A non-diaphragm fiber gas pressure sensor fabricated using a femtosecond laser is proposed and experimentally demonstrated. This sensor is a type of low-precision extrinsic Fabry-Perot interferometer (EFPI) formed by sandwiching a hollow fiber (HF) between a single mode fiber and a no-core fiber. A femtosecond laser is used to drill a micro-channel on the side wall of the HF to allow the gas to enter and exit the HF. When the external gas pressure changes, the refractive index of the gas in the EFPI cavity changes. The gas pressure can be detected by determining the optical cavity length of the EFPI. The experimental results show that the pressure sensitivity in the cavity is 1.02 μm/MPa in the pressure range of 0-5 MPa, and the pressure resolution of the sensor is 2.4 kPa. This pressure sensor has the advantages such as wide measuring range, high sensitivity, high resolution, and high linearity.

This study develops a force-measurement method based on distributed optical fiber strain sensing. A thin-walled ring loading test with the fiber Bragg grating (FBG) and Brillouin optical time domain analysis (BOTDA) techniques is performed. The principles of fiber optic sensing, data fitting, and error analysis are also introduced in detail. The results obtained by FBG and BOTDA are compared and analyzed in the experiment to further verify the feasibility of proposed method. The experimental results show that the load value calculated by the BOTDA data is closer to the actual one. Furthermore, the distributed fiber optical monitoring technology has been proven to meet the accuracy requirements of data acquisition in geotechnical tests.

In this study, a high-speed digital holographic system is established for measuring the amount of burning particles. After processing the hologram, the evolution of the morphology and the fragmentation of bamboo powder are observed. Further, the ability of the system to track the three-dimensional trajectories of particles is validated. The size and speed distributions of the particles before and after combustion are compared, and the reconstruction results display high-quality images of bamboo powders having various morphologies. The statistical results denote that the number of small particles (diameter50 μm) decreases after combustion. Subsequently, the distribution range of the radial speed of the particles is extended, and the axial speed denotes an overall increase with a bimodal distribution. The statistical results are observed to agree very well with the direct observations.

A cascaded 2.5-dimensional (2.5D) convolutional neural network is proposed. The task is divided into three sub-tasks of whole tumor segmentation, tumor core segmentation and enhancing tumor segmentation, and the results are combined to generate the final result. In each sub-task, the three-dimensional (3D) images are horizontally, coronally and sagittally cropped to generate 2.5D images. The 2.5D images are fed into the proposed 2.5D V-Net for training. The 2.5D segmentation results are concatenated as the 3D results to generate the segmentation results of different sub-tasks. The results show that the average dice scores for the segmentation of whole tumor, tumor core and enhancing tumor by the proposed method are 0.9071, 0.8542, and 0.8140, respectively, which basically meet the clinic need.

An adaptive linear transformation image dehazing algorithm based on Gaussian attenuation is proposed. A linear transformation model between the minimum channel of hazy images and that of haze-free images is established. A Gaussian function using the minimum channel of hazy images is constructed to adaptively compensate the transmissivity in the bright region and improve the accuracy of transmissivity. A cross-bilateral filter is used to eliminate the texture effects of transmission, and the image is restored by the atmospheric scattering model. The experimental results show that the proposed algorithm can effectively improve color distortion in the bright regions of images and eliminate the Halo effect at the edge of depth of field. Moreover, the restored image possesses obvious details and suitable saturation.

The rapid cloud detection method in various backgrounds is studied based on the spectral reflection characteristics. The spectral reflection characteristics are combined with the texture features of the clouds, and a comprehensive cloud detection algorithm is proposed based on the combination of dynamic fractal dimension and radiation quantity characteristics. The hyperspectral remote sensing images taken by the Hyperion sensor of the EO-1 satellite are taken as examples to study the cloud-containing remote sensing images of different underlying surfaces, and the thick clouds area and thin clouds area are detected and analyzed. Compared with the two algorithms of remote sensing image cloud detection, the proposed algorithm can identify the thin cloud regions more accurately, which can greatly improve the accuracy of remote sensing image cloud detection and at the same time can meet the requirements of fast cloud detection of satellite-borne hyperspectral images.

For cracks in small bridges, a segmentation method is proposed based on a generative adversarial network. This method introduces a segmental branch into the discriminator structure and combines the generative confrontation network with the semantic segmentation network. In addition, the method is capable of super-resolution image reconstruction and segmentation. To solve the problem of small-bridge-crack segmentation, this method transforms low-resolution small-bridge-crack images into super-resolution coarse-bridge-crack images, which are then segmented. The experimental results show that the proposed method facilitates the identification of small-bridge-crack and its segmentation is accurate. Compared with the traditional segmentation method, the recall rate and mean intersection over union of this method are improved by 6% and 10%, respectively.

A method for stitching light-field images is proposed based on spatial plane segmentation and projective transformation. The spatial plane fitting and segmentation are performed on the target scene according to the depth map, and the initial segmentation result is refined using a Markov random framework model according to the color information. The corresponding segmentation regions are projected to the same viewpoint using the projective transformation matrix, which is calculated using the optical-flow information among different viewpoints. Finally, a minimum energy seam is calculated to stitch together the corresponding overlapped regions, and a Poisson optimization technique is utilized to blend the stitched area. The experimental results show that the proposed method can generate large field of view light-field images without any ray misalignment or noticeable geometric distortion.

Traditional image fusion based on multi-scale transform experiences problems such as low contrast and edge details. A fusion algorithm based on the adaptive fuzzy logic and an adaptive pulse coupled neural network (PCNN) is proposed in the nonsubsampled contourlet transform domain. For the low-frequency sub-band, the fusion is based on the adaptive fuzzy logic. For the high-frequency sub-band, the information about orientation is adaptively utilized as the linking strength of the PCNN and the edge features of the source images are adopted as the input to motivate the adaptive PCNN. Then, the sub-band coefficient is fused according to the pulse ignition amplitude. The experimental results indicate that the proposed fusion algorithm can better highlight the target information of the fusion image, provide richer background details, and achieve a better fusion effect both on the clarity of fusion images and the human vision.

In this study, we applies the non-negative matrix factorization (NMF) algorithm to space object image recognition. First, we obtain the sparse NMF algorithm by improving the iterative rules of two traditional NMF algorithms and separately apply the three algorithms to the two-dimension (2D) and (2D)2 dimensions. Then, we simulate the space optical environment and acquire multiple sets of space-object-scaling model images in the laboratory. After image preprocessing, we establish the training and the testing sample databases, and extract the features of the training samples using different NMF algorithms. Finally, the minimum distance classifier is used to classify the testing samples. The results show that the recognition rates of various NMF algorithms are all above 78%, and the maximum is up to 90%. The experimental results confirm the effectiveness of the proposed algorithm. Compared with the existing methods for space object image recognition, the NMF algorithm is advantageous owing to its high accuracy, fast speed and low resource cost.

The remote sensing data obtained from the Sentinel-1 spacecraft allows the application of the synthetic aperture radar (SAR) polarization feature data to the detection of the sea-ice drift. The difference in the usage of the horizontal-horizontal (HH) and horizontal-vertical (HV) polarization data for the detection of the sea-ice drift is discussed in terms of their advantages and disadvantages. The information detected using these two kinds of polarization data is observed to be complementary. Using feature fusion, the feature information detected using these two kinds of polarization data can be used for monitoring the sea-ice drift. The experimental results indicate that the sea-ice drift vector obtained using the proposed method is improved in terms of the spatial distribution and coverage rate, suggesting that the proposed method can be effectively applied for monitoring the sea-ice drift.

In this study, a classification method of chopped strand mat defects based on convolutional neural network is proposed. In the proposed method, the rotation, translation, and inversion are employed to expand the dataset for solving the overfitting problem caused by the small data samples in the deep convolutional neural networks. Transfer learning is employed to improve the convergence speed and generalization ability of the network. Further, the different network structures are compared, and the most optimal network structure is used to verify the database. The experimental results demonstrate that the proposed method can effectively classify the chopped strand mat defects with an accuracy rate of 93%.

A new palm vein classification method that combines a deep neural network and a random forest is proposed. First, the proposed method extracts the palm vein features using AlexNet, a pre-trained deep neural network model. Then, the principal component analysis is used to reduce the dimensionality of the extracted high-dimensional palm vein features in order to conserve storage space and reduce classification errors. Finally, the random forest is used for classification owing to its high tolerance to noise. Based on the PolyU, CASIA, and self-built databases, the test accuracies obtained are 100%, 97.00%, and 99.50%, respectively. Compared with the traditional methods, the proposed method overcomes the limitations of the manual feature extraction algorithms, effectively reduces the palm vein classification errors, and demonstrates better robustness.

In this paper, the centroid localization error of a star sensor with regard to the radius of defocus image spot and noise intensity is analyzed. The functional model for the optimal Gaussian radius about the two variables, standard deviations of centroid localization error, and noise intensity are established. Under the condition that the centroid localization precision is determined, the one-variable-function relationship between the optimal radius value of defocused Gaussian image spot and the background gray-noise intensity can be obtained. The simulation experimental results show that under the given centroid localization precision, the relationship between the optimal Gaussian radius and the noise intensity follows a functional curve, such as a parabola. After the noise intensity is determined by statistics, the optimal Gaussian radius of image spot can be obtained.

Based on spectral domain optical coherence tomography with ultra-wideband light source, one Doppler shift calibration method suitable for a practical system is proposed, in which the mirror motion in the traditional method is replaced by a moving optical-path module, thereby enabling its applications in actual situations. After acquiring a series of time-varying spectra in motion, the proposed method is used to analyze the spectral distribution range via Fourier transform. Compared with the traditional calibration method, the proposed method is more precise and the axial resolution of the system is improved by 0.34 μm. Thus, this method improves the imaging effect in actual biological tissue imaging.

In this study, a compact rainbow refractometer is developed based on the global rainbow technology. This refractometer adopts a cage structure and a completely sealed design. The overall size of the refractometer, exhibiting a length, width, and height of 0.42, 0.42, and 0.15 m, respectively, is obviously smaller than those of the traditional refractometers. Further, the performance of the rainbow refractometer is validated by the experimental measurement of the refractive index of a single-component droplet, the concentration of a two-component droplet, and the droplet temperature. The accuracy of the refractive index measured using the rainbow refractometer is verified using a deionized water spray, and the measurement error is observed to be approximately 2×10-4. The refractive indices of the water-ethanol droplets with volume fractions from 0 to 100% are measured and compared with the values in the literature, and the error sources are analyzed. The results denote that the developed compact refractometer exhibits a unique function of measuring the refractive indices of droplets; additionally, the compact refractometer possesses the advantages of small volume and high accuracy. Furthermore, it is applicable to industrial environments and exhibits a good application prospect in the measurement of spray fields.

A railway track smoothness detector based on laser reference is proposed, which can detect height irregularity, alignment irregularity and curvature. The detector uses laser as reference and uses gradienter and remote focusing system to make laser irradiate the photoelectric target vertically. The data is collected through the data acquisition unit and is transmitted to the data processing unit. The track irregularity parameters are obtained and displayed on the human-computer interface. The experimental results show that the proposed detector has high reliability and high accuracy. The detection distance can reach 200 m and the measurement accuracy is 1 mm.

A precise positioning technique is proposed based on the speeded up robust features (SURF) algorithm. By identifying the quick response (QR) code laid on the ground, this technique is used to complete positioning prediction and attitude correction. First, the acquired QR image is preprocessed, and the feature point information of the image is extracted by the SURF algorithm. The feature points of both the real-time image and the target image are then matched, and the transformation matrix between these two images is obtained by least square fitting. Finally, the transformation matrix is combined with the visual guidance model of the automatic guided vehicle (AGV), and the precise positioning of the AGV is realized. The experimental results show that for a heavy-duty AGV with a large structural size, the proposed method has robust positioning with a precision of ±1 mm.

Selective laser melting (SLM) is used to prepare 316L stainless steels. Further, the effects of laser power, scanning speed, and scanning spacing on the formation of cracks are studied in detail, the morphologies, chemical compositions, types of the precipitated phases, and grain sizes of the cracks are discussed, and the microstructures and formation mechanism of the cracks at different positions are presented. The results denote that the cracks mainly include micropore aggregation cracks, bubble aggregation cracks, and hot cracks. With the increase of linear energy density, the numbers of micropore aggregation cracks and bubble aggregation cracks are observed to initial increase and subsequent decrease; however, the number of hot cracks is observed to unidirectional increase. Under the optimal process parameters (a linear energy density of 222.2 J/m, a laser power of 200 W, and a laser scanning speed of 900 mm/s), the samples containing a small number of pores but no lacking cracks and no bubbles are obtained.

The physical properties of IN738 alloy powder prepared by the gas atomization technology are fully characterized. The particle size distributions, particle surface roughness, surface morphologies and internal structures of the powder and the selective laser melting(SLM)formed parts are analyzed. The results show that most of the IN738 alloy powder is spherical or spheroidal. The particle size distribution is normal with the powder particle sizes d10 of 14.96 μm, d50 of 28.72 μm, and d90 of 52.85 μm. The time used for 50 g powder flowing through a fluidity tester is 23.4 s. The laser absorptivity of powder increases as particle size and surface roughness increase. The surface of the SLM formed part is well lapped and the density reaches 99.3%.

The 316L stainless steel powder is formed by the selective laser melting technique, and the effects of scanning speed and hatch space on the formation specimen are investigated. When the laser power is 380 W, the layer thickness is 50 μm, the hatch space is 90-130 μm, and the scanning speed is 750 mm·s -1, the density of the formation specimen is up to 99.99%, and the yield strength, tensile strength, and elongation are 625 MPa, 537 MPa, and 38%, respectively. The scanning speed strongly influences the defect formation in the specimens. As for the formation samples, their mechanical properties can be improved and the grains can be refined by suitably increasing the scanning speed.

The microstructures and mechanical properties of laser welded Ti-22Al-27Nb alloy sheet joints are investigated under hot rolling, heat treatment of 1080 ℃/40 min/water-cooling (HT1), and heat treatment of 980 ℃/3 h/oil-cooling (HT2). The well-formed welds are obtained with the process parameters of laser power of 1500 W and welding speed of 1 m/min. The microstructures of hot rolled, HT1 and HT2 sheet joints are B2+O+α2 phase, B2 phase, and B2+O+α2 phase, respectively. From the fusion line to the base metal, the microstructures of heat affected zones for hot rolled and HT2 sheet joints are the B2, B2+α2, and B2+O+α2 phase zones, respectively, however, those of HT1 sheet joints are only single B2 phase. The tensile strengths of hot rolled sheet joints at room temperature and 650 ℃ are 1023 MPa and 675 MPa, and the elongations are 5.48% and 2.56%, respectively. In contrast, the tensile strength of HT2 sheet joints are 860 MPa and 680 MPa, and the elongations are 7.04% and 4.97%, respectively.

A kernel correlation tracking algorithm exhibiting feature-weight and scale adaptation is proposed. The histogram of oriented gradient (HOG) and the color name (CN) features of the target search area are extracted for performing adaptive weight fusion, and the target position is estimated using the peak value of the correlation filter response map of the fusion feature. Further, using the product of the peak value of the correlation filter response map and the peak sidelobe ratio of the large weighted feature as the basis for scale estimation, the rough and accurate estimations of the target scale are performed and utilized to obtain the optimal scale of the target. The results of the simulation experiments performed using the object tracking benchmark (OTB-2013) dataset show that the proposed algorithm exhibits obvious improvements in terms of tracking precision and success rate compared with other five tracking algorithms. The tracking precision and success rate obtained using the proposed algorithm are 0.799 and 0.723, respectively. Furthermore, the proposed algorithm can well adapt to the change of target scale.

A stereo-matching algorithm using weighted guided image filtering based on the Laplacian of Gaussian (LoG) operator is proposed. The algorithm calculates matching cost by fusing weighted absolute difference and gradient. Then, cost aggregation is implemented using an improved guided image filtering based on the LoG operator to ensure that the penalty parameter is self-adaptive. The disparity computation is implemented using the winner-take-all (WTA) strategy, and the final disparity map is obtained using two different interpolation methods. The experimental results show that the average mismatch rate of the proposed algorithm on the Middlebury benchmark standard dataset is 4.32%. The proposed algorithm can process both high and low texture regions effectively; thus, the mismatch rate of the disparity map is reduced.

The NiCrBSi coatings are produced by the laser cladding technique. The microstructures and phases of these coatings are analyzed. The wear tests are carried out at different temperatures. The evolution of microstructures and the wear mechanism at high temperatures of 700 ℃ and 750 ℃ are discussed. The applicable working temperature range of coatings is determined. The results show that the coatings are mainly composed of γ-Ni, Cr3C7, CrB, and Ni-B-Si eutectics. At the temperature of 700 ℃, the reinforced phases of Cr3C7 and CrB are kept stable, and the eutectics are slightly decomposed. The wear resistance is a little lower than that at room temperature, and the wear mechanism is adhesion wear and abrasion wear. In contrast, at the temperature of 750 ℃, Cr3C7 and the eutectics are obviously decomposed, but CrB is kept stable. As a result, the coating is severely worn and softened. Moreover, the fins are formed at its periphery. The coating is subjected to failure due to severe plastic deformation.

CsPbCl3 nanocrystals (NCs) are prepared by the supersaturated recrystallization process at room temperature using ferrous metal chlorides as the chlorine sources. The Cl- ions react with Cs+ and Pb2+ to form CsPbCl3 NCs. The photoluminescence of the as-obtained NCs is not completely quenched by adding Fe3+, Fe2+, Co2+, and Ni2+. However, the as-synthesized NCs emit pure blue photoluminescence excited by 365 nm ultraviolet light. Especially, the quantum efficiency is slightly higher than that of NCs using lead chloride as the chlorine source.

An optically tunable terahertz metamaterial absorber with multiple absorption bands is designed. The CST 2014 simulation software is used to simulate the structure of the designed metamaterial absorber. We designed four metallic bars with varied lengths on the substrate to realize that the perfect absorption of the designed metamaterial absorber which can be controlled from single-band to dual-band. The electric field distributions at four absorption peaks of the metamaterial absorber is simulated to further study the transmission characteristics of the metamaterial absorber. The photosensitive medium in the gap of two metallic bars is further irradiated by a pump laser to realize the optically-controlled tuning of the absorber. The simulation results show that the absorptivity of the designed metamaterial absorber at four absorption peaks all exceed 95%. The resonance mechanism of the designed structure is attributed to the overlapping of four resonance frequencies corresponding to the four metallic bars with different lengths. Hence, the perfect absorber can be dynamically controlled from four-band to dual-band.

This paper presents the design of a high-bandwidth data acquisition and storage system for 3D imaging lidar. To ensure the real-time storage for the system, the data are transmitted to the computer via PCIE 3.0 (peripheral component interface express 3.0), and the echo data are stored in a solid-state drive (SSD) large-capacity disk array. A high-speed signal synchronization trigger circuit and the field-programmable gate array (FPGA) inherent path-delay calibration algorithm are proposed to address the problem that the sampling data of the multiple analog-digital converters (ADCs) in the system are not synchronized. To accurately measure the delay between the laser path trigger pulse and the system clock, a multi-level input-output delay (IODELAY) unit time digital counter (TDC) algorithm is designed with a time resolution of up to 52 ps. The results show that the system has been verified to have a maximum storage bandwidth and capacity of 5.12 GByte/s and 24 TByte, respectively. Further, the system exhibits a strong real-time and synchronization performance for data acquisition and storage and has a high practical applicability.

A novel method for the cloud detection of ZY-3 satellite remote sensing images is proposed based on the fully convolutional neural network (FCN) combined with the conditional random field. The model of a fully convolutional neural network is optimized and the FCN after three times of upsampling (FCN-8s) is upsampled. The momentum combined adaptive algorithm is used for the acceleration of convergence by adjusting the learning rate of parameters. The fully convolutional neural network is combined with the conditional random field, the fully convolutional output image is taken as the first-order potential of the front end, and the Gaussian kernel function is used as the second-order potential of the back end. The mean-shift regional constraints are added to protect the local feature information of images and the posterior probability of the conditional random field model is inferred by the mean field algorithm. The experimental results show that the proposed cloud detection method can increase the identification accuracy rate of an image cloud region to 97.38%, which is 13.42% higher than that from FCN-8s.

High extinction ratio laser can be obtained in non-polarization-maintaining fiber by the active polarization control technology. Compared with the polarization-maintaining laser generated by the polarization-maintaining fiber laser, this method has the characteristics of simple process and low price. This paper mainly introduces the basic principle of active polarization control technology, systematically expounds the development of fiber laser active polarization control technology at home and abroad in recent years, and summarizes the general trend of active polarization control technology development. The algorithm of active polarization control and the phase-locked of polarization coherent beam combining is simply sorted out. Finally, the development direction of active polarization control and the phase-locked of polarization coherent beam combining is prospected.

The photodarkening effect is a crucial factor that influences the stability, reliability, and lifetime of the high-power Yb-doped fiber lasers. The accurate physical characterization and diagnosis of the photodarkening performance of the Yb-doped fibers are the premise and basis for understanding the photodarkening mechanisms and for establishing the photodarkening elimination technologies. This study introduces the main physical parameters of the photodarkening performance. Further, the physical diagnosis methods used to assess the photodarkening performance are also reviewed. Additionally, the advantages, disadvantages, and application scopes of these methods are analyzed. The major factors influencing the photodarkening performance are subsequently discussed. Finally, the research development trend of the physical diagnosis methods is briefly discussed with respect to the photodarkening performance of the Yb-doped fibers.

Selective laser melting (SLM) is regarded as one of the most important additive manufacturing technologies. Based on the principle of discrete stacking, a high-energy laser beam was used to melt metal powder. Subsequently, the dense three-dimensional bulk materials were formed directly via SLM. SLM has unique advantages in some aspects, such as production of complex parts, short processing time, and low cost, which has been widely used in the preparation of various alloy-based materials. This paper reviews the research status and the existing problems in producing aluminum-, titanium-, nickel-, and iron-based materials via SLM at home and abroad. The possible future research directions of SLM in the preparation of metallic materials are also prospected.

Laser additive manufacturing is mainly divided into selective laser melting and laser direct deposition, which can meet the individual needs and has unparalleled advantages in the manufacturing of complex components. The principle and methods of the laser additive manufacturing technology are briefly introduced. Through the investigation and analysis of relevant literatures at home and abroad, the treatment methods and processes corresponding to the internal defects such as residual stresses, cracks, spheroidization and pores of laser additive manufactured components are described. A method for solving defects in laser additive manufactured products is proposed.