2024

• Record 13 of
  
  Title:Genetic Algorithm-Based Optimization of Arrhythmia Classification Model
  
  Author(s):Zhao, Jingpu(1); Feng, Zhengyang(1); Sun, Yiding(2); Xing, Runqiang(3); Hu, Kai(1)
  Source:2024 7th International Conference on Advanced Algorithms and Control Engineering, ICAACE 2024
  Volume:   Issue:   DOI: 10.1109/ICAACE61206.2024.10549254  Published: 2024  
  Abstract:The heart rate typically aligns with the electrical events within the heart, and an electrocardiogram (ECG) captures the electronic signals from the heart as they manifest on the surface of the body. By classifying and predicting the heart rate segments specifically, patients can be resuscitated in time. Currently, most of the commercially available problems on heart rate detection use machine learning and deep learning for automatic classification, and algorithms in related directions are slightly inadequate, resulting in still low accuracy and performance of classification and prediction models in specific situations. In this study, an arrhythmia classification algorithm model is developed for specific six heart rate risk classes by analyzing ECG data. A pruned and optimized decision tree method was first used to determine the danger classes of ECG data to attain real-time assessment and alert for arrhythmias. Then the model was optimized by using genetic algorithm, and by iterating the model parameters and selecting better parameter combinations, the final model accuracy improved by about 6%, the recall rate was increased by 13%, and the F1 value was increased by 16.8%, further improving the comprehensive performance of the model, which will provide reference for subsequent research in related directions and further promote the progress of research in related fields such as heart disease patient monitoring. © 2024 IEEE.
  Accession Number: 20242716656313
• Record 14 of
  
  Title:Motion detection of swirling multiphase flow in annular space based on electrical capacitance tomography
  
  Author(s):Zhao, Qing(1); Liao, Jiawen(1); Chen, Weining(1)
  Source:Proceedings of SPIE - The International Society for Optical Engineering
  Volume: 13090  Issue:   DOI: 10.1117/12.3026097  Published: 2024  
  Abstract:Cyclone multiphase flow in the annular space is widely used in fluid machinery, such as burner and pneumatic conveying. However, the annular flow field is complex, and the related research is not sufficient. To improve the safety and efficiency of equipment, this paper proposes a method for detecting the motion state of swirling fluid in annular space by integrating computational fluid dynamics (CFD) and electrical capacitance tomography (ECT), calculates the motion characteristics of swirling multiphase flow in the annular space using the CFD, and visually measures the distribution and motion state of swirling multiphase flow in the annular space using the ECT. Numerical simulation and experimental results show that the results of the two methods are in good agreement, indicating that the model selected in this paper in the CFD is correct. The CFD effectively reveals the distribution of swirling multiphase flow in the annular pipe, and the ECT can accurately reconstruct the position and size of swirling multiphase flow in the annular space. The combination of these two methods provides a new idea for the study of multiphase flow in annular space. © 2024 SPIE.
  Accession Number: 20241815993004
• Record 15 of
  
  Title:A stitching seams search strategy based on spectral image classification for hyperspectral image stitching
  
  Author(s):Liu, Hong(1,2); Hu, Bingliang(1); Hou, Xingsong(2); Yu, Tao(1)
  Source:2024 9th International Symposium on Computer and Information Processing Technology, ISCIPT 2024
  Volume:   Issue:   DOI: 10.1109/ISCIPT61983.2024.10673327  Published: 2024  
  Abstract:Hyperspectral image data is a form of data that combines images and spectra, and there are information differences between images in different bands when performing cube concatenation of hyperspectral data. A stitching seam search strategy based on hyperspectral spectral image classification is proposed to address the insufficient utilization of spectral dimension information in current data cube stitching methods. The main steps in searching for stitching seams are: Iteratively self-organizing data analysis algorithm (ISODATA) is used to classify two hyperspectral data cubes separately. Perform grayscale changes on the classification result images. Use graph cutting method to search for stitching seams on the transformed image. Apply the stitching seam to all bands to obtain the spliced hyperspectral data. The experimental results of applying this method to unmanned aerial hyperspectral data cubes captured by acousto-optic tunable filter (AOTF) spectral imager at waypoints show that our proposed method has certain advantages in both spatial and spectral dimensions compared to using stitching seams obtained from a single spectral segment image to achieve hyperspectral data cube stitching strategy. © 2024 IEEE.
  Accession Number: 20244117161963
• Record 16 of
  
  Title:CRNet: A Detail-Preserving Network for Unified Image Restoration and Enhancement Task
  
  Author(s):Yang, Kangzhen(1); Hu, Tao(1); Dai, Kexin(1); Chen, Genggeng(2); Cao, Yu(3); Dong, Wei(2); Wu, Peng(1); Zhang, Yanning(1); Yan, Qingsen(1)
  Source:IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops
  Volume:   Issue:   DOI: 10.1109/CVPRW63382.2024.00615  Published: 2024  
  Abstract:In real-world scenarios, images captured often suffer from blurring, noise, and other forms of image degradation, and due to sensor limitations, people usually can only obtain low dynamic range images. To achieve high-quality images, researchers have attempted various image restoration and enhancement operations on photographs, including denoising, deblurring, and high dynamic range imaging. However, merely performing a single type of image enhancement still cannot yield satisfactory images. In this paper, to deal with the challenge above, we propose the Composite Refinement Network (CRNet) to address this issue using multiple exposure images. By fully integrating information-rich multiple exposure inputs, CRNet can perform unified image restoration and enhancement. To improve the quality of image details, CRNet explicitly separates and strengthens high and low-frequency information through pooling layers, using specially designed Multi-Branch Blocks for effective fusion of these frequencies. To increase the receptive field and fully integrate input features, CRNet employs the High-Frequency Enhancement Module, which includes large kernel convolutions and an inverted bottleneck ConvFFN. Our model secured third place in the first track of the Bracketing Image Restoration and Enhancement Challenge, surpassing previous SOTA models in both testing metrics and visual quality. © 2024 IEEE.
  Accession Number: 20244217222948
• Record 17 of
  
  Title:Exploring the Connection between Eye Movement Parameters and Eye Fatigue
  
  Author(s):Sun, Weifeng(1,2,3); Wang, Yuqi(1,3); Hu, Bingliang(1,3); Wang, Quan(1,3)
  Source:Journal of Physics: Conference Series
  Volume: 2722  Issue: 1  DOI: 10.1088/1742-6596/2722/1/012013  Published: 2024  
  Abstract:Eye fatigue, a prominent symptom of computer vision syndrome (CVS), has gained significant attention in various domains due to the increasing diversification of electronic display devices and their widespread usage scenarios. The COVID-19 pandemic has further intensified the reliance on these devices, leading to prolonged screen time. This study aimed to investigate the effectiveness of utilizing eye movement patterns in discriminating fatigue during the usage of electronic display devices. Eye movement data was collected from subjects experiencing different levels of fatigue, and their fatigue levels were recorded using the T/CVIA-73-2019 scale. The analysis revealed that features related to the pupils demonstrated a high level of confidence and reliability in distinguishing fatigue, especially related to pupil size. However, features associated with fixations, such as fixation duration and frequency, did not significantly contribute to fatigue discrimination. Furthermore, the study explored the influence of subjective awareness on fatigue discrimination. By modifying the experimental settings and considering the subjects' subjective perception, it was observed that individual consciousness and self-awareness played a crucial role in fatigue discrimination. The implications of these findings extend beyond the field of computer vision syndrome, offering potential applications in developing interventions and strategies to alleviate eye fatigue and promote eye health among individuals who extensively use electronic display devices. © Published under licence by IOP Publishing Ltd.
  Accession Number: 20241916032392
• Record 18 of
  
  Title:Bracketing Image Restoration and Enhancement with High-Low Frequency Decomposition
  
  Author(s):Chen, Genggeng(1); Dai, Kexin(2); Yang, Kangzhen(2); Hu, Tao(1,2); Chen, Xiangyu(3); Yang, Yongqing(4); Dong, Wei(1); Wu, Peng(2); Zhang, Yanning(2); Yan, Qingsen(2)
  Source:IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops
  Volume:   Issue:   DOI: 10.1109/CVPRW63382.2024.00616  Published: 2024  
  Abstract:In real-world scenarios, due to a series of image degradations, obtaining high-quality, clear content photos is challenging. While significant progress has been made in synthesizing high-quality images, previous methods for image restoration and enhancement often overlooked the characteristics of different degradations. They applied the same structure to address various types of degradation, resulting in less-than-ideal restoration outcomes. Inspired by the notion that high/low frequency information is applicable to different degradations, we introduce HLNet, a Bracketing Image Restoration and Enhancement method based on high-low frequency decomposition. Specifically, we employ two modules for feature extraction: shared weight modules and non-shared weight modules. In the shared weight modules, we use SCConv to extract common features from different degradations. In the non-shared weight modules, we introduce the High-Low Frequency Decomposition Block (HLFDB), which employs different methods to handle high-low frequency information, enabling the model to address different degradations more effectively. Compared to other networks, our method takes into account the characteristics of different degradations, thus achieving higher-quality image restoration. © 2024 IEEE.
  Accession Number: 20244217223292
• Record 19 of
  
  Title:Radiation Pattern Computation with the Variable Resolution for Reflector Antennas using the CZT
  
  Author(s):Zhang, Yang(1); Xiang, Binbin(1); Lin, Shangmin(2); Zhao, Yongqin(1); Mo, Shike(1); Wang, Wei(1)
  Source:2024 IEEE 7th International Conference on Electronic Information and Communication Technology, ICEICT 2024
  Volume:   Issue:   DOI: 10.1109/ICEICT61637.2024.10670916  Published: 2024  
  Abstract:The Fast Fourier Transform (FFT) is widely used in antenna near-far field transformation. When FFT is applied in analyzing signal frequency bands that cannot be densely sampled, the frequency resolution obtained will be relatively low. Applying the Chirp-Z Transform (CZT) to the near-far field transformation highlights the characteristic of the CZT that allows the input points to be equally angle-sampled and then resampled. This expands the sampling quantity to any integer length. By using CZT, the antenna measurement points can be sampled and transformed into an arbitrary number of frequency points. This approach can change the resolution of sample points in the near-far field transformation, allowing for frequency spectrum refinement in specific narrow frequency bands. © 2024 IEEE.
  Accession Number: 20244117164503
• Record 20 of
  
  Title:Splicing Design Method and Accuracy Analysis of the U-shaped Frame
  
  Author(s):Han, Jingyu(1); Li, Xiangyu(2); Xie, Meilin(2); Hao, Wei(2); Lian, Xuezheng(2); Wang, Jie(1,2); Song, Wei(2); Ruan, Ping(2)
  Source:2024 5th International Conference on Mechatronics Technology and Intelligent Manufacturing, ICMTIM 2024
  Volume:   Issue:   DOI: 10.1109/ICMTIM62047.2024.10629508  Published: 2024  
  Abstract:The U-shaped frame is a crucial component of the theodolite. The coaxial accuracy of the bearing holes on both sides of the U-shaped frame directly affects the pitch-axis accuracy and further influences the angular measurement precision of the theodolite. The machining of the bearing holes on both sides of the U-shaped frame with a long span requires the use of methods such as a large-stroke boring machine or the assembly and adjustment of the frame's own structure. However, considering various factors such as the manufacturing cost, manufacturing time, and machining precision of the U-shaped frame, neither of the above methods is the optimal solution. This paper proposes a splicing design method for the U-shaped frame based on the principle of one side and two pins positioning. This method solves the problems of low coaxial accuracy and difficult machining of bearing holes in U-shaped frames with long spans. Through multiple splicing accuracy tests of the coaxiality of the bearing holes in a U-shaped frame with a span of 1500 mm, the average coaxiality error of the bearing holes is 0.023 mm, with a maximum value of 0.038 mm, which is less than the theoretical maximum coaxiality error of 0.058 mm. The coaxiality tolerance precision level is close to Grade 5, which meets the requirements for the use of high-precision theodolite. ©2024 IEEE.
  Accession Number: 20243616988986
• Record 21 of
  
  Title:A 2λ×100 Gb/s Optical Receiver with Si-Photonic Micro-Ring Resonator and Photo-Detector for DWDM Optical-IO
  
  Author(s):Chen, Sikai(1); Xue, Jintao(2,3); Chen, Yihan(3); Gu, Yuean(3); Yin, Haoran(1,3); Bao, Shenlei(2,3); Li, Guike(1,3); Wang, Binhao(2,3); Qi, Nan(1,3)
  Source:Proceedings of the Custom Integrated Circuits Conference
  Volume:   Issue:   DOI: 10.1109/CICC60959.2024.10529008  Published: 2024  
  Abstract:The emerging AI computing system asks for high-speed, large-scale, and power-efficient interconnects. As the system scales-out reaching tens-of-meters, electrical links cannot support higher bandwidth (BW) at affordable power consumption. Silicon photonic (SiPh) technology enables the fabrication of both photonic integrated circuits (PIC) and electronic integrated circuits (EIC) on the same wafer. By integrating optical transceivers into the xPU package, fiber channels could be directly attached to the chip edge, building up the highly integrated optical-IO. SiPh micro-ring resonator (MRR) is an attractive solution due to small footprint and its capability of wavelength selection [1]-[3]. To fit more wavelengths into one full-spectral-range (FSR) of the MRR, dense wavelength-division multiplexing (DWDM) has been adopted, which brings design challenges on anti-aliasing and wavelength stabilizing. © 2024 IEEE.
  Accession Number: 20242216153064
• Record 22 of
  
  Title:Denoising Algorithm based on Event Camera
  
  Author(s):Lv, Yuanyuan(1,2); Liu, Zhaohui(1); Zhou, Liang(1); Qiao, Wenlong(1,2); Zhang, Haiyang(1,2)
  Source:Proceedings of SPIE - The International Society for Optical Engineering
  Volume: 13154  Issue:   DOI: 10.1117/12.3016236  Published: 2024  
  Abstract:The event camera is a novel type of bio-inspired vision sensor inspired by the biological retina. Compared to traditional frame-based cameras, it offers high temporal resolution, high dynamic range, reduced redundancy, and lower transmission bandwidth. These unique features pave the way for innovative solutions in the field of computer vision. However, the heightened sensitivity of event cameras to fluctuations in brightness, along with their susceptibility to environmental factors and hardware limitations, presents a significant challenge. It involves capturing spatiotemporal information from the target signal simultaneously with the generation of a substantial volume of noise events. In applications relying on event cameras, this noise compromises target detection precision. Therefore, event stream denoising is essential before further applications can be pursued. Unfortunately, conventional frame-based algorithms are ill-suited for processing event data due to the distinct format of event cameras. In response to the challenges of event stream denoising, using the event stream generated by Celex-V as an example, this paper categorizes noise events and conducts an analysis of the event noise distribution model. Leveraging the characteristics of noise events, such as randomness and isolation, the paper proposes an event-based cascaded noise processing method. This method involves analyzing events in the spatiotemporal vicinity of arriving events and removing noise events from the event stream data. While ensuring the integrity of data flow information, it achieves rapid and efficient noise removal. The denoised event stream is advantageous for subsequent processing in various applications based on event cameras. © 2024 SPIE.
  Accession Number: 20242016095187
• Record 23 of
  
  Title:DC-KD: double-constraint knowledge distillation for optical satellite imagery object detection based on YOLOX model
  
  Author(s):Yang, Hongbo(1,2); Qiu, Shi(1); Feng, Xiangpeng(1)
  Source:Proceedings of SPIE - The International Society for Optical Engineering
  Volume: 13176  Issue:   DOI: 10.1117/12.3029285  Published: 2024  
  Abstract:Object detection is an important application of optical satellite remote sensing imagery interpretation. Since the objects of interest, such as aircraft, ships, and vehicles, are small in size with obscure contour and texture, it is difficult for object detection in satellite images. The spatial resolution of aerial images is higher than satellite images, and the object detection model can achieve higher precision. Knowledge distillation has been validated as an effective technique by learning the common features of aerial and satellite images to improve the precision of object detection in satellite images. It means that a teacher model pre-trained on aerial image datasets guides the training of a compact student model on satellite image datasets. However, there are data distribution differences between aerial images and satellite images. The distribution differences may cause the teacher model to give guidance signals that deviate from the ground truth, thus leading to sub-optimization of the student model. In this paper, we proposed a new distillation scheme, termed DC-KD, which updates the teacher model using the predictions of the teacher model that are inconsistent with the ground truth, and the rest are used to guide the training of the student model. We achieved a 3.88% mAP50 improvement on the xView dataset based on the YOLOX-S model. © 2024 SPIE.
  Accession Number: 20242316199229
• Record 24 of
  
  Title:Prediction of Cognitive Impairment in Epilepsy Patients Based on EEG Signal with Residual Block-Based Network
  
  Author(s):Rong, Yan(1); Wang, Yuqi(1); Wei, Xiaojie(1); Feng, Li(2); Hu, Bingliang(1); Wang, Quan(1)
  Source:2024 9th International Conference on Computer and Communication Systems, ICCCS 2024
  Volume:   Issue:   DOI: 10.1109/ICCCS61882.2024.10603033  Published: 2024  
  Abstract:Mild cognitive impairment (MCI) is an irreversible, gradual neurological disease and one of the main complications of epilepsy. The scale proves effective in detecting cognitive impairment, but it relies on labor-intensive processes and is easily affected by patients' subjective behavior. Using machine learning methods to analyze EEG data is a promising alternative for detecting MCI. However, the increasing amount of EEG data affects the efficiency of detection. We innovatively propose a novel deep learning (DL) network based on residual blocks to overcome this issue. The DL network aims to efficiently detect cognitive impairment in epilepsy patients by analyzing unique EEG data collected during the Attention Network Test (ANT) and predicting patients' scale score. The suggested network consists of four phases: band-pass filter, spatial convolution block, residual block, and classifier. The entire proposed framework comprises four steps: collecting EEG data, preprocessing the raw data, extracting data features, and classification between MCI subjects and normal ones. Data from 92 patients with epilepsy were used for training and performance evaluation of the network. The classification performance of the proposed network has been compared with that of ResNet18, ResNet34, EEGNet, and FBCNet. The experimental results shows that the proposed network achieved the best classification performance among all five tested networks. The proposed network could also be used for MCI prediction, in which task it achieved about 0.05 in MAE while predicting the score of MOCA for patients, demonstrating a considerable predictive result. Five-fold cross-validation was used to assess the framework's stability. © 2024 IEEE.
  Accession Number: 20243416904442