the holes have a profound impact on afterward processing. According to structure from motion (SFM) data acquisition, we propose a new method to repair holes. Firstly, we use a two-dimensional phase of fringe projection to extract boundary points from the three-dimensional point cloud. Then, we register the data set obtained from SFM and fringe projection to extract supplementary points. Finally, based on the point cloud data set with supplementary points, the repair of point cloud holes is implemented based on the radial basis function to calculate the surface equation. Experimental results show that the algorithm is robust and it can recover the surface information of complex objects effectively.

We propose a new method for stereoscopic image comfort assessment based on convolutional neural network, which does not need to extract specific manual features from images in advance according to specific tasks, but simulates hierarchical abstract processing mechanism of human brain to extract image features autonomously. This method adopts three channel convolutional neural network structure, and the input data sets of the three channel are obtained by reducing the dimension of the original data samples through principal component analysis, and chopping the original data samples into two size image patches (32×32, 256×256), respectively. The network structure of each channel is designed according to the input data sets. In addition, the classification accuracy of this method is improved by introducing dropout and local response normalization, etc. With rectified linear unit as the activation function and Softmax as the classifier in the output layer, experiment results on 400 stereo image samples in TJU database with different comfortable levels show that, the correct classification rate of this method is 94.52%, which is higher than that of the extreme learning machine and support vector machine.

Focused light field imaging has overcome the disadvantages of low spatial resolution in traditional light field imaging, but the induced artifacts at out-of-focus planes due to sacrificing the angle sampling density. This not only affects the visual appearance of the image but also severely decreases the accuracy of defocus response analysis. We present a method based on sliding window fusion and adaptive median filtering, which can improve the angle resolution and eliminate the artifacts at out-of-focus planes. A window with an appropriate size is selected to slide in the sub-images of the four-dimensional light field data with a specific step. Basic light field refocusing algorithm is conducted on each slide to produce multi-view data, which are then fused to get a refocused image at a corresponding plane. The focal stack can be obtained by changing the window size and repeating the above calculation. Each layer of the focal stack is filtered by an adaptive median filter with the kernel size being linearly correlated with the window size. As a result, artifacts can be removed in the whole focal stack. Todor dataset is used to test the proposed method. The results show that the proposed method improves the visual appearance at out-of-focus planes, and the accuracy of defocus response.

The foreground ray has a serious impact on the quality of reconstruction during the traditional synthetic aperture imaging is refocused on the surface of object. For the above problem, we propose a refocusing imaging algorithm via foreground labeling in this paper. The algorithm first estimates the depth range of the scene based on the edge features of the EPI, extraction of foreground edge features and propagation according to the parameters of the specified surface to be rebuilt, thus determining the foreground occlusion corresponding rays. After labeling and removing those corresponding rays, reconstructing a specific surface with light field reconstruction algorithm to achieve the high quality reconstruction of the occluded object.The paper uses datasets provided by Stanford University and Disney laboratory to simulate, the experimental results show that the proposed algorithm can effectively remove the information of the occlusion in the scene and improve the quality of the refocused image.

The underwater polarimetric imaging target enhancement technique based on unpolarized illumination is proposed based on the physical model of image degradation caused by light absorption and backscattering by particles in water. The advantage of this technology is that the unpolarized light ensures that there is always a difference in polarization between the target reflected light and stray light. And, the characteristic parameters of angle of polarization is used to ensures the accuracy of estimation of stray light intensity. Compared with the current underwater polarimetric imaging techniques based on polarized illumination, the proposed method has wide application range and high image recovery accuracy. Experimental results show that the proposed method can effectively improve the visibility and contrast of underwater images, and the contrast is increased by at least 100%. It is suitable for water bodies with different material targets, different imaging distances, and different impurities and turbidity levels. It has potential application in the underwater imaging field.

The structured light is widely used in 3D surface shape measurement. However, it is likely to cause the intensity saturation when projecting certain intensity structured light to the surface with high reflection. The intensity saturation always causes the reconstructed surface with large errors and even hard to be reconstructed. In order to improve the quality of 3D reconstruction of high reflective objects, an active compensation method base on light intensity divisional projection is proposed. Firstly, the position of the saturated region in the projection plane is calculated by the Gray code grayscales in the projection area. Then, the transition compensation region is proposed to reduce the projection intensity of the fringe pattern in the saturated region smoothly. Finally, the divisional projection optimized compensation method for saturated region is verified by the experiments. The experimental results show that the proposed method can reduce the number of projected images needed to calculate the compensation region, achieve a smooth transition of the saturated region boundary, improve the computational efficiency, effectively suppress the reconstruction error caused by the saturation, and improve the accuracy of the 3D reconstruction.

The As2Se3 specimens with different thicknesses, roughnesses and surface smoothnesses are prepared by use of homemade As2Se3 glass rods. The refractive indices are measured by the infrared spectroscopic ellipsometer, and an optical model is built to obtain the refractive indices by fitting. The influences of thickness, roughness and surface smoothness on the refractive indices of specimens are comparatively analyzed. The results show that all factors have obvious influences on the measurement accuracy of the ellipsometer and the surface smoothness is of most importance. If the specimen thickness is controlled to be 1-3 mm, and simultaneously the back side roughness and surface smoothness of specimens are increased, the measurement accuracy of the ellipsometer can be obviously enhanced.

In order to solve the indoor attitude testing problem of the single theodolite, a target board engraved with different angled lines is placed at the focal plane of collimator to simulate the attitude of infinity target. The mathematic model of the relationship between point coordinate on each quadrant of the target board and total station test angle is established, and the test target board is designed. Moreover, the total station is used to sample and test the points of the lines on target board, and the tilt angle of line is calculated through the mathematical model. The experimental results show that the simulation error of angle between line 1# and line 2# is 0.160°, and the simulation error of angle between line 1# and line 3# is 0.046°, which can satisfy the requirement of the single station image with maximum axis slope extraction error of 0.6°.

In order to enhance the integrity and accuracy of moving object detection in complex scenes, a multi-features-based moving object detection method is proposed. The color feature is modeled by using the proposed adaptive Gaussian mixture model (GMM) algorithm. A kind of hysteresis multi-thresholds modeling method is used to model the scene background by adopting the color and improved local binary pattern (LBP) texture feature simultaneously. A neighborhood compensation strategy is adopted to combine the object regions obtained by the two-features extraction. The improved Kirsch edge detection method combined with the Canny thoughts is adopted in the edge extraction which eliminates the mistakenly detected ghost pixels and improves the edges of foreground objects. The experimental results show that the proposed method is superior to the traditional algorithms in the detection integrity and accuracy, and the real-time performance is also better.

A fast method to measure the primary surface dynamic deformation of a radio telescope for improving high-frequency observation efficiency is proposed based on the similarity and overlap of the multibeam patterns, and the phase retrieval method. Firstly, the pointing offsets relative to the center are subtracted from a part of in-focus and defocus patterns of each beam, and these patterns are spliced to construct a group of complete in-focus and defocus patterns of the center beam. Secondly, the sum of squared residuals by which the splicing patterns differ from the theoretical ones is minimized based on Levenberg-Marquardt algorithm, to calculate the coefficients of Zernike polynomials, and the primary reflector dynamic deformation is retrieved. The numerical simulation experiment based on the 7-beam Q-band shows that the proposed method can retrieve the primary reflector dynamic deformation of the radio telescope in 5 min with the measurement accuracy of 55 μm, with the advantages of anti-noise and anti-fluctuation, and can be applied for the astronomical observations below 90 GHz.

Three integer pixel search algorithms of genetic algorithm, particle swarm optimization, and artificial fish swarm algorithm applied in the micro-displacement measurement are studied and compared. The matching precision of the images is evaluated by the size of the correlation coefficient and the normalized cross-correlation function is selected as the correlation coefficient and the objective function of algorithms. The objective function is iteratively solved to obtain the micro-displacement result of the integer pixels. The simulated speckle pattern is taken as the research object, and the matching precision, searching speed, and micro-displacement measurement results for these three algorithms are compared and analyzed. The results show that the genetic algorithm has obvious advantages in the matching precision, searching speed and micro-displacement measurement precision, which can meet the application requirements of the digital image correlation method in the micro-displacement measurement.

We combine the specially designed low threshold five-mirror folded cavity, employ the home-made double wall carbon nanotubes (DWCN) as saturable absorber by vertical growth method, and realize continuous mode locking operation in Tm, Ho∶LLF all solid state laser. Absorbed pump threshold is as low as 59 mW under the continuous-wave laser operation, and the light to light conversion efficiency reaches 30.09%. The absorbed pump threshold is increased to 107 mW after DWCN is inserted in the cavity. When the absorbed pump power is higher than 1562 mW, the laser operation gets into a stable continuous mode-locked state, for which the corresponding mode-locked pulse repetition frequency is 100 MHz, and the pulse width is about 515 ps. When absorbed pump power reaches 2 W, the maximum output power is 234 mW, the central wavelength is 1895 nm, and the maximum single pulse energy is 2.34 nJ.

To solve the problem that the accuracy of visual odometry was decreased in the weak texture environment, an optimization algorithm of binocular visual odometry is proposed. Firstly, the planar constraints are provided for the features without stereo matching by extracting the local feature plane, to increase the quantity of effective stereo features. Secondly, in feature tracking, the uniform acceleration motion model is used during feature tracking to increase the quantity and improve the quality of tracked features. Finally, in pose estimation and optimization, the bundle adjustment method which considers the feature confidence is adopted to reduce the influence of distant feature, and improve the accuracy and robustness of the algorithm. The experimental results based on the datasets and actual scene show that the proposed algorithm has obvious optimization effect on the positioning accuracy under weak texture environment with little computational resources, and has adaptability in other scenes.

In order to effectively utilize the long-term temporal information of video for improving the accuracy of action recognition, a new recognition approach is proposed based on the sequential optical flow image and double-stream convolutional neural networks. Firstly, the Rank support vector machine (SVM) algorithm is used to compress the continuous optical flow frames into a single sequential optical flow image to realize the modeling of the long-term temporal structure of video. Secondly, we design a double-stream convolutional networks containing appearance and short-term motion stream and long-term motion stream. It takes the stacked RGB frames and the sequential optical flow images as input to extract the appearance and short-time motion information and the long-time motion information of the video. Finally, the linear SVM is adopted to integrate C3D descriptor and VGG descriptor for action recognition. The experimental results on HMDB51 and UCF101 datasets show that the proposed approach improves the action recognition accuracy effectively by using the spatial information and the temporal motion information.

In order to polish up the target edge burring and staircase effect in low-textured regions or discontinuous regions, a stereo matching method based on the local binary description and superpixel segmentation is proposed. Firstly, the initial disparity is obtained by space and color features binary cost computation and winner-takes-all method. Then the segmentation results by simple linear iterative clustering method are labeled for each pixel's space and color features. In disparity refinement procedure, the appropriate fixed points are chosen to propagate disparity for both edge and inner pixels of each superpixel. Experiments with Middlebury datasets are mainly carried out in the initial disparity considerations and disparity refinement. The result shows that the disparity maps are much smoother especially in target boundary. The proposed method can achieve more accurate disparity value in non-overlapping and occluded regions between reference image and matching image, which effectively reduces the mismatching rate in non-occluded, all, and discontinuity regions.

Compared with the traditional detectors, the detectors based on large data and deep learning do not require manually designed features and are more robust. Under the background of air defense, we build the images and videos dataset of aerial target for training and test, improve the deep learning-based detector Faster R-CNN, and specialize it in aerial target detection. Aiming at the peculiarities and requirements of aerial target detection, we propose the strategies such as accumulation of dilation, regional amplification, local tagging, adaptive threshold and spatio-temporal context to make up the shortage of Faster R-CNN that small weak or occluded targets can not be detected and improve the detection speed and accuracy. Experimental results show that the improved Faster R-CNN performs well under circumstances such as small weak or multiple targets, clutter, illumination changes, blur and large-area occlusion. Compared to the original Faster R-CNN, the mean average precision is improved by 16.7% on the built dataset, and the speed is 3 times faster.

The precursor of Eu3+-doped zinc molybdenum oxide hydrate is synthesized at the room temperature by using the co-precipitation method. The influences of sintering temperature on the microstructure and photoluminescence performance of this precursor are investigated. The research results show that the Eu3+-doped ZnMoO4 nanostructure can be obtained by sintering the precursor of Eu3+-doped zinc molybdenum oxide hydrate with a low cost and at about the temperature of 400 ℃.

In order to explore the application of hyperspectral technology in the pathological diagnosis of gastric cancer, we combine hyperspectral imaging and microscopy to acquire hyperspectral images of gastric slices. According to the difference of spectral characteristics between gastric cancer tissue and normal gastric tissue in the wavelength of 410-910 nm, we propose a classification method based on convolutional neural network (CNN). The original spectrum is preprocessed by S-G smoothing and the first order derivative. We establish the optimal network structure and parameters by analyzing the spectral data characteristics and the classification efficiency. Experimental results show that the classification accuracy of cancerous and normal gastric tissues is 96.53%, the sensitivity and specificity of distinguishing gastric carcinoma reach 94.29% and 97.14%, respectively. Compared with shallow learning methods, the CNN model can fully extract the deep spectral characteristics of cancerous tissues and effectively prevent over-fitting. The method of deep learning combined with micro-hyperspectral imaging can also provide a new idea for the medical pathology research.

The projection products of pure laser sources are one of the future trends in the development of display technology. In order to use multiple laser diodes as the light source to achieve high power uniform illumination, a laser illumination system is designed in this paper. Two lenses and a round rod are added before the uniform illumination system based on the freeform surface array to further improve the uniformity of target surface. The laser array is collimated and then converged into a round rod for transmission. After being collimated by the lens, it passes through the freeform surface array and is converged on the target surface by the field lens to form a uniform illumination area. With the help of software simulation, it is found that the uniformity of the target surface increased from 77.59% to 90.99% after mixing the light with the round rod. In addition, the shape of the freeform array unit is studied. The experimental results show that, compared with the square structure unit, the freeform surface unit with the same aspect ratio as the target surface can get better uniformity.

A method for label-free imaging to single nanoparticle by evanescent wave in-plane scattering is presented. Using the two evanescent waves of total internal reflection (TIR) evanescent wave and surface plasmon polaritons (SPPs) to interact with a single nanoparticle, respectively, which excites the nanoparticle polarization and scatter. Generated interfacial scattering is interfered with incident evanescent wave, which forms a characteristic imaging of the polarization field of the nanoparticles and the parabolic interference fringes. Single nanoparticle label-free imaging is performed on three polystyrene nanoparticles with the diameters of 500, 200, 100 nm. The imaging results of two evanescent wave in-plane scattering on single nanoparticle are compared. It indicates that the nanoparticle polarization of SPPs in-plane scattering is approximately 10 times stronger than that of TIR in-plane scattering and close to dark field imaging. As a result, SPPs in-plane scattering manifests the better sensitivity to label-free single nanoparticle imaging. This single nanoparticle label-free imaging method can be extended to areas such as virus detection and single bio-molecule imaging.

In the unknown environment, the automatic identification and classification of unmanned aerial vehicle (UAV) landing landforms are of great significance. The traditional natural scene classification uses the information of the middle- and the low-level features, but the UAV landing landform image has complex scene and rich information, which needs high-level semantic features to express more accurate information. A landform image classification algorithm based on discrete cosine transform (DCT) and deep network is proposed. First, the advantage of DCT energy concentration is introduced into the efficient feature representation of convolutional neural network (CNN) to reduce the dimensionality and computational complexity. Then a 14-layer feature learning network is constructed based on the characteristics of landform image, and the CNN structure is improved. Finally, the deep features are input into the support vector machine (SVM) to complete the image classification quickly and accurately. Experimental results show that the algorithm reduces data redundancy and training time greatly, and can automatically learn high-level semantic features. The features extracted by the proposed algorithm have better feature expressions and effectively improve the image classification accuracy.

In the multi-sensor fusion tasks of automatic drive, the strategy and the results of the data fusion are greatly influenced by the uncertainty of each subtask. To keep the whole system run steadily in multiple circumstances, the calculation model must operate with low uncertainty. The existing methods can only obtain uncertainty in the neural network prediction process, and few methods can reduce the uncertainty of the model in a self-learning method. To address the above problems, the concepts of uncertainty learning layer and uncertainty loss term are proposed, and a neural network architecture (ULNN) which can reduce uncertainty by self-learning method is designed to enhance the robustness of neural network model prediction. Experiments on CIFAR-10 and CIFAR-100 datasets show that ULNN can effectively reduce the model uncertainty and obtain 26 and 12 times lower uncertainty on the two data sets respectively. The universality of ULNN is proved by the experimental results of semantic segmentation on CamVid dataset.

Based on the linear superposition of the odd and even modes of the Ince-Gaussian (IG) beam, a novel V-shaped IG beam mode is proposed, called as V-shaped Ince-Gaussian mode (VIG mode), in which each light petal has a V-shaped structure. The experimental and numerical simulation results show that the number of V-shaped petals is twice as much as the order of the VIG mode. Given an initial phase difference between the odd and even modes, each V-shaped petal in VIG modes is split into a large and a small light petals. Moreover, this splitting process is freely controlled and the spatial positions of these two separated petals are able to exchange. The analysis of the force field of the VIG mode shows that the VIG mode is expected to be a candidate beam for cell sorting. Further, if the linear superposition is conducted by two even modes and one odd mode with order ratio of 1∶3∶2, a three branches VIG mode can be generated.

This article presents a hybrid algorithm for improving the accuracy of wavefront reconstruction of a defocused Shack-Hartmann wavefront sensor using the moment method. For a defocused Shack-Hartmann sensor, not only the shift of centroid for each microlens spot but also the second-order moment of each spot are extracted for calculating the local slopes and curvatures of the input wavefront. To improve the accuracy of the wavefront reconstruction, we employ these calculated slopes and curvatures to reconstructure the input wavefront instead of just using the calculated slopes for the conventional Shack-Hartmann sensor. The proposed method is verified by numerical simulations. It shows that the proposed method is useful for wavefront reconstruction with high accuracy. For example, the root mean square value of the residual wavefront error for reconstruciton of the peaks function is 0.0327λ using the proposed method, which is smaller than that, i.e., 0.0903λ, using the conventional method, indicating the proposed method can significantly improve the accuracy of the wavefront reconstruction.

River is a very typical and important geographical target in remote sensing images. The automatic detection of rivers is of great significance in water resources investigation and water conservancy planning. In this paper, a river target detection algorithm based on multi-feature fusion and soft voting method is proposed. The algorithm firstly divides the images into cells and then extracts the local entropy, texture, spectrum, and color feature of the cells. Random forest is used to train and classify. To optimize the rough detection result of machine learning, the morphology operation and the multi-criteria voting method is introduced. For optimized rough detection result, the level set active contour is used to approach the river shoreline. Experiments show that the proposed algorithm has a good detection effect, and the detection accuracy rate of test set reaches 97.44%. In addition, the river can be effectively detected in the complex background.

Aiming to solve the problem that the complex background pixels affect the hyperspectral classification accuracy, the object detection theory is introduced into the hyperspectral image classification domain, and a hyperspectral image classification method based on spectral-spatial feature iteration is proposed. A multi-target constrained classifier (MTCC) is designed by constrained energy minimization method. Based on the detection theory, the MTCC can effectively decrease the influence of complex background data on the classification accuracy. At the same time, to eliminate the over-classification problem caused by the spectral features, the method uses the feedback fusion of spectral-spatial to strengthen the spatial enhancement information so as to improve the classification accuracy gradually. The results of the experiments on the data sets of Purdue, Salinas and Pavia show that the average accuracies of the proposed methods are 98.09%, 97.33% and 84.68% respectively, and the precisions of the proposed method are 96.84%, 95.32% and 79.13% respectively. Compared to other algorithms, the proposed method has higher generalization ability and practicability.

The influence of single-particle scattering on the state transmission of orbital angular momentum (OAM) of Laguerre-Gaussian (LG) beams is studied by using the finite-difference time-domain method. The effects of particle size, particle position, ellipsoidal-particle-radii ratio, orientation angle and initial OAM mode of LG beams on the state degradation of OAM are investigated, respectively. The research results show that the larger the particle radius is, the more serious the OAM state degradation is. The change of the relative position between the scattering particles and the beam also leads to the regular change of the energy proportion. The initial OAM mode, ellipsoidal-particle-radii ratio and orientation angle all have influences on the state degradation at different extents.

The uncertainty of the Mueller scattering matrix of nonspherical aerosol is a significant factor influencing the accuracy of polarization sensing. In the self-developed multi-resolution time-domain (MRTD) aerosol scattering model, the Mueller scattering matrix should be calculated from the electric field in far region. Therefore, the near-to-far field transformation process becomes an important step influencing its simulation accuracy. In order to determine the best near-to-far field transformation scheme for MRTD scattering model, the simulation accuracy of the Mueller scattering matrix corresponding to the surface integration scheme based on Huygens principle and the volume integration scheme based on Helmholtz equations for dielectric medium is compared and analyzed systematically in the case of spherical and nonspherical particles. The results show that, although both of the near-to-far field transformation schemes can achieve near-to-far field transformation effectively and accurately, the near-to-far field transformation scheme based on volume integration principle has better performance based on the calculation error distribution of the Mueller matrix.

The inner porosity of tablets plays an important role in its disintegration characteristics, and then affects its efficacy. However, few researches have been reported to detect tablets porosity so far. The quantitative relationship between the porosity, refractive index and the indomethacin mass fraction of mixed tablets composed by microcrystalline cellulose (MCC) and indomethacin (API) is studied using terahertz spectroscopy based on the time delay effect of terahertz pulse, and its mathematical model is built. The porosity and reflective index are calculated by measuring terahertz spectrum of the tablets. The results show that the relationship between porosity and reflective index or mass fractions of different compositions is a linear correlation, which is consistent with the mathematical model. The average relative error of porosity obtained by terahertz spectroscopy and gas replacement method is 6.0%, and the average relative error of porosity obtained by terahertz spectroscopy and density method is 8.9%. The terahertz spectroscopy can be used to measure the porosity of the tablet, and it is a supplementary measure in monitoring the quality of the tablets.

The changes of hydrogen bond structures and the ion association of (NH4)2SO4 aqueous solution with different mass fractions at room temperature are investigated by Raman spectroscopy and excess Raman spectroscopy. The stretching vibration interval of O-H at 2800-4000 cm-1 and holosymmetric stretching vibration interval of SO42- at 940-1020 cm-1 are studied. The peak splitting of Raman data are handled, and positive and negative peaks integral areas of excess Raman spectra are analyzed. The results show that the mass fraction of 12.00% is the turning point of the hydrogen bond structures and the ion association in the solution. When the mass fraction is greater than 12.00%, the hydrogen bond structures of DDA type and DAA type increase, and the hydrogen bond structures of DDAA type and DA type as well as free OH decrease. The existence of ion pairs changes in the solution and NH4+ forms contact ion pairs with SO42-.

A hyper-resolution spatial heterodyne spectrometer (SHS) used for hydroxyl radical OH detection with limb observations from a satellite is presented. The SHS instrument, consisting of two 1-D imaging channels in two orthogonal observation directions at different time for the same volume with the aid of the movement of the satellite platform, can get the three-dimensional distribution of OH with a series of observation radiation intensities. Each channel, including cylindrical telescope, collimating lens, a monolithic spatial interference unit and imaging lens, has vertical spatial resolution without scanning parts, and it can be field widened with prisms in the interferometer arms to increase the throughout of the instrument. An optical system for measuring mesospheric OH with a spectral resolution of 0.01 nm over the 1.6 nm ultraviolet passband 308.2-309.8 nm is designed, and its optimizing process and results are shown. The ground-based experiment with the prototype instrument is carried out with the use of the light from the sun tracker. The measurement results show that the optical system design meet the needs of OH detection requirements, and the instrument experimental performances are in agreement with the theoretical spectral characteristics. Our research provides a basis for remote sensing of mesospheric atmosphere detection in satellite.

Phytoplankton concentration of functional reaction center is closely related to its growth environment and physiological state. A new method is used to analyze the phytoplankton concentration of functional reaction center based on the biological energy flow theory, in other words, based on the fluorescence kinetic parameters such as initial fluorescence efficiency (F0) and function absorption cross section (σPSII) is presented in this paper. The fluorescence parameters of Cholorella pyrenoidosa are measured under different conditions. The results show that there is a good agreement between the result of fluorescence kinetic parameter method and that of assimilation coefficient method under normal physiological conditions, and the correlation coefficient is 0.999. Compared with the assimilation coefficient method, the fluorescence kinetic parameter method is more accurate to analyze the changes in phytoplankton concentration of functional reaction center induced by photosynthetic activity (Fv/Fm) and photosynthetic unit sizes (nPSII) under abnormal physiological conditions. The phytoplankton concentration of functional reaction center obtained by the fluorescence kinetic parameter method is correlated with Fv/Fm, and the correlation coefficient is 0.920 under the short-term stress condition. The changes of nPSII caused by light can be measured by the fluorescence kinetic parameter method under the long-term light stress condition, which are consistent with the existing research results. This new method can accurately measure phytoplankton concentration of functional reaction center.

Abstract In consideration of technical characteristics of the infrared hyperspectral interferometric spectrometer, we analyze the formation mechanism of nonlinear responses of its detector. The influences of second-order and third-order nonlinear responses on spectra are studied by means of simulating interference data with high-order nonlinear errors. An iterative method is proposed to minimize the out-of-band distortion by cross-iteration, thus the correction coefficient is determined to correct the non-linear responses. The interference data observed by the blackbody at different temperatures are corrected by the cross-iteration method, and then the spectra are recovered. The spectral response of the undisturbed wavenumber is fitted to the blackbody radiance. The results show that the second-order nonlinearity response has a major impact on the out-of-band data, while the third-order nonlinearity response mainly affects the in-band data. So the in-band data will still have residual errors when only the second-order nonlinear response is corrected. The cross-iterative method can correct the nonlinear response of the detector, and the accuracy produced by the third-order nonlinearity correction is improved about 7.26% compared to the second-order correction. The corrected goodness of fit is improved by about 0.4%, and the corrected interference data are more accurate than before.