Chinese Version Light学术出版中心
Latest Issue
Issue 9 ， 2022
-
Bionic thermal imaging with cross-shaped four-aperture partially overlapped field of view
Abstract：To alleviate the contradiction between the field of view and resolution in conventional thermal imaging systems with limited detector technology， the multi-aperture layout mode is studied with image stitching algorithm in the non-overlapped areas and super-resolution processing algorithm in the overlapped areas. The partially overlapped bionic thermal imaging modes of cross-shaped four-aperture and cross-shaped five-aperture field of view are proposed for the multi-aperture layout mode to compare and analyze their uniqueness with respect to the Tian-shaped four-aperture. Using ULIS's miniature uncooled infrared focal plane detector Micro80 GEN2， a system of bionic thermal imaging is developed with cross-shaped four-aperture partially overlapped field of view. This provides a visual mode similar to human vision with high-resolution imaging in the center and large-field search in the periphery， using the non-overlapped areas to increase the imaging field of view and the sub-pixel micro-shift in the overlapped areas to achieve super-resolution. Image processing is performed on the resolution target image， wherein an external scene image is obtained by the experimental system， with blind element removal， non-uniformity correction， image stitching based on perspective transformation， and super-resolution using a spatial integration model for the focal plane detector. The spatially variable resolution images obtained are evaluated both qualitatively and quantitatively. The size of the experimental system is 180 mm×100 mm ×100 mm； the stitching total field of view is 2.36 times that of a single-aperture camera； the overlapping area of the four apertures accounts for 7.78% of the total field of view， and the imaging resolution is improved. The imaging system uses multiple sub-apertures to form multi-view stereo vision to realize stereo imaging. Moreover， it can realize polarization thermal imaging or multi-spectral imaging by installing polarizers or filters， making it suitable for a variety of applications.Keywords：thermal imaging;Multi-aperture;bionic compound eye;layout mode;super resolution- 351
- |
- 81
- |
- 0
Published：2022-05-26 -
Laser interference induced backward transfer to prepare silver micro-stripe structure
Abstract：To optimize the parameters of laser-interference-induced backward transfer （LIIBT） for preparation of micro-structures， this study investigated the influence of laser pulse quantity and fluence on LIIBT， analyzed the formation mechanism of Ag micro-stripes， and imaged the Ag micro-stripes structure on ciprofloxacin using surface-enhanced Raman spectroscopy （SERS）. The silver micro-stripe structures were prepared using two-beam LIIBT at atmosphere in which a sodium-calcium glass and silver film were used as the receiving substrate and target material， respectively. The scanning electron microscopy （SEM） results showed that the boundaries of the silver micro-stripe became clear by increasing the pulse quantity and laser fluence. The energy-dispersive x-ray spectroscopy （EDS） results showed that the micro-stripe structure was composed of silver nanoparticles. In addition， although the concentration of ciprofloxacin was reduced to 10-8 mol/L， its effect on the SERS results was apparent in terms of the silver micro stripes. Finally， the transferring mechanism of silver was discussed based on the secondary pulse. This study optimized the experimental parameters for preparing a silver micro-stripe structure by LIIBT， examined its formation mechanism， verified that its structure had noticeable SERS effects to detect low ciprofloxacin concentrations， and provided a SERS substrate for high sensitivity antibiotic residue detection in environmental pollution and food engineering.Keywords：laser interference induced backward transfer（LIIBT）;micro-stripe structure;silver nanoparticles;surface enhanced Raman scattering（SERS）;ciprofloxacin- 179
- |
- 34
- |
- 0
Published：2022-05-26 -
High-resolution light field imaging system using liquid crystal diffuser
Abstract：A new light field imaging system based on a liquid crystal diffuser was proposed to address the problem of low spatial bandwidth in conventional light field cameras. Using the light-controlled orientation technique， the electronically adjustable liquid crystal diffuser was prepared for light field imaging by eliminating the impurities and electrostatic defects caused by conventional frictional orientation. The applied voltage was changed to obtain light field information under different voltages. Using the proposed algorithm， the liquid crystal diffuser was adapted to fuse the detailed information under high transmittance and scattered light intensity under low transmittance to obtain high-resolution light field information with balanced light intensity and detail. Compared with the conventional light field imaging system， the proposed method improves the PSNR value of the reconstructed high-resolution light field image by approximately 4.98%. In addition， the proposed method has low cost， simple preparation， convenient implementation， making it suitable for application to cost-effective light field cameras that realize high resolution light field imaging.Keywords：light field imaging;liquid crystal diffuser;light-controlled orientation;spatial bandwidth product- 221
- |
- 44
- |
- 0
Published：2022-05-26 -
Automatic attitude measurement of laser tracker based on deep learning and PnP model
Abstract：In view of the urgent demand for attitude measurement in high-end manufacturing applications， such as aerospace and automobile assembly， a fast and high-precision attitude measurement method for a laser tracker was proposed. The method employed deep learning in conjunction with the visual PnP model to realize automatic attitude measurement of the laser tracker. The correspondence between 3D feature points and 2D feature points required by the traditional PnP model were directly determined through a feature extraction network designed to extract high-dimensional features. The joint probability distribution between feature vectors was determined using optimal transmission theory to complete the matching of 3D-2D feature points. Subsequently， Ransac-P3P combined with EPnP algorithm was used to obtain high-precision attitude information； Based on this， the Jacobian matrix of PnP solution process was calculated using implicit differential theory， and the PnP attitude solution model was integrated into the network to guide the training of the network. The complementary advantages of strong depth network matching ability and high attitude solution accuracy of the PnP model improved the solution accuracy of the network. In addition， a dataset with rich annotation information was used to train the attitude measurement network for the laser tracker. Finally， an attitude measurement test was conducted using a high-precision two-dimensional turntable. The experimental results show that the calculation error of pitch angle is less than 0.31°， the rolling angle error is less than 0.03°， and the single measurement takes approximately 40 ms. The proposed method can potentially be applied to attitude measurement scene of the laser tracker.Keywords：laser tracker;attitude measurement;monocular vision;deep learning- 149
- |
- 49
- |
- 0
Published：2022-05-26
Modern Applied Optics
-
Linear yaw compound piezoelectric micro-motion platform
Abstract：To extend the applications of micro/nano technology in the fields of optics， precision machining， and precise positioning， this study introduces the design， analysis， and testing process for a two-degree-of-freedom micro-movement platform to achieve linear and yaw motion. First， the main composition and working principle of the designed platform are introduced. Subsequently， the mathematical models of the two motions are established and simulated to assess motion performance. A prototype is fabricated considering the size of the piezoelectric platform， and an experimental platform is established to test the motion capability and frequency response. Finally， the precise control of the platform is realized via proportional integral derivative method. The simulation and experimental results demonstrate that the platform achieves good motion accuracy and decoupling performance， with the first natural frequency reaching 333.8 Hz. Moreover， the platform can achieve 24.924 μm linear motion and 1.330 mrad angular swing. The designed structure can serve as reference for the design of micro-motion platform and the research of high-precision motion.Keywords：micro-motion platform;linear motion;yaw motion;flexible mechanism;mathematical model- 176
- |
- 38
- |
- 1
Published：2022-05-26 -
Three-dimensional reconstruction and filtration behavior analysis system for nanofiber membrane
Abstract：To realize effective theoretical calculation and analyses of a nanofiber air filtration membrane， an analysis system was developed to calculate the air filtration performance of the nanofiber membrane. The 3D topology reconstruction algorithm， the calculation module of the filtration efficiency， and the calculation method of pressure drops were investigated. First， feature extraction of the nanofiber membrane was conducted using denoising， binarization， thinning algorithm， and discretization， and the three-dimensional reconstruction was realized through the definition， storage， and visual rendering of the feature information of the nanofiber membrane. Subsequently， through collision detection and mechanical analyses of particles and nanofibers， the number of particles captured by the nanofiber membrane was determined to calculate the filtration efficiency. Finally， according to Poiseuille's law and the pressure drop principle， a filtration pressure drop calculation module based on SEM images was realized. The numerical results of both the topological structure and thickness of the nanofiber membrane were compared with experimentally determined values. The comparison results show that the calculation error of filtration efficiency is less than 10%， and the calculation error of filtration pressure drop is less than 20%. The trend of calculation results is consistent with the experimental results， which reflects the difference in the filtration performance of different topologies.Keywords：nanofiber membrane;3D reconstruction;analysis system;filtration efficiency;filtration pressure drop- 101
- |
- 15
- |
- 0
Published：2022-05-26
Micro/Nano Technology and Fine Mechanics
-
Improved BYOL method for predicting epidermal growth factor receptor gene mutations in non-small cell lung cancer
Abstract：The epidermal growth factor receptor （EGFR） mutation status can predict whether a patient has non-small cell lung cancer （NSCLC）. A self-supervised EGFR gene mutation prediction method based on contrastive learning is proposed， which can distinguish between negative and positive images of the patient’s lesion area input to the network， without requiring a large number of expert hand-labeled patient datasets. The self-supervised BYOL network was modified to increase the number of layers of the non-linear multilayer perceptron （MLP） of the network projection layer， and image data of the patient's CT and PET modalities were merged as the input of the network. Negative and positive medical records can be predicted without the need to annotate a large number of patient datasets. Using the non-small cell lung cancer EGFR gene mutation datasets， it is compared with traditional radiomics， supervised VGG-16 network， supervised ResNet-50 network， supervised Inception v3 network， and unsupervised transfer learning CAE. The experimental results show that the instance features of patient lesion area images learned from CT and PET images of patients using contrastive learning can be used to distinguish negative and positive cases， with an area under the curve （AUC） of 77%. The classification results improved by AUC of 7% compared to the traditional radiomics method， and by AUC of 5% compared to the classification results of the supervised VGG-16 network. The AUC is only 9% lower than that of supervised ResNet-50， without requiring a large number of expert hand-annotated datasets and large patient clinical datasets. The improved BYOL network proposed in this paper requires only a small number of labeled patient datasets to obtain more accurate prediction results than some traditional supervised methods， demonstrating its potential to help clinical decision-making.Keywords：medical image processing;deep learning;contrastive Learning;PET/CT;prediction of gene mutations in lung non-small cell lung cancer- 134
- |
- 25
- |
- 0
Published：2022-05-26 -
Multi-level features progressive refinement and edge enhancement network for image dehazing
Abstract：An ordinary differential equation （ODE）-based multi-level feature progressive refinement and edge enhancement network is proposed for image dehazing to provide an effective convolutional neural network framework with an algorithm designed to preserve edges while improving accuracy. The study mainly comprises subnetworks of multi-level feature extraction， ODE-based progressive refinement， and edge enhancement. First， the multi-level features extraction subnetwork is leveraged to extract low-level features with detailed information and high-level features with semantic information from hazy images， to enable the subsequent dehazing operations. Second， a novel Leapfrog module is designed based on the relationship between the residual framework and ODE solver by cascading Leapfrog modules to model an approximation solution for ODEs. Finally， the progressive refinement subnetwork is developed. Notably， each Leapfrog module refines the output of the previous Leapfrog with alternative low/high-level features. Finally， motivated by the effectiveness of edge enhancement via second-order differential operators in the edge enhancement network， the edge of the dehazing result predicted by the last Leapfrog module is calculated using the pretrained UNet and added back into dehazing to enhance edges and preserve details. The experimental results demonstrate that the proposed method outperforms the existing methods on both synthetic images and real images quantitatively as well as qualitatively. The dehazing accuracy is improved by 5% and the runtime is only 0.032 s. Hence， the proposed method can be incorporated into practical dehazing applications in engineering.Keywords：image dehazing;ordinary differential equation;leapfrog method;edge enhancement;convolutional neural network- 198
- |
- 35
- |
- 0
Published：2022-05-26 -
Straw coverage detection of conservation tillage farmland based on improved U-Net model
Abstract：To meet the technical requirements of straw return monitoring in conservation tillage， an improved U-Net semantic segmentation algorithm was proposed to detect straw coverage rate. First， a novel convolution module was developed to replace the convolution module in the original U-Net framework； Second， the inception structure was improved， and stripe pooling and efficient spatial pyramid hollow convolution modules were introduced for the design of a new Gception structure. Finally， an attention mechanism was introduced into this module. The effectiveness and progressiveness of the optimized U-Net model were evaluated on farmland surface images collected by a drone， and comparative experiments were conducted with the U-Net， PSP-Net， Link-Net， Res-Net， DSRA-Unet， and DE-GWO algorithms， obtaining mean intersection over union of 80.05%， an average pixel accuracy of 91.20%， and an average coverage error of 0.80%. The experimental results demonstrate that the segmentation capability of the improved U-Net model was superior to those of the other algorithms， indicating the effectiveness of feature extraction and the integrity of global features， by effectively eliminating tree shadows and other interference factors over the farmland. These results prove that the proposed model achieves better segmentation not only in complex farmland scenes with agricultural machinery and tree shadow interference but also in large-scale images. Moreover， the model provides efficient algorithms for large-area straw coverage detection and other related image detection methods.Keywords：straw image;coverage detection;semantic segmentation;U-net model;attention mechanism- 137
- |
- 39
- |
- 0
Published：2022-05-26 -
Adaptive octree 3D image reconstruction based on plane patch
Abstract：In this study， an adaptive octree convolutional neural network based on plane patches is proposed for effective 3D shape encoding and decoding. Unlike volume-based or octree-based convolutional neural network （CNN） methods， which represent 3D shapes with the same voxel resolution， the proposed method can use planes and adaptively represent the 3D shapes of octree nodes with different levels. The patch models the 3D shape within each octree node， whereby the patch-based adaptive representation is utilized in the proposed adaptive patch octree convolutional neural network （O-CNN） encoder and decoder for the encoding and decoding of 3D shapes. The adaptive patch O-CNN encoder takes the plane patch normal and displacement as input and performs three-dimensional convolution on the octree nodes of each level， whereas the adaptive patch O-CNN decoder infers each level. The shape occupancy rate and subdivision state of the octree node as well as the best plane normal and displacement of each leaf octree node are estimated. As a general framework for 3D shape analysis and generation， adaptive patch O-CNN not only reduces memory and computational costs but also exhibits better shape generation capabilities than existing 3D-CNN methods. Shape prediction is performed using a single image to verify the efficiency and effectiveness of the generation task of the adaptive O-CNN. The chamfer distance error is 0.274， which is lower than that of OctGen （0.294）， resulting in a better reconstruction effect.Keywords：computer vision;three-dimensional reconstruction;convolutional neural network;neural network- 105
- |
- 27
- |
- 0
Published：2022-05-26 -
Infrared and visible image fusion based on alternating gradient filter and improved PCNN
Abstract：To overcome the problem of image blur at the edge of object in the process of infrared and visible image fusion， an image fusion method based on the alternating gradient filter and improved pulse coupled neural network （PCNN） was proposed. In this paper， a novel alternating gradient filter （AGF） was proposed based on the gradient filter， which combines the rolling guide filter （RGF） and the smooth iterative recovery filter （SIRmed）， with corresponding characteristics of local strength retention and edge recovery. The source images were decomposed into the approximate layer and residual layer using the AGF. The multi-scale morphological operator and the fusion rule of maximum region energy for the source images were adopted for the approximation layer， and then the residual layer was fused with the improved parameter adaptive PCNN fusion rule. Finally， the fusion result was reconstructed by an alternating gradient filter. The experimental results show that compared with the other five fusion methods， the objective evaluation indices of this method， that is the average gradient （AG）， standard deviation （STD）， information entropy （EN）， spatial frequency （SF）， edge intensity （EI）， and visual fidelity （VIFF） increase by 18%， 10%， 2.8%， 16%， 51%， and 11.2%， respectively. In addition， the results demonstrate that the proposed AGF fusion method can overcome the shadow on the edge of the object and reserve the brightness， edge， detail， and texture information of the source images.Keywords：image processing;infrared and visible image fusion;alternating gradient filter;multiscale morphological operator;pulse coupled neural network- 166
- |
- 47
- |
- 0
Published：2022-05-26
Information Sciences
Loading...
0