In order to solve the high-precision interactive control problem of laser Raman spectrometer,this paper proposes a reliable and stable multi-channel communication interactive system.The system integrates the required functions into the visual interface through human-computer interaction technology,so as to enhance the convenience of Raman spectrometer operation.In the communication technology of the multi-channel Raman spectrometer,stable and reliable communication technology and phase-locked loop technology are adopted.Aiming at the clock drift problem caused by the aging of the crystal oscillator and electromagnetic interference in the actual communication process,this paper uses the clock Kalman filter algorithm to estimate the error,and finally performs error compensation in the program design,which greatly improves the accuracy of information data transmission.Experiments show that this communication interactive system has stable and reliable data transmission and convenient control functions.

In recent years,the rodent damage has been increasing on our country′s grassland year by year.Grassland rodent damage not only aggravates the process of soil erosion and desertification,but also causes plague.The number of rat holes is an important index for rodent damage monitoring and grade evaluation in our country.Now the manual survey method has many problems,such as low precision,time-consuming and labor-consuming,high investigation cost,only suitable for small area investigation,and so on.It is difficult to meet the requirements of large area real-time monitoring and research.Dynamic and real-time monitoring of the number of rat holes is an important means to effectively formulate anti-rodent measures and prevent the occurrence of plague.In this study,the data of desertified grassland was collected by high spectrometer carried by unmanned aerial vehicle (UAV),and the rat hole index (RHI) was proposed to identify the rat hole in the grassland.The results show that the overall accuracy of identifying prairie rat holes by RHI index can reach 97% and the Kappa coefficient can reach 0.93.Compared with normalized difference vegetation index (NDVI),soil-adjusted vegetation index (SAVI) and ratio vegetation index (RVI) vegetation index models,this model has higher recognition accuracy.The proposal of RHI can effectively improve the accuracy and efficiency of rat hole identification in grassland,and provide an effective method for rodent control and grassland degradation monitoring and research.

Hollow-core photonic Bragg fibers have a wide range of sensing applications.In this paper,a hollow-core photonic Bragg fiber with defect layers is proposed and analyzed for bacterium detection in the terahertz (THz) frequency range.The hollow core of the fiber is surrounded by a periodic sequence of high/low refractive index (RI) multilayers including the photosensitive resin and air.The analytes are deposited on the inner surface of the THz hollow-core Bragg fiber to induce the defect layer.The performance of the proposed surface bio-sensor is numerically investigated.The result shows that the proposed surface bio-sensor provides a high core power fraction.The confinement loss of the proposed surface bio-sensor takes on two peaks in the frequency range of 0.3—0.4 THz and 0.45—0.55 THz,respectively.The distinguishable characteristic frequencies of the proposed surface bio-sensor are obtained with different bacterium deposited on the inner surface of the THz Bragg fiber which can be used to recognize the kind of the bacterium.Furthermore,the thickness-independent surface sensing can be realized in the high frequency range.The proposed surface bio-sensor based on THz hollow-core Bragg fiber can be used to recognize the kind of the bacterium and has a great potential in microorganism detections.

From March to November 2019,the largest on-station comparison observation calibration (OCOC) in China was carried out so far,based on the mobile REAL standard aerosol lidar certified by the European lidar calibration center.In this paper,a method for OCOC is firstly introduced.Then,the first OCOC between REAL lidar and 12 aerosol lidars distributed in Beijing,Shanghai and Guangzhou were carried out separately at 532 nm wavelength.The 11 radars which were calibrated in advance according to method of hardware and software calibration established by the meteorological observation center (MOC) were quantitatively evaluated according to the preset quality criteria,and the results show that the 11 radars basically meet the requirements.The system deviation (SD) of 9 radars P/S channels within a range of 2—5 km was all in ±20%,and the SD of more than 50% of the calibrated radars P channels within a range of 0.5—2 km is basically in ±10%.The main problems are P/S channel crosstalk and detector saturation,etc.This work determines the reliable range of data of calibrated radars and provides support for operational quantitative application of aerosol lidar.

Aiming at the problem that the target resolution in unmanned aerial vehicle (UAV) tracking scene is low and it is difficult to track the target robustly due to the influence of UAV flight attitude,speed change and other factors,an adaptive spatial-temporal regularization algorithm for UAV target tracking is proposed.Based on the spatial temporal regularized correlation filter (STRCF) algorithm,the cost of spatial regularity in AutoTrack is introduced,and the spatial-temporal regularization parameters are dynamically updated online by using the peak side lobe ratio and local response variation vector to improve the accuracy of the tracker.Occlusion processing module is embedded in the tracker to solve the tracking drift problem after the target is occluded.Experimental results on several UAV benchmark datasets show that the proposed algorithm has higher accuracy and faster processing speed than the benchmark algorithm Autotrack.The tracking accuracy and speed of DTB70 data set are improved by 1.5% and 74.4% respectively.In the classification comparison of nine attributes on UAVDT data set,this algorithm achieves high performance on seven attributes,such as scale variation (SV),object blur (OB),etc.It can be seen that this algorithm can better meet the requirements of UAV applications.

In order to effectively identify students′ online classroom action,a human skeleton action recognition model integrating global attention mechanism and spatiotemporal convolution network is proposed.Firstly,a global attention module is added between the spatial graph convolutional network and the temporal convolutional network of the Spatiotemporal graph convolutional neural network,and the spatial feature map output by the spatial graph convolutional network is used as the input of the attention module;Secondly,average pooling and maximum pooling operations according to the time dimension are introduced to increase the ability of the model to learn global feature information.Finally,three spatiotemporal graph convolutional neural networks and class activation map (CAM) added to the attention mechanism are used to construct a rich activation map convolutional network with stronger ability to recognize occlusion data (RA-GCNv2-A) model,and realize student online classroom action recognition function through transfer learning.Experimental verification is performed on the NTU-RGB+D and NTU-RGB+D120 two datasets.Compared with the RA-GCNv2 model,the recognition accuracy on the NTU-RGB+D dataset is increased by 1.3% (cross-subject,CS),1.2% (cross-view,CV),the recognition accuracy on the NTU-RGB+D120 dataset is increased by 1.6% (cross-subject,CSub),1.4% (cross-setup,CSet) respectively.The experimental results show that the proposed method is an effective way to recognize students′ online classroom action.

Deformation image registration (DIR) is a key process of contour derivation and dose accumulation in adaptive radiotherapy (ART).This study compares the changes of contour derivation and dose accumulation in the process of adaptive radiotherapy with three different deformation registration algorithms.The radiotherapy imaging data images of patients with head and neck cancer were selected for this study.After rigid registration,B-spline elastic registration and LCC-Demons registration algorithms were used.The B-spline elastic algorithm studied two optimization indexes:normalization Cross-correlation,normalized cross-correlation combined with bending energy penalty (BEP) regularization.Dose evaluation uses dose parameters D95,Dmin,Dmean,Dmax and uniformity index (HI) to evaluate the difference between cumulative dose and initial dose.Geometric evaluation uses Dice correlation coefficient (DCS) and Hausdorff distance (HD) to evaluate the degree of overlap between volumes.After experiments,the Dice coefficients of the planning target volume (PTV) target area,left parotid gland and right parotid gland obtained by geometric evaluation are respectively 0.82±0.07,0.83±0.07,0.86±0.08 in the two optimization algorithms of B-spline elastic registration and the LCC-Demons algorithm.Among them,the Dice coefficient obtained under the LCC-Demons algorithm reached the maximum value of 0.91, and the obtained Hausdorff distances were all between 1 and 3 mm.The DSC overlap result and the indicated deviation distance were both within the clinically acceptable range.For dose evaluation,the corresponding HI values are:1.089,1.082,1.081,LCC-Demons also reached the minimum value of 1.081.In this experiment,LCC-demons achieved good results in contour derivation and dose mapping,and it is also clinical it provides a certain reference for the evaluation of registration quality in the environment.

This paper proposes a surface defect detection model on a copper plate and strip based on YOLOv4.Aiming at the problem of surface defects in the production process of copper metal plate and strip that are difficult to locate and identify due to their various forms and random positions, a big data-driven deep learning strategy is adopted.Using the copper strip surface defect image as the training sample,the YOLOv4 target detection model is trained.The experimental results show that the improved model recognizes the copper strip surface defect with a full-category mean average precision (mAP) of 93.37%,which is higher than the original YOLOv4.The model has an average accuracy of 91.46% for all categories and a detection speed of 49 frames per second.Compared with the two-stage detection model faster region-based convolutional neural network (Faster R-CNN),it can improve the detection speed while ensuring the detection accuracy,which can meet the needs of online detection.Defect detection task in industrial production process is suitable for completing copper strips.

In order to solve the problem that low-dose computerized tomography (CT) introduces a lot of noise in the proqcess of image acquisition,which leads to the serious degradation of image quality,a low-dose CT denoising algorithm based on residual attention mechanism and composite perceptual loss is proposed in this paper.In this algorithm,the Generative Adversarial Networks is used to complete the denoising of low-dose CT images.The multi-scale feature extraction and residual attention module are introduced into the network framework to fuse the information of different scales in the image,improve the ability of the network to distinguish noise features,and avoid the loss of image details in the process of denoising.At the same time,the composite perceptual loss function is used to accelerate the convergence speed of the network and promote the denoising image to approach the original image perceptually.Experimental results show that the proposed algorithm can effectively suppress noise and recover more texture details in low-dose CT images compared with existing algorithms.Compared with the low-dose CT images,the peak signal-to-noise ratio (PSNR) value and structural similarity (SSIM)value of the CT images processed by the proposed algorithm are increased by 31.72% and 13.15%,which can meet the higher requirements of medical imaging diagnosis.

Aiming at the low precision of the problem of high-speed photography technology at a longdistance measurement of the vibration response in the test,a method combining bicubic interpolation with Sobel operator edge detection image technology is proposed to improve the measurement precision.At first,the shaking table test of pipe system with two measuring systems,high-speed photography and accelerometer,is designed.Secondly,the image measured by the high-speed camera system and the acceleration vibration response measured by the accelerometer are processed,and the displacement vibration response data is extracted.Finally,the displacement vibration response data measured by high-speed photography and accelerometer are compared in time domain and frequency domain,and analyzing the accuracy of data measured by high-speed photography.The results show that the image measured by optimizing high-speed photography technology combined with bicubic interpolation and Sobel operator edge detection image processing can obtain high-precision displacement vibration response data to meet the measurement and use requirements.

Aiming at the problems of prior knowledge and slow algorithm convergence for feature selection in the quantitative analysis of laser-induced breakdown spectroscopy (LIBS) technology,this paper proposes a quantitative analysis method for automatically selecting features with a combination of Pearson correlation coefficient-based ranking,principal component analysis and L1 regular term.This method first selects the feature that has the greatest correlation with the target element,then compresses the feature dimension to within the number of samples,and finally sparses the feature weight coefficient and establishes a quantitative analysis model.This method is used to screen the characteristic subsets of Co elements in the soil and establish a quantitative analysis model.The R2 (coefficient of determination) of the training set and test set of the model reached 0.995 and 0.991,root mean square error (RMSE) were 4.634 mg/kg and 6.078 mg/kg,mean absolute error (MAE) were 6.100% and 6.441%.The number of features is reduced from 42 870 of the original spectral data to 5,which takes only 0.97 s.The results show that the method proposed in this paper can reduce the dimension of feature subsets and improve the generalization and accuracy of quantitative analysis models,providing an efficient method for feature selection in quantitative analysis of LIBS technology.

Signal filtering,spectrum analysis and adaptive signal processing have been widely used in all-phase signal processing.In this paper,based on one-dimensional all-phase signal processing method,the mathematical formula of two-dimensional all-phase transformation is given,and the corresponding signal processing system diagram is constructed.Secondly,the sufficient and necessary condition for linear phase of all-phased transform is analyzed to present all-phased wave-packets.Thirdly,Bartlett and Blackman base window are selected to simulate 20 all-phased wave-packets,respectively.Finally,by means of these packets, it′s successfully to decompose two tested images.Experimental results demonstrate that all-phased wave-packets can be used to analyze the image in different layers,thereby achieving sub-images with multi-resolution in frequency domain and time domain.This novel method will potentially be applied in image compression,edge detection and adaptive signal processing fields.

Halide perovskites have properties of high photoluminescence quantum yield and narrow emission band,which can be used as a new X-ray detection material recently.But it also has problems of low light and thermal stability and lead toxicity.Here,we report an inorganic lead-free perovskite nanocrystal material of Cs3Cu2I5 prepared by using a solution method.Under X-ray radiation,we analyze its luminescence characteristics and explore the relationship between the conversion efficiency of Cs3Cu2I5 nanocrystal materials in different thicknesses and the X-ray source tube voltage.In the imaging experiment,we design an efficient Cs3Cu2I5 material scintillator conversion screen,then develop an X-ray imaging system based on it.Finally,we perform X-ray imaging of the razor blade and the snail shell respectively.The experimental results show that the emission peak of perovskite Cs3Cu2I5 reaches its peak at a wavelength of 460 nm under X-ray radiation;when the thickness is 0.30—0.50 mm,the conversion efficiency of Cs3Cu2I5 material is higher;the imaging results show that the image outline is clear and the internal structure of the object structured.Our results show that the lead-free perovskite Cs3Cu2I5 nanocrystal material can be used as a new type of scintillator,which has huge application potential in X-ray detection and imaging.

In a single-source modulation retro-reflector (MRR) free-space optical (FSO) system,the downlink signal causes crosstalk to the uplink remodulation signal of MRR terminal received by transceiver terminal,causing the system′s bit error rate performance to deteriorate.In this paper,we propose to use combinatorial logic (CL) inverse pulse position modulation (IPPM) scheme to eliminate the crosstalk of the downlink signal to the uplink remodulation signal to reduce the bit error rate of the single-source MRR FSO communication system.In this study,we built a CL-IPPM-based single-source MRR FSO communication experiment platform,and did simulation and full-duplex communication experiment based on it.The results show that the single-source MRR FSO communication system using the CL-IPPM scheme has a gain of about 5 dB.Furthermore,when the transceiver terminal and MRR terminal adopted IPPM, CL-IPPM scheme has the best bit error rate performance.In addition,we used the pseudo-phase conjugation property of the micro-corner cube retro-reflector array to improve the robustness of the MRR FSO system, and combining the CL-IPPM and the micro-corner cube retro-reflector array, the full-duplex communication based on single-source MRR FSO system can be realized with a bit error rate of 10－3 or less under weakly and medium turbulent environment.

In order to solve problems of long encoding time and high hardware acceleration resource occupancy rate caused by the advanced residual prediction (ARP) algorithm which does not make full use of the data characteristics of depth map in three-dimensional high efficiency video coding (3D-HEVC),a fast ARP selection algorithm based on reconfigurable hardware implementation is proposed.Firstly,the depth map is divided into three regions according to its data characteristics,and then the threshold is set to select the advanced residual prediction algorithm in different regions quickly,so as to reduce the coding time.The experimental results show that compared with the standard platform HTM16.1,the fast selection algorithm reduces the encoding time by 8.10% when the average peak signal to noise ratio 〖ＷＴＢＸ〗(PSNR) loss is only 0.019 dB.Finally,the dynamic programmable reconfigurable array processor (DPRAP) is used to accelerate the ARP fast selection algorithm in parallel,and then a reconfigurable implementation scheme is proposed based on the reconfiguration mechanism of the array processor,so as to accelerate the algorithm and reduce the hardware resource occupancy.The experimental results show that compared with the parallel scheme,the total number of the process element (PE) and instructions are respectively reduced by 50% and 33.23%,and the average speedup is 1.9.The algorithm before and after optimization is compared with disparity estimation,and the average speedup is 2.5.Therefore,this study has a certain reference value for real-time video coding of 3D-HEVC algorithm.