As an effective detection tool for observing wind field information, a coherent Doppler lidar is widely used in wind field and atmospheric observations in different scenarios. In August 2019, researchers used the coherent Doppler lidar to continuously carry out a wind profile observation experiment for 13 d at the Shenzhen Shiyan Comprehensive Meteorological Observation Base. The similarity and difference of inversion results of the five-beam, three-beam, and two-beam methods in the Doppler beam scanning mode and the influences of these methods on wind measurement accuracy and data acquisition rate are analyzed. The data comparison results show that the inversion results of the three-beam method and the two-beam method are in good agreement with the results obtained by the five-beam method. In view of the low signal-to-noise ratio of individual beams caused by the system or atmospheric conditions, such as clouds, fog, and precipitation, in actual observation of the lidar, it is necessary to flexibly select the beams to invert the wind field and increase the data acquisition rate. The average increase in detection height obtained is 100--400 m.

Differential wavefront lidar is a lidar system for measuring the atmospheric optical turbulence intensity based on wavefront differential jitter variance of a laser echo detected by the two-aperture telescope. To evaluate and optimize the detection performance of this system for atmospheric optical turbulence, numerical simulations have been carried out. According to the wave optics theory, atmospheric turbulence phase screen model, and atmospheric extinction model, the propagation of laser beam in vertical atmospheric path is simulated. Combined with the optimized design of grid sampling, the intensity distribution of laser beam at different positions along the propagation path is obtained. Based on the principle of incoherent light source imaging, the distributions of two spot images on the detector are obtained according to the backscattered light intensity distributions at different transmission paths. According to the simulation results, we find that the degree of beam wavefront distortion is deepened with the increase of turbulence intensity. The diameter of the imaging spot decreases with the increase of the detection height. When the detection height is 10 km, the diameter of the imaging spot decreases to 2.45×10 -4 m. By comparing the simulated inversion results with the results of the input HV5/7 (Hufnagel Valley 5/7) in simulation, we find that the two have high consistency, which preliminarily proves the reliability of the principle and method of atmospheric turbulence detection by using the differential wavefront lidar.

The sun glint is caused by the mirror reflection of sunlight from the sea surface. The high radiation intensity of glint saturates the sensor pixel, making the contour of the target on the sea surface lost in the sun glint, which affects the detection of the target. A method for suppressing the sun glint is proposed based on the polarization time-domain characteristics of the sun glint. The method uses the real-time polarization imaging system to polarized radiant sea surfaces in real-time and to obtain the polarization video information along three polarization directions of 0°, 45°, and 90°. Stokes vector is calculated according to the corresponding frame image with different polarization directions, and the sequence of spatial-domain glint-suppression polarization-radiation patterns is generated. The generated image sequence is fused in the time domain to suppress the sun glint. In this study, a real-time surface polarization imaging observation-experiment platform was built to conduct a detection experiment of the target surface under the glint background. The experimental results show that the signal to clutter ratio of glint suppression image obtained using the proposed method is improved compared with that of the ordinary linear polarization image; the glint suppression effect is obvious, and the saliency of the sea target in the image is enhanced.

We report the numerical simulation on the spiderlike photoelectron momentum distributions (PMDs) of a hydrogen atom ionized by a linearly polarized laser field using the semiclassical rescattering model (SRM) and the time-dependent Schr?dinger equation (TDSE), and focus on the extraction of scattering-amplitude phases. Based upon the previous research results, we draw a conclusion that the time parameters obtained from the SRM and the saddle-point theories are approximately equal. In addition, we propose a novel method for phase extraction. Moreover, we establish two first-order approximate formulas to analytically describe the scattering amplitude phase. The phase obtained by the proposed method is related to the longitudinal and transverse momenta. The new method obviously enhances the extracting precision, although it increases the computation load.

In the light-induced drift phenomenon, ground state hyperfine pumping has an obvious effect on the population at each energy level of an atom and further on the light-induced drift velocity. The research on the influencing mechanism of ground state hyperfine pumping is helpful to optimize the drift velocity and the selectivity of isotope separation. We build the atomic light-induced drift rate equations and utilize the strong collision model to describe the collision between atoms and buffer gases. The rate equations are solved numerically, and the influence of ground state hyperfine pumping on light-induced drift is investigated. The research results show that ground state hyperfine pumping has an obvious suppression effect on light-induced drift. The larger the ground state hyperfine splitting energy difference is, the stronger the suppression effect on light-induced drift is. Increasing laser power density and laser linewidth and simultaneously using the second laser to re-pump atoms populating at the other ground state hyperfine energy level to the excited state energy level can weaken the effect of ground state hyperfine pumping and enhance the light-induced drift velocity. The research results are basically consistent with the qualitative analysis based on the physical principles and also the experimental results from some related literatures.

Due to its low contrast ratio (CR), liquid crystal display (LCD) is greatly challenged by OLED. Dual-layer LCD can effectively reduce the light leakage and improve the CR. However, its two LC panels can display two layers of images, and displayed ghosts may appear in the case of oblique view, which deteriorates the display quality. To this end, we proposed a viewing-angle-compensation-based image segmentation algorithm. Then, the positions of light at different viewing angles that passed through the two LC panels were recorded and the mapping matrix was therefore established. The transmittance of the dual-layer LCDs was optimized to obtain the minimum error between the initial image and the reconstructed image at all the set viewing angles. The simulation results show that at five viewing angles, the proposed algorithm presents larger peak signal to noise ratio (PSNR) than the other existing algorithms, which proves that this algorithm can achieve a better tradeoff between image quality and viewing angles.

This study combined the Kmeans clustering algorithm with the traditional K-nearest neighbor (KNN) algorithm and proposed a Kmeans-KNN fusion algorithm suitable for energy self-sustaining indoor visible light positioning (VLP) systems. This algorithm considered both low complexity and high precision. Based on using the Kmeans clustering algorithm to divide the specially designed fingerprint library to achieve coarse positioning, the KNN algorithm was used for precise positioning. This study further introduced the proposed Kmeans-KNN fusion algorithm into an energy self-sustaining VLP system and analyzed the positioning performance of the system under different conditions. The results show that compared with the traditional KNN algorithm, the Kmeans-KNN fusion algorithm's positioning accuracy is significantly improved; the average positioning error of the system is 0.141 m. In addition, the calculation amount of the proposed algorithm is reduced by 94.7%. Therefore, the system energy consumption is significantly reduced, which is conducive to the realization of high-precision energy self-sustaining of the VLP system.

In this paper, we proposed a multi-objective optimization model of bandwidth resources in the fiber Bragg grating (FBG) sensor network which is applicable to asymmetric overlapping spectra. Then, the non-dominated sorting genetic algorithm II (NSGA-II) was used to obtain the Pareto front of bandwidth resources, from which appropriate solutions were selected to determine the Bragg wavelength allocation at each FBG node. Experiments show that this method can save at least 40% of the bandwidth of the light source. While improving the multiplexing ability of the sensor network, it minimizes the overlapping of adjacent FBG spectra, maintaining high measurement accuracy. This method provides a reference for the allocation requirements in practical engineering applications.

A novel seven-core chalcogenide glass fiber was fabricated via extrusion. Firstly, two glasses, As2Se3 and As2S3, were prepared by distillation purification and melt-quenching. Then, an improved isolated extrusion method was used to fabricate the optical fiber preforms. In addition, a novel seven-core chalcogenide glass fiber of a complete structure was obtained with the protection of the high-temperature polymer layer, with the numerical aperture (NA) in the range of 1.20 to 1.45. Its loss was measured by the cutback method, with a minimum of 1.9 dB/m@4.4 μm. Finally, pumping with femtosecond laser from an optical parametric amplifier (OPA) yielded a supercontinuum covering 1.5--12 μm in a 14-cm long seven-core fiber, demonstrating excellent nonlinearity of this fiber. In summary, this fiber shows great potential in both mid-infrared research and technical applications.

Aiming at the influence of the focus point position of optical path in the infrared solar occultation flux method alignment system on the spectral quality and inversion accuracy of the gas mass concentration, the reference coordinates are established, the geometric optical path in the interference cavity is derived, and a new infrared spectral correction algorithm is proposed in this paper. Using ZEMAX software to simulate and analyze the off-axis condition of the optical path in the interference cavity, and the results show that the change of focus point position will cause the size, position and intensity of the light spot on the mercury cadmium telluride detector. The results of field experiments show that the change of off-axis angle of the beam will cause the change of interference intensity and signal-to-noise ratio of the beam. In the case of static measurement, the average drift of SF6 wavenumber of the traditional algorithm and the algorithm are 0.2602 cm -1 and 0.1146 cm -1, respectively. In the case of dynamic measurement, the average drift of SF6 wavenumber of the traditional algorithm and the algorithm are 0.2355 cm -1 and 0.0860 cm -1, respectively, and the accuracy of inversion of gas mass concentration is higher by using the algorithm.

A holographic wave-front sensor (HWFS) is a modal wave-front sensor, which utilizes a multi-mode composite holographic element and a position sensitive photodetector to extract the amplitudes of specific Zernike modes from the incident light. However, disturbing light from high order diffraction of the multi-mode composite holographic element is inevitable. The disturbing light disperses the power of the incident light and decreases the signal-to-noise ratio of the sensor. Here, an intensity modulation method is introduced to eliminate the interference light generated by the diffraction of binary phase holographic elements. The proposed method can be used for the elimination of interference light in the holographic mode wavefront sensor, and it is suitable to other applications based on binary phase holographic elements, such as holographic imaging, phase-stepping interferometers, and beam deflectors.

In this study, a fast calculation method of the computer generated integrated color rainbow holography based on the light field image is proposed, and the color three-dimensional (3D) display of the hologram is realized through optical experiments. First, the hologram plane is divided into several continuous unit line hologram planes, and the 3D coordinates of the unit line hologram planes projection points are calculated according to the vertex coordinates of each unit line hologram plane and the virtual slit vertex coordinates. Then, using the projection point as a virtual pinhole, the 3D object is projected through this pinhole; the light field image on the plane of the unit line hologram and the phase of the spherical waves converging at the virtual pinhole are used as the object light amplitude and phase in the computational hologram, and the reference light code is used to obtain the unit line hologram. Finally, all unit line holograms are combined to form a color rainbow hologram. The experimental results show that only takes 43 min to realize a hologram with a size of 50 mm×50 mm and a resolution of 157232 pixel×157232 pixel using the method, which has broad application prospects in the fields of holographic packaging and 3D advertising.

The speed of star map recognition determines the attitude update rate. To solve the time-consuming problem of cyclic displacement in the star map recognition algorithm based on log-polar transformation, we improve the algorithm. First, we search for the star point nearest to the center of the field of view, and construct a new rectangular coordinate system. The rectangular coordinates are then converted to polar coordinates. After projecting the star coordinates onto the distance axis, we construct a feature vector of star patterns. Because the polar coordinate transformation is rotationally invariant on the distance axis, the time-consuming problem causing by the cyclic displacement is avoided. In a simulation evaluation, the average recognition time of the proposed algorithm was reduced to 8.4% that of the traditional method, but this performance was slightly degraded by positional noise. The recognition rate of the proposed algorithm was higher than that of the traditional algorithm when the data were affected by pseudostars and missing stars. Under the influence of noise, star patterns are more changeable than triangle patterns, so the ecognition rate of the proposed algorithm is generally lower than that of the triangle-recognition algorithm.

With the aid of Fourier ptychographic microscopy, we can synthesize single-frame images with more details after spectrum expansion, thus reconstructing high-resolution images in a large field of view. However, the ubiquitous aberrations in imaging systems often lead to blurry images and reduce resolution of the reconstructed images. To solve the above problems, we proposed an innovative aberration correction method based on ptychographical iterative engine. Specifically, when updating the spectrum and pupil function, we adaptively selected the optimal ratio between their current and maximum values to improve the quality of iterative reconstruction. Then, we reconstructed the simulated images with mixed aberrations by taking peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) as evaluation metrics. The simulation results showed that compared with the traditional embedded pupil recovery algorithm, our method sharply increased the PSNR and SSIM (14.9% and 1.4%, respectively). To further verify the effectiveness of our method on real images, we collected human blood cell samples for reconstruction. The reconstructed images were clear, which contributed to distinguishing the cell contours accurately.

In an automatic focimeter that uses the Hartmann method for measurement, the Hartmann diaphragm can select light and determine the measurement resolution. Its aperture and hole spacing directly affect the quality of the imaging spots on the CCD, resulting in a measurement error in the spot position. In this paper, we established an objective function model according to the arrangement of circular holes on the Hartmann diaphragm and a variable model through the diffraction and interference theory and Matlab simulation. Then, taking the enumeration method for extremum search, we designed an optimization program for the optimal aperture of circular holes. As a result, the optimal aperture of the Hartmann diaphragm in a certain measurement area was obtained. Furthermore, we developed an experimental platform for lens measurement based on the Hartmann method. Comparative experimental results prove that this method can get reliable optimal apertures.

The precise extraction of the centroids of laser footprint spots has important practical significance for improving the plane accuracy of laser altimetry by domestic satellites. This paper compared three traditional extraction algorithms of spot centroids, i.e., gray centroid algorithm, Gaussian fitting algorithm, and ellipse fitting algorithm. Then, we proposed a algorithm in which the grayscale and distance thresholds were set by the Gaussian fitting algorithm and Canny edge detection and the spot centroid was extracted by the gray centroid algorithm. The tests of 100 sets of simulated spots reveal that the mean square deviation of the centroid offset extracted by the proposed algorithm is less than 0.05 pixel, and the accuracy is higher than that of the other three algorithms. Furthermore, this algorithm was used to extract the spot centroid of a certain segment of continuous laser data acquired by the Gaofen-7 (GF-7) satellite. The results show that the two laser footprint spots of the GF-7 satellite are within 0.55 cm in the case of ground jitter. Our work can provide a certain reference for the subsequent usage research on laser altimetry data from the GF-7 satellite.

Based on the displacement measurement system of grating interference fringes, this paper proposed a fringe image displacement detection algorithm based on gray interpolation, i.e. grayscale extreme value algorithm. First, the optical displacement measurement system is introduced. Then, the principle of grayscale extreme value algorithm is explained. In order to verify this algorithm, the fringe image of the optical system was simulated and the algorithm simulation was used to locate the image at sub-pixel level. The traditional image correlation method and the optical flow method were compared and analyzed, and the relative error curve was obtained. The proposed algorithm accuracy is 1.797 pixel. Finally, a grating interferometric optical system was built, and the moving stripe data were collected and demodulated using the grayscale extreme value algorithm to realize sub-pixel level displacement demodulation with an experimental accuracy of 1.093 pixel under a large range. The experimental results show that the grayscale extreme value algorithm possesses a high measurement accuracy in a large range and can be used to realize sub-pixel level positioning. Moreover, it possesses the characteristics of short calculation time, little influence of noise, and strong robustness. The proposed method is also suitable for image-based displacement signal demodulation of other systems, such as self-coherent moire fringe system and photoelectric shaft encoder.

With some offshore waters in China as the study area, relying on the bidirectional long short-term memory (Bi-LSTM), we established a model for optimizing and inverting the Sentinel-2A image data on August 20, 2017, and the measured water depth data by a single-beam sounding system. Then, the water depth was compared with that of traditional inversion methods in terms of accuracy. The results show that the model proposed in this paper achieves the optimal evaluation indexes, with root-mean-square error (RMSE) and mean relative error (MRE) of 0.85 m and 18.93%, respectively. The MRE is 14%--29% higher than those of other methods. In particular, the RMSE in the 0--2 m offshore shallow water region displays a sharp rise of 0.68 m. Furthermore, 5-fold cross validation was used to test the network model cyclically, demonstrating the highest and most stable accuracy of Bi-LSTM. This model can invert coastal water depth with high accuracy and stability, and also apply to the depth inversion of regions without training samples, meeting the marine needs such as resource development, engineering construction, and environmental protection.

In order to obtain rail longitudinal direction parameters of a rail profile measurement system, a rail longitudinal calibration method based on the planar target is proposed. The Z-axis of the target coordinate system is always vertical to the rail longitudinal direction when the plane target is close to the rail surface. The plane target is translated or rotated on the rail surface and is closely attached to the rail surface. A group of plane target images with different postures are collected by the camera, and the Z-axis unit vector of the target coordinate system is transformed into the world coordinate system. After the transformation, the coordinates of the unit vector are fitted to the plane, and the longitudinal direction vector of the rail is obtained by calculating the normal of the fitting plane, so as to complete the calibration of the longitudinal parameters of the rail. The effectiveness of the proposed method is verified in the laboratory environment, and the rail longitudinal calibration error is less than 0.1°. When the number of target images is more than 10, the optimal calibration effect can be achieved. This method provides a theoretical basis for the accuracy improvement and reliability evaluation of the rail profile measurement system.

In this paper, we developed a miniaturized adaptive dual-comb system. Two erbium-doped fiber oscillators with repetition frequency of about 77.2 MHz and frequency difference of 180 Hz were used as the seed source of the dual-comb. Their output pulses were injected into a highly nonlinear fiber for spectral broadening after passing through two-stage erbium-doped fiber amplifiers. As a result, the spectral coverage range was extended to 1200--1700 nm. Furthermore, we applied two continuous lasers centered at 1550 nm and 1564 nm and two fiber oscillators to beating and thus obtained four beat signals. This aimed to compensate for the instability of differences in carrier-envelope offset frequency and repetition frequency. Piezoelectric transducer voltage feedback and multi-stage temperature feedback were introduced to the above signals to achieve their stability. As such, the standard deviation of the beat signals was 0.19 MHz. The developed adaptive dual-comb system has a size of 80 cm×67 cm×8 cm and overall weight less than 30 kg, which provides a solution for a wider range of outdoor applications.

A concave quartz window (concave lens) is installed at the front end of a cavity receiver to improve the uniformity of energy flow and gain advantage of the redistribution ability of the functional curved surface of an optical concave lens to the transmission of solar light. A combined ray-tracing method and genetic algorithm optimization design method are used to optimize the geometric size and installation position of the concave lens and improve the uniformity of the energy flow of the dish system. Twelve types of concave lenses are combined with four distinct quadric surfaces and three curved surfaces. The benefits and characteristics of concave lenses in improving the uniformity of energy flow of the dish system are analyzed in detail, and the effectiveness of the proposed method is verified. The results show that adding an optimized concave lens to the dish system can significantly improve the energy flow uniformity of the cavity receiver, reduce the local energy flow peak, and obtain excellent optical and energy efficiency. The optimized double cone lens showes a significant reduction of the non-uniformity factor from 0.59 to 0.11.

In view of the security and practicability requirements of an electronic payment system as well as the problems of complicated steps and trivial signature processes and others in the traditional electronic payment protocol based on controlled teleportation, we propose an electronic payment protocol based on quantum dense coding. The proposed protocol uses quantum operations such as quantum key distribution, single particle measurement, Bell measurement, and unitary operator to perform message blinding, authorization, signature and verification processes in order, and thus completes the whole electronic payment process. In the new protocol, the three-particle entangled state is used as the quantum channel, which can use less resources to complete transactions among different banks. Moreover, by using dense coding instead of controlled teleportation, it is possible to transmit two-bit classical messages with only one qubit. The security analysis shows that the protocol can guarantee the blindness of the purchase information, and satisfy the undeniability, unforgeability and unconditional security.

The current remote sensing images object detection methods based on deep learning are difficult to achieve real-time detection on satellite with limited computing resources due to its complexity and large amount of calculation. To solve this problem, a light-weight and embedded-based method is proposed. Based on YOLOv3-tiny, the network structure is optimized by simplifying the network and improving the multi-scale prediction. Then, the spatial attention module is introduced to enhance the characteristics of remote sensing objects. The experimental results show that under the input size of 608×608, the mean average precision, recall rate, and F1 value of the proposed method are 76.70%, 75%, and 78%, respectively, which are 3.61%, 8%, and 6% higher than that of YOLOv3-tiny. Meanwhile, its computation and model volume are reduced by 39.67% and 71.26%, respectively, compared with YOLOv3-tiny. In addition, the proposed method can achieve a real-time detection speed of 32.5 frame/s on the embedded platform NVIDIA Jetson Xavier NX, which can meet the requirement of real-time detection when run on the embedded platform.

The field widening of a sweep mirror field-widened Fourier transform imaging spectrometer is achieved by adding a sweep mirror before its optical system. However, the addition of the sweep mirror causes the cross distribution of rays from different fields and different channels, which presents a difficulty in the suppression of stray light, and thus the traditional stray light suppression design is no longer valid. To solve this problem, a new design of stray light suppression is proposed. Based on the stray light suppression design, the point source transmittance (PST) stimulation and analysis of the visible and near-infrared optical system and the shortwave infrared optical system is presented for the sweep mirror at different field of view locations of 0°, 15°, and -15°. The analysis results show that as for all sweep mirror locations, the PST along the spatial direction and spectral direction at 0.5° off-field angle can be reduced to 10 -3. Based on the observation mode, the signal source and stray light source of the on-orbit imaging spectrometer are analyzed. In addition, based on the PST stimulation results, the stray light irradiance at the focal plane is studied and the signal-to-stray ratio is calculated. It shows that the signal-to-stray ratio of the visible and near-infrared optical system is 0.1% and that of the shortwave infrared optical system is 0.6%. The results indicate that the proposed stray light suppression method is effective and can meet the requirement of the space-borne imaging spectrometer.

In this paper, we developed a near-infrared laser sensor for the detection of methane (CH4) isotope abundance with tunable laser diode absorption spectroscopy for gas detection and analysis applications in petrochemical and other fields. To improve the stability of isotope abundance detection, we adopted a customized pressure control module with two electronic proportional valves to dynamically control the pressure of the gas cell. With the target pressure being set to 13.332 kPa, the average monitoring pressure in 30 min was 13.326 kPa, and the 1σ standard deviation was 25 Pa. To improve the detection accuracy of isotope abundance, we applied a linear regression algorithm instead of the traditional absorbance maximum ratio method. The detection experiment was carried out with 5×10 -3 (volume fraction) standard CH4 gas, and the results of the two methods were compared. The results demonstrate that the average isotope abundance (δ13CH4) obtained by the absorbance maximum ratio method is 1.633%, and the 1σ standard deviation is 0.962%. The results largely deviate from the theoretical values and have poor stability. The 1σ standard deviation obtained by the linear regression algorithm is 0.367%, and the average value is -4.652%, which is consistent with the isotope abundance results of natural CH4. The stability of the linear regression algorithm detection results is three times that of the maximum ratio method. The pressure control scheme and linear regression algorithm proposed in this article lay a foundation for the development of practical isotope abundance sensors and show potential application prospects in environmental protection, resource exploration, and other fields.

For NOx pollutions in Wuhan during the outbreak of COVID-19, we applied a mobile MAX-DOAS and a portable ultraviolet DOAS to cooperatively measure NOx concentration in the third ring road of Wuhan from Feb. 29 th to Mar. 14 th, 2020. The mobile MAX-DOAS acquired the vertical column concentration (VCD) distribution of NO2 along its course and the portable ultraviolet DOAS measured the NO and NO2 concentrations to calculate [NOx]/[NO2]. Then, the NOx emission flux and its error of the third ring road of Wuhan were calculated in conjunction with the data of the wind field. The results show that the NOx emission flux during the measurements in the third ring of Wuhan ranges from 7.78 mol/s to 15.71 mol/s, about 10.78 mol/s on average. Compared with the average [NOx]/[NO2], the real-time [NOx]/[NO2] along the route of the mobile MAX-DOAS derived from the portable ultraviolet DOAS could effectively reduce the error of NOx emission flux caused by the [NOx]/[NO2] error. However, this method is not recommended in scenarios with substantial near-surface NOx emission sources.

To study the emissivity spectrum characteristics of high-temperature targets in the short-wave infrared band and the correlation law between emissivity and temperature, a high-temperature graphite plates is measured using a spectrometer in a darkroom environment, and its surface is measured using a thermocouple based on the black body radiation law. Spectral emissivity of the high-temperature graphite plate in the short-wave infrared band (1300--2500 nm) at nine different temperatures is obtained. At the same time, the variance analysis method is used to analyze the differences in emissivity at different temperatures. The experimental results show that the shape of the emissivity spectrum curve is the same at different temperatures. The emission valley and emission peak appear at the wavelengths of 2200 and 2380 nm, respectively. As temperature increases, emissivity tends to decrease. In case of a confidence level of 0.05, the emissivity of different temperatures is significant; emissivity changes with temperature. In the temperature range of 559--855 K, the value of the emissivity from 0.77--0.83 to 0.62--0.70. The change law can be fitted with a linear model, and the coefficients determination are all greater than 0.87.

To increase the classification and recognition rate of sex hormones in a complex water environment, we proposed to combine three-dimensional fluorescence spectroscopy with a model of improved chicken swarm optimization based support vector machine (ICSO-SVM). An FS920 fluorescence spectrometer was used to analyzed the fluorescence characteristics of single-component solutions and mixed solutions of three typical sex hormones, i.e., estrone, estradiol, and estriol. On the premise of severe spectral overlap, the ICSO-SVM model was established to classify and identify the three sex hormones. The proposed model has stable training, fast convergence, and 100% sex hormone recognition rate for the test set, and thus it outperforms the PSO-SVM model. In conclusion, three-dimensional fluorescence spectroscopy combined with ICSO-SVM model is effective for sex hormone detection.

As the main technology of defect nondestructive testing, radiographic testing has unique advantages in defect characterization, but it is difficult to measure the size of the defects along the X-ray penetration direction. For this reason, taking steel plate as the research object, based on the radiographic imaging theory, the relationship between the image gray and the penetration process conditions, linear attenuation coefficient, and penetration thickness was studied. Numerical integration and curve fitting methods are used to convert the linear attenuation coefficient into tube voltage and penetration thickness expressions. The final image grayscale-penetration thickness relationship model is obtained. By using the experimental platform to obtain the experimental data, the unknown parameters in the model are obtained by software fitting, and the accuracy of the model is measured by the steel step test block, and the actual thickness and the calculated thickness of the step test block are compared. The results show that the actual thickness and the calculated thickness are close, the error range is 0.18%--1.49%. The experimental results prove the accuracy of the model well, and the thickness of the defect in the steel material can be calculated according to the actual penetration conditions and the image gray level, and the size of the defect along the ray penetration direction can be obtained. Combined with the digital imaging image to measure the size of the defect perpendicular to the ray direction, the defect can be evaluated in an all-round way.