For high-precision situation that the impact of angle error on precision is comparable with the impact of position error, a camera pose estimation algorithm that can fuse the angle and position errors is proposed. The covariance of position error is projected onto a unit sphere and fused with the covariance of angle error in the iterative process of position estimation algorithm of traditional least squares adjustment camera. The problem that the projection progress depends on the parameters to be estimated is solved with block relaxation iteration method. The weighted least squares adjustment equation is constructed using the covariance. From the adjustment equation, the estimated values of camera pose in the current iteration are obtained, which combine the position and angle errors. The proposed algorithm unifies the measurement models of the position error and angle error. Experimental results show the effectiveness of the proposed algorithm applied in orientation system for missile vehicle.

Terahertz wave at frequency range from 0.1 THz to 10 THz has advantages such as large transmission capacity, good direction, high transmission efficiency and so on. Studying the application to the communication field plays a very important role in satisfying the demand of higher transmission rate for users. The terahertz time-domain spectroscopy (THz-TDS) system is used to measure the terahertz wave transmission data from 0.1 THz to 2.4 THz in atmosphere with different humidities for a 0.6 meters transmission distance. And based on the classical model of extracting optical constants proposed by Dorney et. al., the data such as delay time, power spectrum, amplitude spectrum and absorption coefficient are obtained. The results show that the water vapor in the atmosphere has obvious absorption and attenuation effects on terahertz, and the attenuation rises with the increase of humidity, especially in absorption peak. At the same time, there is transparent window of weak attenuation which can be applied to terahertz communication.

Using the improved strong field approximation method, the effects of long-range potential and short-range potential in above-threshold ionization energy spectra are investigated and the origin of the low-energy structure is also found. Meanwhile, in order to prove whether it is reasonable to use pure molecules instead of molecular ions in calculating the differential cross sections (DCS) of electrons and molecular ions, the DCS of electrons and atomic ions are calculated in some different incident electron energies conditions. The results of theoretical calculation and experiments show that the effect of long-range potential can be neglected and using pure molecules instead of molecular ions is reasonable with large-angle scattering in the large incident electron energies condition.

A kind of thermoelectric cooling charge-coupled device (CCD) detector is designed and its performance is analyzed. The vacuum refrigeration structure, electric circuit, application software of this CCD detector and their design thoughts, functions and effects are introduced. The dark current noise and the readout noise of this detector are tested. The results show that, the readout noise is 56 e- and the dark current noise is 0.232 e-·(pixel·s)-1 at the temperature of -30 ℃ and the reading frequency of 250 kHz. The spontaneous Raman scattering signal of the ambient atmosphere is obtained by this detector. Under the temperature of -30 ℃ and the exposure time of 10 s, the relative heights of the spontaneous Raman scattering peaks for the O2, N2 and H2O are 3303, 7768, 843 ADU, respectively, in which the signal-background-ratio and the signal-noise-ratio of the N2 peak are 14.8 and 24.9, respectively. The designed CCD detector possesses a detection ability for the weak spontaneous Raman scattering signals.

A kind of novel sampled fiber Bragg grating (SFBG) sensor for temperature and axial strain sensing is proposed and demonstrated. In order to fabricate the SFBG, we make a single helical twist structure on a standard single-mode fiber by a special fiber fusion splicer, and then inscribe a fiber Bragg grating on the twist structure based on the UV-side illumination and the phase mask technique. The reflective spectrum of the SFBG has the characteristics of equal spacing and narrow bandwidth, and can be freely modified by adjustment of the twist rate of the fiber. We experimentally investigate the temperature and axial strain responses of the SFBG, and the results show that when helical twist period is P=504.0 μm (torsional rate is α=12.47 rad/mm), the period of FBG is Λ=544.6 nm, and the device length is L=5.0 mm, the measured sensitivities for temperature and axial strain are 10.12 pm/℃ and 1.12 pm/με, respectively. Compared with the counterparts, the SFBG has the advantages of simple fabrication, high flexibility, high stability and low cost. It has potential applications in multi wavelength fiber lasers and multi-channel spectral filters.

The polarization control of non-polarization maintaining fiber amplifier is an effective way to obtain high power linear polarization fiber laser output. In this paper, we introduce the principle of optical fiber amplifier polarization control based on stochastic parallel gradient descent (SPGD) algorithm, and also numerically analyze the effects of the parameters, such as algorithm performance evaluation function, influence of disturbance voltage distribution types, gain step and the disturbance amplitude on polarization control. Then we use field programmable gate logic array (FPGA) to perform SPGD algorithm, and control the polarization state of the output light. Finnaly, the obtained polarization extinction ratio of the output laser is greater than 11 dB.

We analyze the differences of static and dynamic power transfer functions (PTFs) and Q-factor transfer function (QTF) for clock-pumped four-wave mixing (FWM) fiber regenerators with the optical signal degraded by spontaneous emission noise. The influence of extinction ratio on the dynamic PTF curve is revealed. The dynamic PTF and QTF characteristics of all-optical regenerator are studied emphatically and the simulation results are in good agreement with the experimental data. It is shown that the input optical power corresponding to the PTF′s output saturation point is larger than that to the QTF′s optimal operating point by 1 dB for the given wavelength configuration and optical fiber parameters. Besides, the point is basically independent of the pump optical power. This conclusion means that the optimal operating point can be derived by measurement of the corresponding dynamic PTF curves.

Image segmentation is usually needed to extract the features in order to recognize images of terahertz inline digital holography reconstruction. The images of terahertz inline digital holography reconstruction contain various degree of noise, which affects the quality of segmentation. In this paper, the composite segmentation method based on the clipping, mirror expansion, filtering and histogram polynomial fitting is adopted for the characteristics of terahertz holographic reconstructed image. By using this method, the 2.52 THz images of terahertz inline digital holography reconstruction of gasket and gear are processed and compared step by step with the basic global threshold method and the Otsu method, and objectively evaluated with Mathews correlation coefficient. Experimental results show that the composite method can get better segmentation results.

Digital image correlation method is widely studied and applied due to its advantages, such as non-contact, whole measurement, and simple structure. The traditional calculation method of strain fields takes the local least squares fitting process of the displacement field as the core, and uses the unified subset size of the whole field to be calculated. However, there is a biggish algorithm error in the process of heterogeneous deformation measurement. Larger strain subset size will smooth local deformation gradient, while smaller strain subset size cannot effectively reduce the displacement field error. Therefore, according to the local gradient intensity of the displacement field, a novel algorithm is proposed to correct displacement fields and dynamic select the strain for calculating the subset size, and simulate the results of the speckle images. The results show that the proposed method effectively reduces the calculation error of the digital image correlation method in heterogeneous deformation, and while the system precision is improved. The dynamic subset selection algorithm based on the local gradient intensity of the displacement fields has a simple principle and high calculation accuracy.

A novel method of opening-hole virtual repairing in point cloud is proposed based on modified curve contraction flow, which aims to improve the effect of existing means. Firstly, on the basis of calculating the surface normal vector and tangent vector of the hole boundary point, the whole boundary is formed according to the feature line. Secondly, the cross-point of each boundary point is obtained by the intersection of the vertical plane of the tangent vector and the boundary, and a cross-vector is formed with the corresponding boundary point. The growth direction of the boundary point is calculated by using the weighted sum of the inward vector and the cross-vector, and the growth step is determined according to the density of the original point cloud so as to be repaired. Finally, the Laplace smoothing algorithm is utilized to smooth and uniform mesh for obtaining final repairing result. The experimental results show that the similarity of Gaussian curvature distribution of our approach improves by 26.8%, compared with the base surface mesh addition method,wave front method and geometric feature repaired method.

At present, the super-resolution reconstruction methods based on convolutional neural network have the defects of large amount of parameters, low timeliness and loss of edge detail information. In order to solve these problems, we propose a super-resolution reconstruction algorithm of multiscale convolution neural network based on edge correction. Firstly, in the training phase, we set the parameter sharing layer by using the redundancy of low frequency information, In other words, the same set of filters applied to different magnification training networks to build the multi-task learning framework. In the reconstruction phase, the edge correction coefficient of high-resolution image is learned from the sample training library. The neighborhood pixel difference is used to fuse the edge coefficient and the reconstructed high resolution image, and to correct the deviation of the edge information and make up for the missing details. Finally, according to the stochastic gradient descent and back-propagation, we use the gradient to continuously update the weight parameters to make the network reach the maximum optimization. Experimental results show that the proposed algorithm has the significant reconstruction effect, high edge sharpness, elimination of blurring and aliasing, and greatly reduces the amount of parameters through parameter sharing to meet real-time requirements.

To overcome the shortcoming of bright speck phenomenon in four order totally variational model and the non-unique optimal solution in sparse regularization model of regularization image super-resolution reconstruction, an infrared remote sensing image super-resolution reconstruction model based on total generalized variation regularization is proposed in combination with the actual demand of infrared remote sensing image super-resolution reconstruction. The advantages and feasibility are analyzed with the concept of zero-order tensor space and the relaxation solution. Combined with the self-fissility of this model, the reconstruction model is split into two sub-problems by alternating direction multiplier method. The conjugate gradient method and the fast Fourier transform method are used to solve the sub-problem in numerical solution process, respectively. From the analysis of the testing results, the proposed model has a significant improvement in the resolution of the reconstructed image for both the simulated image and the real image. The objective evaluation is better than the method used in the literature, in which the peak signal to noise ratio can are increased by 1, 0.02 and 0.1 unit, respectively.

An algorithm combining principal component analysis (PCA)-linear discriminant analysis (LDA) with support vector machine (SVM) is proposed for the real-time and robustness requirements of multi-branch path identification and tracking in the process of automated guided vehicle(AGV) visual guidance. Firstly, the image features are obtained by image preprocessing algorithm and PCA-LDA algorithm. Next, the image is identified by SVM classifier, which is optimized by gray wolf optimization algorithm. In the aspect of path tracking, the lateral deviation and course deviation are calculated by using the least square fitting method. The experimental results show that the rate of path recognition is 99.3% and real-time requirements are achieved by using the algorithm combining PCA-LDA with SVM, and the path tracking error is within 20 mm to meet the general industrial environmental needs.

The existing methods of traffic sign detection and recognition based on low dynamic range (LDR) images can achieve ideal results in good exposure environment. But they are vulnerable to the limitation of lighting and weather conditions, leading to weak robustness. For this reason, we propose a recognition method based on the high dynamic range (HDR) technology. The captured LDR images under different exposure conditions are adaptively stretched in the luminance range by the improved inverse tone mapping algorithm, generating two sub-images separately. Then an HDR image produced by the multi-exposure fusion algorithm is used instead of the original LDR images for recognition. The experimental results show that the proposed method can greatly improve the accuracy of traffic sign detection and recognition.

For image matching, the KAZE feature detection and description algorithm has demonstrated a number of advantages. However, the solution of Perona-Malik (P-M) model adopted by KAZE is not unique, and the weak edges of image are prone to be smoothed in scale spaces by nonlinear diffusion filter function when the feature points are detected. To overcome these problems, an improved KAZE feature detection and description algorithm for image matching (CKAZE) is proposed. Firstly, an adaptive diffusion filter is built based on the principle of KAZE and energy functional. Then, the solution uniqueness and the edge preserving capacity of the proposed adaptive diffusion filter function are studied during filtering process. Finally, the CKAZE is constructed and its performance is validated through image matching experiments on Mikolajczyk benchmark image dataset. The results demonstrate that the correct rates of feature matching through CKAZE is 4.555%, 2.138%, 0.656% and 1.981% higher, respectively, than those by KAZE for Gauss blurring, illumination, rotation zoom and visual transformation, which indicate that the accuracy of feature detection and description is improved by CKAZE.

Aiming at the problems in content-based image retrieval, such as inconsistence between low-level visual features and the user′s high-level semantics for image understanding, low image retrieval accuracy, and the inability of a single distance measurement method for complete reflection of the similarity degree between images, we propose a remote sensing image retrieval method based on improved fuzzy C-means clustering and convolutional neural network (CNN). This method makes full use of the characteristics of remote sensing images. It adaptively processes the noise of remote sensing images by using Retinex algorithm, and uses CNN to supervise the remote sensing images by multi-layer neural network to extract remote sensing image features. Besides, the modified fuzzy C-means clustering is adopted for feature clustering analysis. Meanwhile, the top-k sorting algorithm which combines the quick sorting algorithm with the distance position weights is applied to improve the retrieval accuracy of the remote sensing images. Experimental results show that this method can significantly improve the performance of remote sensing image retrieval.

In video quality assessment, most researchers manually extract the features first, and then use machine learning to predict video quality score, which leads to unideal result. Since the VGG-16 net has excellent robustness in feature extraction, we use the network model and migrate parameters to construct the end-to-end video quality assessment network. The experimental results on LIVE video database show that the assessment score of this method is consistent with the subjective assessment score, and its assessment indexes of Spearman rank correlation coefficient and Pearson correlation coefficient reached 0.867 and 0.843, respectively, which indicated that the performance of the proposed method is better than most of the current video quality assessment algorithms based on manual feature extraction.

The problem of image reconstruction in the sparse system of radio frequency tomography with multiple scattering path is presented, and a self-adaptive sparse reconstruction method based on subspace tracking is proposed. The initial value and step length of the sparse degree are dynamically adjusted according to the characteristics of the target signal. And the attenuation coefficient of multipath linear model is sparse by using the subspace tracking algorithm. In the process of reconstruction, the supporting set is updated by the sparse estimation to reconstruct the target image. Compared with the other reconstruction algorithms, this method can effectively reduce the influence of the ghost on image definition and realize the clear reconstruction of tomography with unknown sparse. The reconstruction of the system and the probability of the ghost are presented. The experimental results show that the proposed algorithm can accurately estimate sparsity and the high precision image with low calculation amount, which can be used in other fields of radio frequency imaging.

The numerical simulation shows that the signal of field change detection is sensitive to the optical thickness and effective scale of the cloud in the range of 5° of the forward scattering angle of the sun. We develop a solar photometer (VFOVSP) based on image tracking and automatic fast changing field of view. It can quickly measure the direct radiation of different fields of view and provide a new technical mean for ground measurement of cirrus. The instrument makes use of image tracking technology to solve the problem of four quadrant tracking failure under the condition of thin cloud. A programmable variable field aperture is used to measure the different fields of view in a short time. The aerosol optical thickness (AOT) measured by VFOVSP is compared with that by a solar photometer (POMO2). The root mean square error of the AOT measured by VFOVSP is less than 0.5%, which indicates that the instrument has a high accuracy in the aerosol measurement. The change of the field ratio is analyzed under the different weather conditions. The change of the ratio is related to the particle size and the optical thickness, which provides possibility for retrieving optical properties of cirrus clouds.

The factors causing the positioning error of mobile instrument of vision measurement machine are analyzed. The positioning accuracy results are compared under the conditions of line ruler compensating the positioning error and without error compensation. In order to overcome the aiming error and alignment error caused by manual sampling when using line ruler compensation, we propose a method of using a template with array solid circles as the compensation standard precision parts to compensate for planar comprehensive positioning error. The method firstly carries out a template alignment. The solid circle image is obtained by image acquisition. And the circle center coordinate is calculated and drawn in the collected image in real time by image processing method. The circle center is marked, which is used as alignment reference to adjust the standard template position on the stage in order to make the motion coordinate system and the template coordinate system parallel. After alignment, the machine is move to the position where the circle center of each array is coincident with the center of the image. The corresponding coordinate point pair is recorded. And the fitting coefficient is obtained through the polynomial fitting method and used for machine positioning error compensation. The method is tested on the VMS300 vision measuring machine. After the compensation, the positioning accuracy of the instrument reaches 2 μm, and the accuracy is effectively improved.

The non-contact measurement of space surface profile based on line laser is an effective method for the accurate on-line dimension detection of present industrial parts. In order to improve the non-contact triangulation accuracy, we need extract the section characteristic line of measured contour accurately. However, for the surface of industrial parts with irregular surface, it is difficult to express the contour feature accurately with a simple smoothing processing method. Therefore, a polynomial curve fitting method is proposed, which takes the contour feature points as the boundary points and then puts all the points together to produce a continuous contour curve. By comparing the collected data set of tested objects and corresponding fitted curves, we find that the proposed method can not only solve the jagged problem generated in the laser line extraction process, but also can improve the measurement accuracy. The fitted curve can truly express contour features of tested surface, which has important practical value.

A rotation detection system of three-dimensional (3D) joint surface profile based on projection light knife is designed. The rapid calibration of the detection system is realized by the use of the camera pinhole model. The center-line of the projection light knife is extracted through the optimized center of gravity algorithm, and thus the contour line of the joint surface is acquired. The oblique correction of section profile is realized by the analysis of the angle between the projection light knife plane and the rotation axis. The 3D profile data of joint surface are accurately reconstructed. The results show that the detection system possesses the advantages such as non-contact and high precision, which can be used to solve the detection problem of prosthetic joint surface profile.

An autonomous measurement method of height based on the asynchronous vision of a single feature point is proposed, in which the internal and external matrix are calibrated separately and thus the computational efficiency is high. This method is verified in the constructed flight simulation device with six degrees of freedom and the results show that, below the testing height of 2 m, the absolute measurement error is less than 0.02 m and the relative error is superior to 1.5%. This method requires a less computing load, and is particularly suitable to the application of fast decoding in planes, and can also be extended to the measurement application of the ground target positioning.

Based on the working principle of pulsed laser ranging, the general equation of the light echo power of laser ranging for the flight targets is derived. Based on the pointing error theory of the target detection and tracking system and the ranging equation, a computation model of the detection probability for the range system is established. The influences of the laser divergence angle, the system tracking precision, the target scale, the detection distance and the system receiving aperture on the light echo stability of pulsed laser ranging are simulated and analyzed by the Matlab simulation software. A system of pulsed laser ranging for the flight targets is designed, the system parameters are determined according to the theoretical analysis results, and the experimental platform is built. The detection performance of this system is confirmed by the tracking and ranging experiment for the unmanned aerial vehicles. The effectiveness of this model is verified by the comparison between the detection probability and the theoretical derivation one for the target boards with different sizes.

In order to gain the lifting posture parameters of optomechanical module of amplifier real-timely, we research a method used to calculate the field installation lift attitude of optomechanical module by inputting the load mass and lifting height simultaneously based on existing optomechanical module field installation device. The structural mechanics model based on structure information of existing lift device and load information is simplified reasonably and established to calculate the lifting posture parameters of optomechanical module in the method. In the following, solution formula including two parameters namely load mass and lift height is derived under condition of matrix displacement method adopted. The validity of the calculation formula of the amplifier optical module lifting posture is verified through the software simulation and the attitude testing of the lifting platform of optomechanical module of amplifier with laser tracker.

In order to improve the forming quality of medical Ti6Al4V titanium alloys by selective laser melting (SLM), the process parameters of laser partition scanning are optimized. By the orthogonal experiment and single factor experiment, the microstructure characteristic, efficiency of space filling, dimensional accuracy and surface quality of formed parts under the partition scanning strategy and the traditional overall scanning strategy are compared and analyzed. The typical surface morphologies of medical Ti6Al4V titanium alloys formed by SLM are summarized. The research results show that the partition scanning strategy can not only increase the dimensional accuracy and surface quality of titanium alloys, but also significantly improve the forming effect of overhanging surfaces.

The S136 mold steel samples with different Cr contents are fabricated by the selective laser melting (SLM) technique and the effect of Cr content on microstructures and performances of mold steels is investigated. The results show that the addition of Cr makes the numbers of voids, cracks and other defects in samples increase, which results in the reduction of relative density. With the increase of Cr content, the martensite content in the samples decreases first and then increases. Benefitting from the refinement of Cr, the higher the Cr content, the finer the microstructures of the samples, but the smaller the microhardnesses. The addition of Cr can generate a dense passivation film on the surface of samples to improve their anti-corrosion resistance. However, when the Cr content is too high, the corrosion resistance is reduced due to the increasing number of cracks.

The titanium alloy samples are formed by selective laser melting (SLM) based on different scanning strategies, and the effect of scanning strategy on forming precision is analyzed. The results show that the roughness on lateral surface of forming part is strongly affected by the scanning strategy. The roughness on lateral surface of forming part by one-way scanning or Z-scanning strategies parallel to the scanning direction is better than that on the lateral surface perpendicular to the scanning direction. The degree of bending curvature of forming part by the spiral scanning strategy is relatively small and its shape structure is relatively accurate. Compared with the design model, the length and width of forming part by these three scanning strategies are reduced by 1-2 mm, while the other sizes are close. The whole forming precision of forming part by spiral scanning strategy is better than those by one-way scanning and Z-scanning strategies.

A femtosecond pulsed laser with wavelength of 800 nm is used to process micro-pits on the surface of cemented carbide YG6. The morphology parameters and surface contact angles of these micro-pits are measured to obtain the optimal processing parameters. The effects of micro-pit shape, distribution density and single micro-pit area on the surface contact angle are analyzed. Based on the middle state theory of Wenzel and Cassie models, the influence mechanism of micro-pit shape on surface contact angle is analyzed. The results show that the micro-pit surface morphology is relatively good and there exists a linear relationship between the micro-pit depth and number of scanning when the average power is lower than 200 mW. The micro-pit surface contact angle decreases with the increase of micro-pit density and increases with the increase of single micro-pit area. The shapes of micro-pits in the order of excellence of surface wettability are as follows: regular triangle, regular square, regular hexagon and circle.

The NiAl-SiC composite coatings with different SiC contents are prepared on the surfaces of 1Cr17Ni2 steels by the laser cladding technique, and the phases, microstructures, chemical compositions and friction properties of the cladding layers are investigated. The results show that the cladding layers mainly contain AlFe0.23Ni0.77 and AlNi phases. The NiAl-SiC composite coatings have some defects, such as micro cracks, voids and falling off. The average friction coefficients of the cladding layer with a SiC mass fraction of 3% and the substrate are 0.731 and 1.260, respectively, indicating that the wear resistance property of the cladding layer is superior to that of the substrate.

The output characteristics of the dual-frequency laser (DFL) based on the Nd∶YVO4/Nd∶GdVO4 combined crystal have been investigated experimentally. In the experiment, the pump current at 14.5 A is set, the heat sink temperature of the combined crystal rises from 5 ℃ to 40 ℃ with 5 ℃ intervals, and a DFL signal with super-large frequency difference up to 310 GHz is achieved. The experimental results show that the DFL signal power has a negative correlation with the heat sink temperature. The fitting rates of the left and right peaks power with heat sink temperature are -0.0190 ℃-1 and -0.0082 ℃-1, respectively. The DFL reaches power balance at the heat sink temperature of 32.36 ℃. In addition, the experiment also shows that the wavelengths of the DFL signal are red-shifted linearly with the heat sink temperature increasing. The measured red-shift rates of the left peak and right peak are 9.70 pm·℃-1 and 6.12 pm·℃-1, respectively.

Aiming at the problem of monocular camera pose estimation under the matching relations of unknown feature points, a new camera pose estimation algorithm is proposed based on the Softassign algorithm. Based on the advantages of orthogonal iteration and Softassign algorithm, the collinear error of the object space in three-dimensional (3D) feature points and two-dimensional (2D) feature points is used as the basis for determining the matching relationship between feature points and the objective function value for calculating the camera pose. A iteratively way determines the matching relationship of feature points and the pose of the camera. It can handle the case where the 3D/2D feature points are in one-to-one correspondence, and the case which occlude 3D feature points and erroneous 2D features. The results of the synthetic image experiments show that the success rate of this algorithm is over 82% in one case with image noise, occlusion and clutter. The experiment of real image can verify the property of algorithm.

A kernel correlation filter (KCF) with step-by-step association framework is proposed to aim at the various challenges, such as camera sudden movement, occlusion, false detection and appearance similarity in multi-target tracking algorithms. Firstly, the accurate detection results are obtained by utilizing a target detector based on the convolutional neural network. Then, a fast tracker based on the KCF algorithm is established for each target by weighted fusion of the tracking results of the three features to predict the motion state of the target. In addition, in order to effectively reduce the number of fragmented trajectories, this algorithm associates the trajectories step by step through the confidence of tracklets, and uses the online random fern to re-detect the target in the case of occlusion. Finally, the scale in the KCF algorithm is adaptively updated by using the associated successful detection information. Experimental results illustrate that the proposed algorithm displays powerful and efficient tracking performance under various complicated conditions compared with the existing excellent algorithms.

Currently, the commonly used multi-angle fusion three-dimensional (3D) reconstruction algorithm mainly includes the iterative closest point (ICP) algorithm and the structure from motion (SFM) algorithm. Aiming at the shortcomings of the above algorithms, we propose a multi-angle fusion 3D reconstruction algorithm with binocular stereo based on ICP and SFM. Firstly, the n groups of photos are taken around the target with binocular cameras by using the SFM algorithm. Then, we manually select the matching feature points of target in each group of binocular images, and calculate the 3D coordinate of matching feature points to generate the n groups of 3D point cloud. Subsequently, the rotation matrix and translation vector within the n groups of 3D point cloud are calculated and optimized by ICP algorithm. Finally, the n groups of 3D point are fused, and the 3D geometry of target is recovered by Delaunay triangle. The experimental results show that, the proposed algorithm takes advantages of the binocular cameras, overcomes the disadvantages of ICP and SFM algorithm, and has good vision effect for 3D reconstruction of target.

A major challenge of pedestrian detection is to detect different-scale pedestrians in complicated scenarios, especially for far-scale pedestrians. Motivated by the experiment that pedestrians with different scales exhibit dramatically different visual features, we propose in this paper a multi-scale aware pedestrian detection algorithm. Firstly, we introduce deformable convolutional layers in full convolutional network structure to expand the receptive field of feature maps. Secondly, we use cascade-region proposal network to extract multi-scale pedestrian proposals and introduce discriminant strategy, and define a multi-scale discriminant layer to distinguish pedestrian proposals category. Finally, we construct a multi-scale aware network, use the soft non-maximum suppression algorithm to fuse the output of classification score and regression offsets by each sensing network to generate final pedestrian detection regions. The experiments show that there is low detection error on the datasets Caltech and ETH, and the proposed algorithm is better than the current detection algorithms in terms of detection accuracy and works particularly well with far-scale pedestrians.

In the task of unmanned vehicle infrared video colorization, considering the uniqueness of a single frame and the continuity of the entire infrared video, a dual-channel cycle-consistent adversarial network (DcCCAN) based colorization method is proposed. The dual-channel generation network we proposed is on the basis of cycle-consistent adversarial network (CCAN) and has good image feature extraction ability, which can automatically extract the features of the frame in the video, and at the same time can extract the features of the previous frame generated. By joint training of adversarial loss and cycle-consistent loss, the function from infrared domain image to color domain image can be learned by unsupervised learning methods and the colorization of the infrared video can be realized. The experimental results show that the proposed method can provide natural color information and texture information for the infrared images in the video, and meet the real-time requirements.

To solve the problem that jitter video moving object detection is not accurate, we propose a moving object detection method based on block gray projection, background difference and continuous inter-frame difference. By dividing the image frames into block processing, the algorithm combines the discrete decision mechanism to remove the target regions with low gray gradient and the local motion, so as to improve the accuracy of global motion vector estimation. A cross correlation calculation is done for the gray projection of the block area and the image correction is completed. For the corrected sequence frames, we propose a fusion strategy based on background difference and continuous three frame difference method, which can deal with and enhance the moving target area. The selfadaptive thresholding segmentation for the differential image smoothing fusion processing and Otsu method is used to detect foreground moving targets. In jitter video sequence experiment, compared with different algorithms, the proposed algorithm can effectively detect moving targets in the jitter scene, and ensure better detection results and faster detecting speed.

The geometric structure, energy band structure, density of electronic state, and optical properties of rare-earth elements La, Y single-doped and co-doped Ca2Si are calculated by the first principles. The calculation results show that, the La single doping and La/Y codoping make the cell volume of Ca2Si increase and the bandgap become narrow, while the Y doping does make the cell volume decrease and the bandgap become wide. After the La, Y single doping and codoping, the Fermi level enters the conduction band, and Ca2Si becomes an n-type semiconductor, and the dielectric function, extinction coefficient, and absorption edge of Ca2Si all move towards the low energy direction, and both the refractivity and reflectivity increase.

In order to obtain a tunable ultra-broadband absorption spectrum under transverse electric (TE) waves and a high reflectivity under transverse magnetic (TM) waves, an electromagnetic absorber is designed based on the plasma metamaterials and the lumped resistors. Its absorptivity, reflectivity, surface electric field diagram, surface current diagram and energy loss diagram are calculated by the full wave simulation method. The influences of its structural parameters and resistance on the absorptivity and reflectivity are investigated. The research results show that, by means of the excitation for different plasma resonance ranges, not only the absorption property is improved, but also the tunable absorption spectrum can be obtained. As for the designed electromagnetic absorber, not only the ultra-broadband absorption of TE waves can be realized, but also the polarization splitting of TE and TM waves can be achieved.

The geometrical structures, electronic structures and optical properties of the two-dimensional (2D) SiC doped with P with different concentrations are investigated by the first principle method. The results show that, the lattice constant and the bandgap of 2D SiC doped with P gradually decrease with the increase of P doping concentration. The valence band is mainly composed of the hybridization of the electrons of C-2p, Si-3p and P-3p states, while the conduction band is mainly composed of the electrons of Si-3p state. The P doping weakens the covalency and increases the ionic property for the C—Si bond. The P doping expands the optical absorption range of 2D SiC, and makes the absorption coefficient and the refractive index increase with the increase of doping concentration, which indicates that P doping can effectively improve the absorption for both the visible and infrared light.

At present, two theoretical analysis methods, nonlinear coupled-mode theory and nonlinear Lugiato-Lefever equation (LLE) model, are used to investigate the generation of Kerr optical frequency comb in micro-ring resonators. In previous theoretical studies, only dispersion and optical nonlinear effects were generally considered, and the thermal effects in micro-ring resonators under strong pumping were ignored. In this paper, in view of the thermal effect in the micro-ring resonator induced by the traveling light beam, the nonlinear LLE is modified, and based on the modified equation, the generation of Kerr optical frequency comb is investigated for the two cases of fixing and adjusting the initial detuning. Generation mechanisms of the Kerr optical frequency comb based on these two different ways are analyzed. The characteristics of the Kerr optical frequency comb generated by the two ways are compared, and the reason that causes their difference is investigated.

An infrared-visible confocal optical system for simultaneous extraction of human fingerprint and finger vein images is designed, which consists of ten spherical lenses with system focal length of 10 mm, total field of view of 34° and F number of 2.4. In order to reduce the instrument volume, the conjugate distance of this system is controlled within 120 mm. After optimization by the ZEMAX software, the design results show that the lights with different wavelengths strictly image on the same image plane. The value of modulation transfer function at a spatial frequency of 164 lp/mm required by this system is greater than 0.4. The sizes for all defocused spots in view field are smaller than the smallest pixel size of the image device and the distortion is less than 1%. The tolerance analysis of the system shows that this design meets the manufacturing requirements.

A novel structure for the focus-variable liquid lens driven by the bending vibration of thin films is proposed, and the mechanism of this proposed novel lens is theoretically analyzed and experimentally investigated. The results show that, under the driving of the bending vibration of the thin films, the pressure in this proposed lens is large in its center and small in its edge, which indicates that this lens is a convex lens. Meanwhile, its focal length decreases with the increase of input voltage, and the focal length range is 10-60 mm. This proposed lens has a simple and thin structure, and a larger zoom range than that of other liquid zoom lenses, and thus has the strong practical application value.

Nowadays, cell phone lens is developing towards high resolution, large aperture and ultra-thin thickness. According to the design requirement of cell phone lens, we study the relationship between the design parameters and constrain conditions and objective function in Delano diagram. With the Mente Carlo-particle swarm optimization (PSO) hybrid optimization algorithm, the automatic generation of the cell phone lens initial configuration is realized. Taking the proposed method, we design a 5 pieces-twenty million pixels and ultra-thin tele-lens of cell phone. The design results show that the modulation transfer function (MTF) is greater than 60% within 0.8 field of view at the 1/4 Nyquist frequency 125 lp/mm. Compared with the 6 pieces lens with the same design requirement parameters, which is based on PWC method, the yield rate is increased by 10% at the edge of view by the tolerance simulation. By the TV line test of the 5 pieces-twenty million pixels lens after the fabrication, the resolving power of 2300 LW/PH and 1700 LW/PH at the centre and the 0.8 field of view are obtained. The outcome of design, manufacturing and TV line test show that, the solving method of initial configuration for cell phone lens based on Delano diagram can help to improve the design efficiency and get higher yield.

All-optical OR gate is a necessary basic technology in all-optical logic signal processing, and the output optical wave of the OR gate based on periodically poled lithium niobate (PPLN) waveguide in former schemes is mixing wave. In this paper, the all-optical OR gate with single wavelength which consists of three cascaded PPLN waveguides is proposed based on sum-frequency generation and difference-frequency generation (SFG+DFG) in quasi-phase-matching consideration. The signal waveforms and eye-diagrams are obtained by numerical calculation and simulation, and the performance of all-optical OR gate with single wavelength is analyzed by the parameters such as extinction ratio, pulse width and delay time of peak power. The results demonstrate that the logic function of OR gate with single wavelength can be achieved well by this configuration, so that its output optical wave can be used directly in optical domain, thus the connection of optical logical devices can be improved, the speed of processing can be increased, and the critical foundation of developing new all-optical logical devices can be provided.

In order to improve the light extraction efficiency of flip-chip light-emitting diode (LED), we propose to prepare a SiO2 dielectric grating on the sapphire substrate surface and form a surface grating flip-chip LED structure. The surface grating flip-chip LED model is established by the CAD module of RSOFT software. Then, the LED module of RSOFT software is used to simulate and optimize the surface grating flip-chip LED. Simulation optimization and theoretical analysis show that, when p-GaN layer thickness hp=220 nm, n-GaN layer thickness hn=100 nm, sapphire substrate thickness hs=130 nm, grating period p=260 nm, grating thickness hg=20 nm, grating duty cycle f=0.02, the light extraction efficiency of the surface grating flip-chip LED can reach to 49.12%, compared to the best normal flip-chip LED light extraction efficiency (30.56%), the efficiency raises by 63%. The research can provide theoretical research methods for the future design of LED with high light extraction efficiency, and provide theoretical guidance for the preparation of devices.

We design an electronic control device based on graphene metamaterial with grating structure and study its polarization modulation properties in the terahertz region by finite element analysis method. The results show that in a wide band range of 0.1-2.0 THz, a broadband modulation occurs in the transmission of polarized terahertz wave parallel to the band (TE). It can be attributed to the intraband determined Drude conductivity of graphene. However, because of the plasma effect of graphene metamaterial, the transmission of polarized terahertz wave perpendicular to the band (TM) is almost transparent in the same terahertz region. More calculations also prove that with the increase of Fermi level of graphene, the modulation of TE will be enhanced largely while the plasma peak of TM will be blue shift. Meanwhile, the modulation can be further enhanced by the increase of the area ratio of graphene and the number of graphene layers in the device. In addition, the increase of the band width can cause a red shift of the plasma peak. With 2 μm period and 1.5 μm band width, a tunable maximum polarization degree of 0.89 can be achieved with 6 layer graphene device at Fermi level of 0.8 eV. This work paves a way for new graphene based terahertz polarization modulator.

For the most commonly used Mie scattering laser radar in the environmental field, through the process discussion of the inversion of aerosol optical characteristics, we analyze the uncertain factors which affect the accuracy of inversion results. The result shows that in order to obtain the results with high stability and reliable detection, the system use a reasonable signal denoising method to set up accumulative frequency, signal denoising method, aerosol extinction backscatter ratio, calibration height and calibration value. In addition, the system needs to calibrate the inversion algorithm regularly, including geometric overlap factor calibration, balloon borne extinction instrument calibration, Rayleigh scattering calibration, visibility meter calibration, sun photometer calibration, humidity calibration, particle concentration calibration and so on. With this series of configuration and calibration, multiple lidars data can achieve high consistency and accuracy.

Conventional fixed imaging sensors are mostly used to analyze the changes of spectrum resulting from changes in the biochemical parameters of the ground objects. However, the computer simulation model used to study the bidirectional reflectance characteristics can not complete the simulation of multiple types of ground objects because of the large amount of view factors that need to be calculated when constructing a scene. Therefore, there is less connection between them. To deal with these problems, based on the Radiosity Applicable to Porous Individual objects (RAPID), we simulate the reflectance of the Yuhuayuan area in Changchun City and analyze the sensitivity of the reflectance to the environment factors. The results show that the field of view of the sensor has a strong influence on hot spot. The solar zenith angle and sky light ratio have a strong influence on all directions of the visible and near-infrared bands. Simulating the influence of environmental factors on the imaging spectrum is a good way to provide the basis for inversion of the ground objects of the biochemical parameters of fixed imaging sensor.

Injection-seeded all-solid-state single-frequency pulse lasers (SFPLs) with narrow linewidth, long coherent length, and high power are desirable laser sources applied in fields such as Doppler wind lidar, gravitational wave detection, and laser spectroscopy. They have wide applications in military and civil fields. The research advancement of the injection-seeded all-solid-state SFPLs and the corresponding techniques are reviewed around injection seeding techniques and cavity length controlling techniques. The prospects of them are discussed.

With development of the optical fiber sensing technology, the distributed optical fiber sensing technology has been used in the structure crack monitoring in recent years, mainly including the quasi-distributed optical fiber sensing technology of fiber Bragg grating, and distributed fiber optic crack sensing technologies such as optical time domain reflectometry, Brillouin optical time domain reflectometry, Brillouin optical time domain analysis, Brillouin optical correlation domain analysis, and the latest high spatial-resolution pulse pre-pump Brillouin optical time domain analysis and differential pulse-width pair Brillouin optical time domain analysis. Based on the applications of these techniques in structural crack monitoring, their respective advantages and disadvantages and applications of distributed fiber optic crack sensing all through the world are summarized. Additionally, mechanics, distorted light signal and experimental researches are analyzed in the distributed optical fiber crack monitoring, and some opinions on the issues are proposed.

In recent twenty years, the quantum positioning system has developed rapidly as a new type of navigation technology because of its unique advantages in information transmission. After the introduction of the satellite navigation and inertial navigation systems and their respective problems, we mainly elaborate the proposition, fundamental principles, advantages and classification of quantum navigation. The research status of quantum navigation is also summarized. Finally, we put forward our views on current problems and prospects of the quantum navigation system.

The Scheimpflug lidar (SLidar) technique has been recently developed for atmospheric remote sensing. By utilizing high-power continuous-wave laser diodes as laser sources and area image sensors as detectors, the SLidar technique can measure range-resolved atmospheric backscattering signal when the optical layout satisfies the Scheimpflug principle. This paper presents the principle, features, system architecture as well as the signal processing methods of the SLidar technique. The developments and applications of the SLidar technique in atmospheric aerosol sensing and gas monitoring are summarized in detail. The challenges of the SLidar technique are also discussed. Finally, the future work and the outlook of the SLidar technique are presented.

In the analysis of Raman spectroscopy, the sample and contaminants often produce strong fluorescence signal which seriously affects the detection of Raman spectrum signal produced by the sample, and limits the application of the technology. In recent years, a number of methods have been developed to solve the problem of fluorescence interference. In this paper, the fundamental principle, implementation and performance characteristics of various methods were reviewed and summarized, and the application of several common fluorescence suppression methods(fluorescence quenching method, light bleaching method, ultraviolet light excitation method, infrared light excitation method, shifted-excitation Raman difference spectroscopy and wavelet transform)were also concretely analyzed in different fields including excitation wavelength, fluorescence source, variation of fluorescence signal, the effect of fluorescence suppression, the influence of the Raman signal and so on. It not only shows the advantages of each method, but also illustrates its disadvantages.

Aiming at the effect of surface scattering on the imaging performance for astronomical telescope optical system, we propose a method of calculating the encircled energy in image plane based on the power spectrum density (PSD) of the mirror surface. The relationship between the intensity distribution of the image surface and the power spectral density of the mirror surface is proposed based on Harvey-Shack scattering model. The propagation of the scattering in the off-axis three mirror optical system on each reflecting surface is described. The relationship between irradiance distribution and the ratio of the effective root-mean-square (RMS) to wavelength is discussed. The surface power spectral density in different spatial frequency bands of 1.5 m-diameter optical surface is obtained by different measurements, and the full-band one-dimensional and two-dimensional power spectral density are fitted by k-correlation model. It shows that the effective RMS of surface shape on the processed sample is 13.7 nm. Then the encircled energy distribution of the image plane at different wavelengths is compared. The processing requirements of a large space astronomical telescope in our country including scattering conditions are given: the effective RMS of mirror surface must be less than 10.3 nm, and the RMS of low frequency error needs to be less than 8 nm.

The fluorescence spectra of the aromatic hydrocarbons are investigated by the alternating trilinear decomposition (ATLD) method. The complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) decomposition and the wavelet soft threshold algorithm are used for spectral denoising, and the signal-to-noise ratio of 28.51 and the root-mean-square error of 3.52×103 are obtained after the treatment. With the ATLD algorithm, the 1-naphthol, 2-naphthol and naphthalene are successfully recognized with a recovery rate of 96%-103.3%, 97.24%-103.9%, and 97.2%-103.6%, respectively. The results show that the ATLD algorithm possesses a good predictability for the aromatic hydrocarbons.

Laser induced breakdown spectroscopy (LIBS) is proposed to discriminate origins of navel oranges. A series of 10 origins of navel oranges which from four counties of Ganzhou in Jiangxi and other six provinces are selected. After cleaning the skin of navel oranges, the images of plasma are collected by ICCD camera and LIBS spectra are obtained by spectrometers to qualitatively analyze the feasibility of origin identification of navel oranges. Furthermore, fifteen points smoothing and multiple scattering correction (15SM+MSC) are utilized to preprocess the LIBS data. And principal component analysis (PCA) and PCA-MLP(multi-layer perceptron) are used to discriminate the origins of navel oranges. The investigation shows that the PCA-MLP model coupled with suitable data processing methods can not only identify origins of navel oranges which from seven provinces in a large area, but also identify origins of navel oranges which from four counties within a small area. The accuracy is 97.8% for origins of navel oranges which from seven provinces by evaluating LIBS spectra in training set, and 95.3% in test set. And the accuracy is 100% for origins of navel oranges which from four counties by evaluating LIBS spectra in training set, and 96.2% in test set. It is potential in differentiating origins of navel oranges by analyzing LIBS spectra.

In order to solve the distribution and content of each substance in hyperspectral images, we introduce the hyperspectral classification into the endmember extraction to propose a new endmember extraction method. Firstly, the number of endmembers is determined by the virtual dimension. And the thought of hyperspectral unsupervised classification is performed by K-means clustering algorithm which classifies each pixel into classifications. Then the pixel with the largest spectral value is extracted from each kind of class. According to the theory of simplex, the pixels of hyperspectral image are used to form a simplex in the high-dimensional space, and the vertexes of the largest simplex are the endmembers which are extracted. The simulation and real data have shown that this method of endmember extraction has the advantages of high efficiency and accuracy compared with the traditional method.

Laser-induced breakdown spectroscopy (LIBS) technology is an element analysis technology based on atomic emission spectroscopy and plasma emission spectroscopy. In this study, LIBS is used to detect the lead (Pb) concentrations in water. The strongest spectral line of Pb 405.8 nm is selected as the analytical line and Si 390.6 nm is used as internal standard element. The detection limit of Pb obtained by linear fitting is determined to be 7.40×10-6 . A quantitative analysis model based on biogeography-based optimization (BBO) algorithm is established. Using this model, we establish the LIBS spectra of 35 samples with different Pb concentrations. Among them, 30 sets of data are used to train the BBO quantitative analysis model, and the remaining 5 sets of data are used as test sets to evaluate the analytical ability of the model. The results show that the relative standard deviation (RSD) and the mean absolute percentage error (MAPE) of the model are quite good when using the BBO algorithm model to predict the Pb concentration in water.

The optical thin films of SiO2, TiO2 and Al2O3 are prepared on GaAs substrates by using the electron beam evaporation method under different substrate temperatures and ion source energies, respectively. The stress distributions of these thin films are measured, and the refractive indexes of these thin films prepared under different ion source energies are also tested. The results show that, the surface stress distributions of these three kinds of thin films are non-uniform, and the substrate temperature and ion source energy can be adjusted to reduce the stresses of these thin films effectively. The average minimum stresses of SiO2, TiO2 and Al2O3 thin films are 2.9, 8.4 and 25.1 MPa, respectively.

The surface condition and spectral absorption characteristics of objects can be accurately reflected by their spectral reflectance data. The information of spectral distribution distinguishes the camouflaged objects which are difficult to be found and identified by human eyes, and measures the difference between a target and the background quantitatively. According to colorimetry, there are two characteristic parameters, color difference and spectral matching coefficient. Using these two evaluation parameters as reference indices, we extracted color characteristics of the target ship and the south of the East China Sea, analyzed their spectral reflectance distribution, and evaluated the effectiveness of the camouflage system. Meanwhile, we proposed the probability formula of target recognition for more effective evaluation of ship camouflage effect and obtained corresponding experiment results. The results show that the color difference and spectral matching coefficient can be used as evaluation parameters for evaluating the camouflage effect of target ships effectively.