最新刊期
期 7 ， 2021
- 摘要：In the Taiji mission， the tilt-to-length （TTL） coupling noise caused by satellite jitter is the second largest noise source that affects the accuracy of the laser interferometer. In order to ensure measurement accuracy， it is necessary to suppress this noise to ≤ ±25 µm/rad@±300 µrad. Based on Fermat's principle， an imaging system composed of biconvex lenses was designed， and the effect of the imaging system on suppressing TTL coupling noise was simulated and analyzed. A physical experiment platform was built to verify the effectiveness of the imaging system in suppressing TTL coupling noise. The simulation results showed that the TTL coupling noise was ≤ ±10 µm/rad@±300 µrad after the biconvex lens imaging system was added； the experimental results showed that the TTL coupling noise was ≤ ±15 µm/rad@±300 µrad after the biconvex lens imaging system was added. These results satisfied the requirements of the Taiji mission in terms of the level of TTL coupling noise. Moreover， the entire interference optical path was easy to control， and high measurement accuracy was achieved. The suppression effect of TTL coupling noise was verified by buliding a physical experiment platform， which laid the foundation for physics experiments of the Taiji mission laser interferometer.关键词：laser interference measurement;gravitational-wave observation;Taiji mission;laser interferometer in space
- 293
- |
- 164
- |
- 1
发布时间：2021-08-09 - 摘要：To determine the polarization orientation of the sky in cloudy weather， a method of cloud segmentation using an improved U-Net deep neural network combined with a polarization degree threshold was proposed. This method was based on Rayleigh scattering theory and obtains the angle between the direction of the sun and the body axis to determine the polarization orientation. The Respath module containing three residual structures was used to integrate the coding and decoding areas of the U-Net neural network and to add the MultiBlock module for downsampling， so that the features of the input image could be accurately learned. Next， two binary segmentation maps， one predicted by the neural network and another with a polarization degree threshold set to 0.1， were ORed to obtain a binary image that can more effectively separate clear and cloudy skies. Finally， this binary graph was used as a mask template to remove the abnormal pixels in the polarizing azimuth graph， so that the retained pixels conformed to Rayleigh scattering theory. The angle between the direction of the sun and the body axis could then be obtained. Results of experiments in actual cloudy weather indicate that the RMSE error of the solar azimuth in the carrier coordinate system obtained using this method is 0.42°， which can meet real-world navigational requirements.关键词：polarization guidance;deep neural network;polarization degree threshold;Rayleigh scattering theory;cloudy weather
- 238
- |
- 54
- |
- 1
发布时间：2021-08-09 - 摘要：To improve the accuracy of photometric measurements and the attitude and shape inversion of space targets， a photometric measurement method entailing extinction coefficient fitting was proposed based on existing measurement methods. In conventional photometric measurement methods， the gray value corresponding to a “0” magnitude was calculated using the gray values of stars whose magnitude was known； then， the magnitude of targets was calculated using their gray values while considering the effect of atmospheric extinction. In this study， more accurate photometric measurement data was obtained using the extinction-correction fitting method， based on actual measured data. Then， to reduce the impact of anomalous data due to atmospheric instabilities， the method of obtaining the iteration mean was used for calibrating the gray value corresponding to 0 magnitude. Experimental results indicate that the error of the improved photometric measurement method is approximately 0.11 magnitude. Compared with previous photometric measurement methods， this improved photometric measurement method has a higher degree of precision， and the measured results can be better applied to subsequent inversion work.关键词：photometric measurement;space objective;atmospheric extinction;iteration mean method
- 168
- |
- 48
- |
- 1
发布时间：2021-08-09 - 摘要：An integrated F-P interferometer vibration measurement system based on the ultra-small GRIN fiber probe and its performance were studied. First， a high-precision nano-stage was used as the vibrating target to be measured and a corresponding vibration measurement system was built to measure and analyze the small vibrations. This measurement system was then used to perform multiple measurements of the tiny vibrations generated by the vibrating target. The experimental data were then processed to analyze the performance of the F-P interferometer. The results show that under the given experimental conditions， the F-P interferometer exhibits desirable performance in the vibration amplitude range of 200-300 nm； its minimum linearity is 0.42%， corresponding sensitivity is 7.507 V/μm， and maximum repeatability standard deviation is 0.232 V. Further， in the vibration amplitude ranges of 10-100 nm and 350-500 nm， the F-P interferometer has better repeatability and its standard deviation is less than 0.102 V. Therefore， the F-P interference vibration measurement system based on the ultra-small GRIN fiber probe can be used to measure small vibrations， thereby providing an experimental basis for further research on its application in the precise measurement of small vibrations and displacements.关键词：optical fiber sensing;F-P interferometer;ultra-small fiber probe;micro vibration measurement
- 192
- |
- 124
- |
- 2
发布时间：2021-08-09 - 摘要：An interferometric gravimeter monitors the trajectory of a free-falling object using a Michelson laser interferometer. The reference mirror of the interferometer， relative to the center of the earth， must be static. However， its vibration， according to ground vibrations， is inevitable and causes a deviation between the dropper displacement relative to the center of the earth and that monitored via optical interferometry. This directly impacts the measurement accuracy for a single drop and many scattered drops. Therefore， a vibration compensation method， which compensates the dropper displacement measured via optical interferometry according to ground vibrations monitored by a high-precision ultra-low frequency vibration sensor， is proposed herein for a NIM-3A absolute gravimeter. Experimental results indicate that type A uncertainty of NIM-3A was 7.8 times optimized， without system deviations. The proposed method effectively improves the measurement accuracy for a single drop and many scattered drops and type A uncertainty of NIM-3A without system deviations.关键词：interferometric measurement;absolute gravimeter;Vibration compensation method;Type A uncertainty;Dispersion
- 159
- |
- 60
- |
- 0
发布时间：2021-08-09 - 摘要：The inversion accuracy of carbon monoxide （CO） concentration measurement using tunable diode laser absorption spectroscopy （TDLAS） technology is affected by various measurement errors， such as those pertaining to laser intensity， temperature， and pressure. To improve the measurement accuracy of TDLAS and reduce the influence of multiple errors on measurements， an error distribution method for each measurement error was proposed. First， an error model for CO concentration inversion was established. It included common error terms such as the peak value of the second harmonic， laser intensity， driving current， temperature， pressure， and optical path， based on the second harmonic measurement principle. Second， the influence of various error sources on the measured concentrations was studied separately using a randomized numerical fitting method， enabling the relationship between the concentration error and each error to be derived and the sensitivity of the concentration error to each error to be compared. Lastly， the various errors were effectively distributed by considering the sensitivity relationships， concentration measurement requirements， available devices， and technical complexity. Simulation of the error distribution scheme was performed using the Monte Carlo method. The simulation results show that when the gas volume fraction is 2.5%， the absolute error of concentration measurement is less than 0.025% under the error distribution scheme. Experiments to measure CO with a volume fraction of 2.5% show that the absolute error is less than 0.01%， demonstrating the high inversion accuracy inversion of CO concentration measurement. The results of this research show that the proposed error distribution method ensures high measurement accuracy for a TDLAS concentration detection system under the influence of multiple errors. In addition， the method can effectively decrease the design difficulty and cost of the systems， which is significant for the development and application of such instruments.关键词：tunable diode laser absorption spectroscopy （TDLAS）;carbon monoxide （CO）;concentration inversion;error distribution;Monte Carlo method
- 193
- |
- 78
- |
- 1
发布时间：2021-08-09 - 摘要：Passive ranging of oxygen absorption under weather conditions such as rain， snow， and haze is severely affected by complex background spectra such as those of the atmosphere and aerosols. Mixed pixel decomposition technique was used to improve the range accuracy under such complex backgrounds. Firstly， the pixel mixing mechanism of hyperspectral image was analyzed， and the mixed pixel of the target and background under complex weather conditions was established by taking the endmember of a rain drop as an example. Then， the method of extracting the target spectrum under complex weather conditions was discussed， and the basic flow of passive ranging was proposed. Finally， the passive ranging experiments were carried out for halogen tungsten lamp targets under the conditions of light rain， heavy haze， and moderate snow at different distances. Experimental results show that the mixed pixel decomposition method can quickly extract the target spectrum and improve the ranging accuracy compared to the multi-cycle acquisition average method and background elimination method. Ranging accuracies can be improved to 3.39%， 5.81%， and 4.36% under the three different weather conditions， respectively. This method can meet the requirements of fast acquisition， quick processing， and high ranging accuracy of the radiation spectrum for actual flight targets.关键词：passive ranging;oxygen absorption;mixed pixel decomposition;target extraction;extreme weather
- 154
- |
- 44
- |
- 0
发布时间：2021-08-09 - 摘要：To meet the strict requirements of stability and consistency of the optical axis for the current three-field TV detection system with a photoelectric search and tracking device， a three-field TV detection system with a rotary switching mechanism was studied. Through the analysis of the system indicators， an optical system for cut-in and cut-out field of view switching was selected， and the structure of the rotary switching mechanism was presented. Based on this， a precise rotating shafting， position-limit mechanism with a powerful neodymium magnet and a high elastic plunger， and worm drive mechanism with self-locking characteristics were designed. The selection process and results of the limit mechanism and the drive mechanism were given. A method of optical axis consistency adjustment with theodolite and cross-reticle was proposed. With this design and adjustment process， a high-stability three-field TV detection system was realized. The test results show that the stability of the optical axis of the M-FOV is less than 8"， and the consistency of the optical axis of the three fields of view is within 1 pixel size， which meets the indicator requirements. The results of laboratory and outdoor experiments show that the imaging quality of each field of view of the system is excellent， and the detection range is long， which can meet the requirements of use.关键词：detection system;three fields of view;optical axis stability;rotary switch;imaging
- 137
- |
- 53
- |
- 0
发布时间：2021-08-09
Modern Applied Optics
- 摘要：The machining precision of a cylindrical microlens array is usually very high， whereas the processing efficiency is very low. However， the processing efficiency can be considerably improved using a profile grinding wheel with a fine profile structure. To predict the surface shape error and surface roughness of a workpiece machined using a profile grinding wheel， a grinding simulation model was built for the profile grinding wheel. Then， the skewness distribution characteristics of the protrusion height of grains in the profile grinding wheel surface with a fine structure were analyzed and simulated using the filtering method. Subsequently， combining the outline of the grinding wheel topography and the run-out error of the grinding wheel， the entire grinding wheel in the space was reconstructed. Thereafter， a kinematics model of abrasive grains in the grinding wheel surface was created to simulate the surface topography of the grinding workpiece. Finally， a grinding experiment was conducted to verify the effectiveness of the simulation model. Relative to the experimental results， the errors in the PV， Ra， and Rz values of the simulated surface were 5.78%， 17.3%， and 12.9%， respectively. The proposed grinding simulation model can effectively predict surface shape errors and the roughness of grinding surfaces.关键词：precision grinding;cylindrical microlens array;grinding surface simulation;grinding wheel modeling;profile grinding wheel
- 454
- |
- 106
- |
- 0
发布时间：2021-08-09 - 摘要：Robot calibration is an important process used for enhancing the positioning accuracy of robots. In this paper， a modified differential evolution algorithm was proposed for the identification of robot kinematic parameters. The generalized Metropolis acceptance criterion was used in the selection operation of the proposed algorithm to explore more regions of the search space for a better convergence. In addition， the population evaluation function established in this study could greatly enhance the global-optimization ability of the modified differential evolution algorithm. Herein， the kinematic model was developed based on the product of exponentials （POE） formula to describe the relationship between the kinematic parameter errors and the end-effector positioning errors. To verify the efficiency of the proposed algorithm， simulations and experiments were performed using a six-degree-of-freedom robot and laser tracker. Through the kinematics parameter calibration, the average distance precision of robot is reduced from 2.906 0 mm (calibration) to 0.095 2 mm. The experimental results proves the effectiveness of the modified differential evolution algorithm for calibration of robot kinematics parameters.关键词：Robot calibration;Modified differential evolution algorithm;kinematic model;positioning accuracy
- 255
- |
- 61
- |
- 3
发布时间：2021-08-09 - 摘要：This paper proposes a variable-bandwidth linear active disturbance rejection control algorithm to realize the precise low-speed control of a high-precision air flotation turntable and improve its positioning accuracy and dynamic response performance. First， the control system model is established according to the mathematical model of a turntable and driving motor by considering disturbances. Second， a linear extended state observer is constructed to analyze the effect of an initial error on observation accuracy， and an observer bandwidth adjustment method is proposed to enhance the observation accuracy. Subsequently， according to the design concept of controller bandwidth， a control law with disturbance compensation is specified. Finally， the performance of variable-bandwidth linear active disturbance rejection control is verified using the experimental platform of a high-precision air flotation turntable. The experimental results show that the tracking error is less than 0.0008° under the low-speed condition. The angle positioning error is less than 0.0007°， and the overshoot is reduced from 0.0017° to 0.0014° compared with that of traditional linear active disturbance rejection control. Notably， the turntable can maintain a high positioning accuracy under a single mismatch of the moment of inertia condition and motor inductance parameters. Further， it meets the requirements for a high-precision air flotation turntable in terms of positioning accuracy， speed stability， and disturbance immunity.关键词：active disturbance rejection control;air flotation turntable;variable bandwidth;bandwidth design;low-speed control
- 223
- |
- 56
- |
- 0
发布时间：2021-08-09 - 摘要：The performance of the wireless energy transmission （WPT） system of gastrointestinal robots is easily affected by many factors such as size and safety. In order to make gastrointestinal robots able to receive energy safely and stably under these restrictions， this paper has established a mathematical model， and use intelligent optimization algorithm to optimize it. First， a flat spiral coil is used as the transmitting coil of the system， and a mathematical model for the optimization of the energy transmission system is established. Then， analyze the constraints of the system according to the safety， power requirements， etc.， and use the efficiency product as the optimization function. Then， improve the particle swarm optimization algorithm and introduce the cuckoo search strategy. Finally， the improved particle swarm algorithm is used to optimize the system， and the optimized parameters are verified by experiments. The experimental results show that the transmission efficiency of the system is 10.2%， and the received power is 637mW， reaching the expected results.The optimized WPT system basically meets the energy transmission requirements of gastrointestinal robots.关键词：Gastrointestinal robot;wireless power transmission;transmission efficiency;particle swarm optimization algorithm
- 195
- |
- 91
- |
- 0
发布时间：2021-08-09 - 摘要：Pendulum components are the key parts of a miniature wire-suspended pendulum accelerometer. The accuracy of the assembly directly affects the measurement accuracy of the accelerometer. An assembly equipment for pendulum components is developed in this study. In this equipment， the upper and lower constant fixtures and removable fixtures are designed. The pendulum components are fixed on corresponding constant fixtures through the respective removable fixtures. A linear stepper motor is used to drive a clamp to thread the suspension wire through glass tubes， where the tension of the suspension wire is controlled by a micro-force sensor mounted on the clamp. A vacuum adsorption mechanism is integrated into a 3-D precision motion platform to form the upper operation module. It realizes the picking up， movement， and positioning of the upper fixture. A locking device is integrated into a rotating platform to form the lower operation module. It realizes the fixing and posture adjustment of the lower fixture. After suspension wire threading， the final assembly is achieved through the alignment and combination of the two constant fixtures. During the assembly process， a single camera is set as a stereo vision system via a reflecting prism. Experimental results show that the positioning and parallelism errors of assembly parts are all less than 15 µm. It can meet the requirements of stability， reliability， and high precision of automatic assembly equipment.关键词：Pendulum components;miniature wire suspended pendulum accelerometer;assembly equipment;lass tubes
- 168
- |
- 273
- |
- 1
发布时间：2021-08-09 - 摘要：Planetary grinding technology can effectively improve material removal efficiency by improving the relative velocity of grinding contact points. However， the uneven wear of the traditional grinding disk （or grinding plate） results in the continuous change of the shape of the grinding disk， which affects the stability and accuracy of the removal function in the grinding process， and also limits the application of this technology. In this study， based on the planetary motion mode of a small grinding head， the wear function was established and the curvature radius of the grinding disk was preset to ensure that the removal amount at each point of the grinding disk was equal after a single processing cycle to improve the removal function stability. Experimental verification was also carried out， and the grinding removal function was found to be consistent with the simulation calculation results obtained using the model， which verified the accuracy of the model. Finally， an efficient and stable removal function was obtained by using the optimized grinding disk. A SiC grinding disk with a diameter of 40 mm was used to grind the SiC workpiece. The experimental results showed that the surface shape changed by less than 1% after comparing the wear of the disk before and after machining， and the volume removal error was less than 2.3%， which meets the stability requirements of the optical grinding removal function. Under the conditions of revolution at 100 r·min-1 and rotation at -100 r·min-1， the volume removal rate reached 6.879 mm3·min-1. Compared with the single rotation grinding with the same parameters， the removal amount of 40.9% was increased， which proved that the planetary grinding technology can obtain a highly stable and efficient removal function based on parameter design， and provided reliable theoretical guidance for the application of planetary motion grinding technology in the efficient processing of SiC mirror.关键词：Optical processing;Planetary motion grinding;Wear model of grinding plate;Stability of removal function
- 177
- |
- 79
- |
- 2
发布时间：2021-08-09 - 摘要：Existing structural safety monitoring methods cannot fully meet the requirements of bollard structure safety monitoring under complex mooring conditions. Herein， the concept and index of the standard state mooring force are proposed according to the structure and mooring characteristics of bollards. The standard state mooring force inversion model is derived according to the elastic principle. According to the range of the mooring force direction and requirements of the strain monitoring points of the inversion model， the multi-point strain arrangement scheme on the surface of the bollard is studied， and a structural safety monitoring method for bollards is proposed based on multi-point strain fusion. Additionally， a strain-equalized fiber Bragg grating sensor-based surface welding sensing solution is proposed based on the strain distribution characteristics and long-term monitoring requirements of the bollards. Furthermore， five strain sensors and one temperature sensor were successfully deployed on the 1500 kN bollard at Ningbo Daxie International Merchants Container Terminal， and long-term continuous monitoring was carried out. By analyzing the inversed cable force under the standard conditions and actual values， the maximum error of the inversion of the standard conditions was found to be 8.1%. It is proved that this method has high feasibility and operability and can effectively perform monitoring and provide early warnings regarding the safety of large mooring column structures in harbors.关键词：bollard;standard condition mooring force;the inversion model
- 171
- |
- 34
- |
- 0
发布时间：2021-08-09 - 摘要：A numerical method is established to analyze the influence of the pendulum angle and rotation speed on the static characteristics of the air-floating turntable of an ultraprecision full-aperture polisher. Substituting the air-film-thickness equation， which contains the pendulum angle， into the Reynolds equation results in an equation that includes the pendulum angle and speed terms. This equation is then solved by numerically combining the finite difference method with the flow balance equation. The calculation results show that in the case of a single pendulum， the maximum limit of the pendulum angle of the air-floating turntable is 5×10-5 rad， and with an increase in the pendulum angle， the maximum value of the air film thickness gradually increases and its minimum value gradually decreases. Furthermore， the pressure distribution changes and the load capacity gradually decreases. In the case of combined pendulums， the maximum limit of the pendulum angle of the air-floating turntable is 4.9×10-5 rad. The effect of the combined pendulums on the static characteristics of the air-floating turntable is similar to that in the single-pendulum condition， but more significant. The failure speed of the air-floating turntable is 2 300 r/min. As the speed increases， the thickness of the air film does not change directly. However， the thickness of the air film changes when the load capacity decreases owing to pressure distribution. It can be concluded that a change in the pendulum angle and rotation speed changes the load capacity of the air-floating turntable. Therefore， the position of the turntable will change accordingly to maintain the balance of force. This research provides a theoretical basis for the subsequent study of dynamic rotation errors.关键词：ultra-precision air-floating turntable;angle pendulum;rotation speed;finite different method;static characteristics
- 334
- |
- 89
- |
- 2
发布时间：2021-08-09 - 摘要：To realize full-field measurements of the deformation of slender and large specimens， a novel stereo measurement method based on digital image correlation is proposed. By establishing a joint constraint relationship for the multi-camera network， any two CCD cameras can be bound as a whole to complete partial measurements of the specimen for subsequent mapping onto a unified reference coordinate system. Consequently， the influence of large nonlinear imaging distortions caused by rough full-field measurements is diminished， while the effective spatial resolution of the acquired images is maintained. In addition， an optimized joint constraint relationship accurately restricts the search for stereo matching points between image pairs to the epipolar line instead of the entire image， thereby reducing the search space and improving the search speed. Moreover， by optimizing the position and altitude parameters between the specimen and the multi-camera network， the accuracy of the reconstructed three-dimensional coordinates of the speckle space is further improved. Finally， the displacement and deformation measurement experiments illustrate the effectiveness of the proposed strategy. Within the spatial range of ［0.8 m， 0.5 m， 0.6 m］， the translation measurement error of the target is less than 0.044 mm and the out-of-plane displacement measurement accuracy is high （±0.06 mm）， which meets the application requirements of aircraft ground physical simulation tests.关键词：machine vision;multi-camera network;joint constraint optimization;digital image correlation;deformation measurement
- 174
- |
- 53
- |
- 1
发布时间：2021-08-09
Micro/Nano Technology and Fine Mechanics
- 摘要：The traditional multi-projection color correction and brightness fusion are performed in the RGB color space； the brightness and chroma interfere with each other in the correction process， which is a major concern. To solve this problem， a color correction and brightness fusion method based on the Lab color space is proposed. The brightness and chroma are extracted separately， and the brightness and chroma conversion relationship between the original images of each projector and the projected display screen is established using the B-spline curve model to eliminate the color difference between the display screen of each projector， and the brightness of the overlapping area of the screen is adjusted using the gamma correction fused with the B-spline curve. The experimental results show that， compared with the RGB color space， color correction in the Lab color space reduces the difference in color intensity of the two projection display images by 4.14， 2.41， and 2.44 in Lab three channels. Compared with gamma correction in the RGB color space， gamma correction and B-spline curve are fused to adjust brightness in Lab space， which reduces the difference in color intensity between the overlapping and non-overlapping areas by 6.12， 4.64， and 3.47 in Lab three channels. The multi-projection color correction and brightness fusion technology based on Lab can correct the brightness and chroma of the projected display image more accurately， and obtain the gamma coefficient corresponding to different color intensities.关键词：Lab color space;color correction;brightness fusion;Gamma correction;B spline curve
- 148
- |
- 53
- |
- 1
发布时间：2021-08-09 - 摘要：From stellar background images obtained by using the Stellarium astronomy software， small targets with different motion directions are selected randomly. The small targets are randomly assigned， shifted in random directions， and modified in size within the different frame images to simulate the morphological and motion characteristics of the space targets in the stellar background， and the sequence frame image is pre-processed to remove noise. After a series of operations， viz. image segmentation， mass determination， image registration， and maximum projection， an image containing the motion target trajectory and the space target is generated. Subsequently， the motion direction of the space target is detected using the Radon transformation. The experimental results show that the absolute error of both the calculated direction and the actual direction of motion is 0.0222°. Therefore， the process can be used to effectively detect the motion direction of the space target.关键词：Radon transformation;starry background;space target;movement direction detection
- 213
- |
- 136
- |
- 1
发布时间：2021-08-09 - 摘要：To solve the problem of image blur caused by camera jitter， the relative motion between objects， and other factors， a multi-scale alternating-connection residual network is designed in this study for image deblurring， and the “coarse to fine” multi-scale method is used to gradually restore the clear image. First， a multi-scale residual module is proposed to expand the network width， and to extract and fuse the feature information between different scales. Second， an alternating-connection residual module based on dilated convolution is proposed to gradually recover the high-frequency information of the fuzzy image. Finally， a convolution layer is used to reconstruct the feature map. The experimental results show that the peak signal-to-noise ratio （PSNR） and structural similarity （SSIM） of the proposed method are 32.3136 dB and 0.9425， respectively， better than those obtained by the current image deblurring techniques. The evaluation index and subjective effect suggest that the proposed deblurring method has stronger image restoration ability， richer texture details， can effectively improve the image deblurring effect， and has higher practical value.关键词：image deblurring;multi-scale residual;dilated convolution;alternate connection residual;receptive field
- 173
- |
- 100
- |
- 0
发布时间：2021-08-09 - 摘要：To address the problem of inadequate feature extraction and high feature dimension when CT modal medical images are trained with convolutional neural networks， this paper proposes a method for lung tumor identification using Multi-scale DenseNet-NSCR based on non-negative sparse collaborative representation classification by fusing multi-scale images. First， the parameters of the pre-trained dense neural network model are initialized using migration learning； the lung images are then pre-processed to extract multi-scale lesion ROI. Subsequently， the DenseNet is trained using a multi-scale CT dataset to extract feature vectors at the full connection layer. To address the problem of the high dimensionality of the fused features， a non-negative， sparse， and collaborative representation （NSCR） classifier is used to represent the feature vector and solve the coefficient matrix； the residual similarity is then used for classification. Finally， a comparison test is conducted with the AlexNet， DenseNetNetNet-201 model， and a combination model of three classification algorithms （SVM， SRC， NSCR）. The experimental results show that Multiscale-DenseNet-NSCR classification is better than other models； all evaluation indexes such as specificity and sensitivity are higher， and the method has better robustness and generalization ability.关键词：DenseNet;multi-scale medical image;transfer learning;non-negative， sparse， collaborative representation classifier
- 193
- |
- 91
- |
- 4
发布时间：2021-08-09 - 摘要：To reduce the training calculation time and optimal training times of a neural network model and ensure high classification accuracy of handwritten digital images， compressive sensing technology was introduced， and a fast classification algorithm of handwritten digital images based on compressive sensing and a single-hidden layer feedforward network （Compressive sensing and single-hidden layer feedforward network， CS-SHLNet） was proposed. First， a Gaussian random matrix is used to obtain a linear measurement of the handwritten digital image with sparseness， and the high-dimensional image signal is projected to the low-dimensional space to obtain the measurement value. Second， using the error backpropagation （BP） algorithm， the weights of the neural network are continuously adjusted to establish a single-hidden layer feedforward network model suitable for the measurement values， which are embedded into the neural network for image feature extraction. Finally， a single-hidden layer feedforward network model is used to classify handwritten digits， and the model is quantitatively evaluated by the time-consuming training calculations， the optimal training times， and the classification accuracy. Experimental results show that—in contrast to using a single-hidden layer neural network and deep learning for high-dimensional image signal classification of MNIST handwritten numeral datasets—through the CS technology， the Gaussian random matrix linear measurement number， i.e.， M=235， is first used to obtain the image measurement value； then， the single-hidden layer feedforward network is used for image classification. The training calculation time of the network model is reduced to 13.05 s， the best training times are reduced by a factor of three， and the classification accuracy is 97.5%. The compressive sensing linear measurement in the algorithm can effectively reduce the computation time of the training and the optimal training times of the neural network model for handwritten digital datasets and the classification accuracy can be ensured.关键词：image classification;handwritten digits;compressive sensing;neural network
- 190
- |
- 134
- |
- 1
发布时间：2021-08-09 - 摘要：In order to achieve accurate foreground motion estimation， a coded marker-based high-accuracy motion estimation method is proposed in this paper. First， several circular coded markers are pasted on the foreground and background of the measurement environment. Then， several images are captured after each foreground motion. Finally， based on the captured images， all the camera poses and the spatial coordinates of the fixed markers on the background and the moving markers on the foreground are estimated simultaneously to obtain the 6 Degrees of Freedom （DoF） parameters （rotation and translation） of foreground motion. To this end， a high-accuracy motion estimation pipeline is proposed in this paper， which includes RANdom SAmple Consensus （RANSAC）-based coded marker detection and recognition， graph clustering-based coded marker and camera partitioning， incremental Structure from Motion （SfM）-based background marker and camera initialization， graph optimization-based foreground marker and foreground motion initialization， and generalized Bundle Adjustment （BA）-based global optimization. The experimental results show that the accuracy of the proposed foreground motion estimation method is approximately 0.3 mm， which is satisfactory for high-accuracy foreground motion estimation.关键词：High-accuracy motion estimation;coded marker;graph clustering-based partitioning;generalized BA-based global optimization
- 174
- |
- 37
- |
- 2
发布时间：2021-08-09 - 摘要：Focus measure is easy to be interfered by the noise in the autofocusing process for a camera and electronic microscope. In order to solve this problem， we combine the slant transform in image processing with the variance in statistics to propose a novel focusing measure with noise robustness. First， the image is divided into small blocks to facilitate the subsequent acquisition of the sharpness index of each local sub image. Then， we perform the slant transform on each sub image to obtain the frequency coefficients. In the transform domain， we calculate the absolute value of the mid-frequency coefficient， and perform the summation operation to obtain the sharpness index of each sub image. Finally， the variance of each sub image sharpness index is calculated， and the result is taken as the final focus measure of the whole image. By extracting the mid-frequency information of local sub images and solving the global variance， the proposed focus measure has strong noise robustness. The experiments on LIVE image database indicate that compared with the typical focus measure， the noise robustness of the proposed method is better than the existing classical algorithms， where the sharpness detection ability （SDA） and the discreteness evaluation indexes are improved by 20.27% and 125.61% averagely.关键词：image processing;focus measure;slant transform;variance
- 163
- |
- 62
- |
- 0
发布时间：2021-08-09 - 摘要：Video-based non-contact photoplethysmography （PPG） can achieve non-contact heart rate monitoring. To improve the quality of non-contact PPG signal and the accuracy of non-contact heart rate， a novel method based on adaptive region of interest （ROI） was studied. First， independent vector analysis （IVA） was used to process the face by blocks. Then normalized cuts （Ncuts） was used to select blocks with the highest signal-to-noise ratio and correlation as adaptive ROI to obtain the heart rate. Finally， the non-contact PPG signal was obtained by weighted average of the adaptive ROI and frequency domain processing. Compared with pre-selecting ROI method， the mean and standard deviation of the heart rate error was reduced from （4.72±6.46） beats per minute （bpm） to （0.52±1.49） bpm， the root mean square error （RMSE） was reduced from 7.96bpm to 1.50bpm， and the average error rate was reduced from 9.45% to 1.73% in static situation. The error of this method under head movement was （1.02±2.91） bpm， the RMSE was 2.11 bpm， which was reduced by above 50%. Bland-Altman and correlation analysis was used to compare the heart rate obtained by this method and using a contact PPG instrument， the 95% confidence interval was -2.44~3.48 bpm at static and -2.76~4.79 bpm at move situation. Compared with the waveform of the contact PPG signal， it was showed the method obtains the PPG signal with complete details. Experimental results showed this method has significant advantages in the quality of PPG signal and the accuracy of heart rate.关键词：photoplethysmography;non-contact heart rate;adaptive region of interest;independent vector analysis;normalized cuts;signal to noise ratio
- 170
- |
- 53
- |
- 0
发布时间：2021-08-09
Information Sciences
- Address：No.3888 Dong Nanhu Road, Changchun, Jilin, China Postal code：130033
- Tel：0431-86176855 Email：gxjmgc@ciomp.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 吉ICP备11002662号-17
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0