2024

• Record 25 of
  
  Title:Consumer Camera Demosaicking and Denoising With a Collaborative Attention Fusion Network
  
  Author(s):Yuan, Nianzeng(1); Li, Junhuai(2); Sun, Bangyong(3,4)
  Source: IEEE Transactions on Consumer Electronics  Volume: 70  Issue: 1  DOI: 10.1109/TCE.2023.3342035  Published: February 1, 2024  
  Abstract:For the consumer cameras with Bayer filter array, raw color filter array (CFA) data collected in real-world is sampled with signal-dependent noise. Various joint denoising and demosaicking (JDD) methods are utilized to reconstruct full-color and noise-free images. However, some artifacts (e.g., remaining noise, color distortion, and fuzzy details) still exist in the reconstructed images by most JDD models, mainly due to the highly related challenges of low sampling rate and signal-dependent noise. In this paper, a collaborative attention fusion network (CAF-Net), with two key modules, is proposed to solve this issue. Firstly, a multi-weight attention module is proposed to efficiently extract image features by realizing the interaction of spatial, channel, and pixel attention mechanisms. By designing a local feedforward network and mask convolution aggregation of multiple receptive fields, we then propose an effective dual-branch feature fusion module, which enhances image details and spatial correlation. Accordingly, the proposed two modules significantly facilitate our CAF-Net to recover a high-quality image, by accurately inferring the correlations of color, noise, and the spatial distribution of the CFA data. Extensive experiments on demosaicking, synthetic, and real image JDD tasks prove that the proposed CAF-Net can achieve advanced performance in terms of objective evaluation index metrics and visual perception. © 2023 IEEE.
  Accession Number: 20235115239885
• Record 26 of
  
  Title:A Detection Method for Typical Component of Space Aircraft Based on YOLOv3 Algorithm
  
  Author(s):He, Bian(1,2,3); Jianzhong, Cao(1,3); Cheng, Li(1,3); Junpeng, Dong(1,3); Zhongling, Ruan(1,3); Chao, Mei(1,3)
  Source: 2024 IEEE 3rd International Conference on Electrical Engineering, Big Data and Algorithms, EEBDA 2024  Volume:   Issue:   DOI: 10.1109/EEBDA60612.2024.10485846  Published: 2024  
  Abstract:A solar panel recognition method based on YOLOv3 deep learning algorithm is proposed to address issues such as inaccurate recognition of traditional algorithms in space solar panel detection. First, this paper scales the dataset images to 416 × 416, then uses Labelme to annotate the data and transform the bounding box position information, and finally uses the YOLOv3 algorithm framework for model training. The results show that the recall, F1 score and accuracy of YOLOv3 algorithm are all above 80%. The YOLOv3 deep learning algorithm meets the requirements for real-time detection of solar panels in terms of accuracy. © 2024 IEEE.
  Accession Number: 20241715982706
• Record 27 of
  
  Title:High-performance architecture for real-time high-definition short-wave infrared streaming video processing and its field programmable gate array prototype
  
  Author(s):Zhou, Feng(1,2,3); Chen, Zhiqiang(1,2,3); Xie, Qingsheng(1,3); Kong, Fanzi(1,2,3); Chen, Yaohong(1,3); Wang, Huawei(1,3)
  Source: Optical Engineering  Volume: 63  Issue: 2  DOI: 10.1117/1.OE.63.2.023103  Published: February 1, 2024  
  Abstract:Image detail enhancement is critical to the performance of short-wave infrared (SWIR) imaging systems. Recently, the requirement for real-time processing of high-definition (HD) SWIR video has shown rapid growth. Nevertheless, the research on field programmable gate array (FPGA) implementation of HD SWIR streaming video processing architecture is relatively few. This work proposes a real-time FPGA architecture of SWIR video enhancement by combining the difference of Gaussian filter and plateau equalization. To accelerate the algorithm and reduce memory bandwidth, two efficient key architectures, namely edge information extraction and equalization and remapping architecture, are proposed to sharpen edges and improve dynamic range. The experimental results demonstrated that the proposed architecture achieved a real-time processing of 1280 × 1024@60Hz with 2.7K lookup tables, 2.5K Slice Reg, and about 350 kb of block RAM consumption, and their utilization reached 12.5%, 19.2%, and 12.5% for the XC7A200T FPGA board, respectively. Moreover, the proposed architecture is fully pipelined and synchronized to the pixel clock of output video, meaning that it can be seamlessly integrated into diverse real-time video processing systems. © 2024 Society of Photo-Optical Instrumentation Engineers (SPIE)
  Accession Number: 20241115712686
• Record 28 of
  
  Title:An efficient multi-scale transformer for satellite image dehazing
  
  Author(s):Yang, Lei(1,2); Cao, Jianzhong(1,2); Chen, Weining(1); Wang, Hao(1); He, Lang(3,4,5)
  Source: Expert Systems  Volume:   Issue:   DOI: 10.1111/exsy.13575  Published: 2024  
  Abstract:Given the impressive achievement of convolutional neural networks (CNNs) in grasping image priors from extensive datasets, they have been widely utilized for tasks related to image restoration. Recently, there is been significant progress in another category of neural architectures—Transformers. These models have demonstrated remarkable performance in natural language tasks and higher-level vision applications. Despite their ability to address some of CNNs limitations, such as restricted receptive fields and adaptability issues, Transformer models often face difficulties when processing images with a high level of detail. This is because the complexity of the computations required increases significantly with the image's spatial resolution. As a result, their application to most high-resolution image restoration tasks becomes impractical. In our research, we introduce a novel Transformer model, named DehFormer, by implementing specific design modifications in its fundamental components, for example, the multi-head attention and feed-forward network. Specifically, the proposed architecture consists of the three modules, that is, (a) multi-scale feature aggregation network (MSFAN), (b) the gated-Dconv feed-forward network (GFFN), (c) and the multi-Dconv head transposed attention (MDHTA). For the MDHTA module, our objective is to scrutinize the mechanics of scaled dot-product attention through the utilization of per-element product operations, thereby bypassing the need for matrix multiplications and operating directly in the frequency domain for enhanced efficiency. For the GFFN module, which enables only the relevant and valuable information to advance through the network hierarchy, thereby enhancing the efficiency of information flow within the model. Extensive experiments are conducted on the SateHazelk, RS-Haze, and RSID datasets, resulting in performance that significantly exceeds that of existing methods. © 2024 John Wiley & Sons Ltd.
  Accession Number: 20241315812824
• Record 29 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 30 of
  
  Title:NVPCA Image Enhancement-Based Detection Method for Sidelobe Peak Parameters in Weak Signal Regions
  
  Author(s):Wang, Zhengzhou(1); Wang, Li(1); Duan, Yaxuan(1); Li, Gang(1); Wei, Jitong(1)
  Source: Zhongguo Jiguang/Chinese Journal of Lasers  Volume: 51  Issue: 6  DOI: 10.3788/CJL231185  Published: 2024  
  Abstract:Objective The primary application of the host device involves research in high-energy density physics and inertial confinement fusion, handling energies up to 100000 joules. A significant challenge encountered during these experiments is the simultaneous detection of strong and weak signals in the far-field focal spot. Specifically, accurately measuring weak signals in the sidelobe area of the far-field focal spot has proven difficult. To address this, we introduce a peak parameter detection method for weak signal regions in the sidelobe, leveraging neighborhood vector principal component analysis (NVPCA) for image enhancement. Methods Our optimization strategy includes several steps. First, we treat each pixel in the sidelobe image and its eight neighboring pixels as a column vector to construct a 9-dimensional data cube. The first dimension post-PCA transformation, the NVPCA image, is then selected. Next, we employ angle transformation to detect various peak parameters of the one-dimensional sidelobe curve in all directions, facilitating the quantification of energy distribution in the sidelobe’s weak signal area. Subsequently, we identify the maximum position points of each sidelobe peak in all directions, linking these to form a maximum ring for each peak and calculating the grayscale mean of these rings. The smallest grayscale mean exceeding the LCM target separation threshold is identified as the minimum measurable signal for the entire sidelobe beam. Results and Discussions 1) We propose a sidelobe weak signal detection method using NVPCA image enhancement. This approach successfully isolates and extracts the minimum measurable signal from the 5th peak ring on the sidelobe image’s periphery, increasing the dynamic range ratio to 1.528 times. This method enhances the peak’s maximum value in any direction, ensuring the extraction of the minimum measurable signal from the peripheral 5th peak loop. 2) The LCM target detection threshold formula is employed to segregate the minimum measurable signal. This formula, tailored to the characteristics of far-field focal lobe images, effectively separates background noise. 3) We validate the one-dimensional curve peak parameters in various directions using a two-dimensional plane display method. Combining two-dimensional and one-dimensional displays, this method not only showcases the peak parameter distribution of one-dimensional sidelobe curves from multiple perspectives but also differentiates adjacent sampling angles’peak positions. The validation using equations (11) – (13) yields rising edge, falling edge, and pulse width consistent with those in Table 5, confirming the two-dimensional display method’s efficacy in verifying one-dimensional curve peak parameters. Conclusions Addressing the challenge of extracting the smallest measurable signal in the sidelobe image’s periphery for strong laser far-field focal spot measurements, we introduce a sidelobe weak signal region peak parameter detection method based on NVPCA image enhancement. Our findings demonstrate this method’s capability to isolate and extract the minimum measurable signal from sidelobe image peripheral peaks, increasing the dynamic range ratio to 1.528 times. This approach is crucial for accurately measuring weak signal areas in sidelobe beams, understanding their energy distribution, and laying the groundwork for future precise measurements of strong laser far-field focal spots in large-scale laser devices. © 2024 Science Press. All rights reserved.
  Accession Number: 20241215768417
• Record 31 of
  
  Title:Characterization of primary silicate minerals in Earth-like bodies via Raman spectroscopy
  
  Author(s):Huang, Shuaidong(1,2); Xue, Bin(1,2); Zhao, Yiyi(1,2); Yang, Jianfeng(1,2)
  Source: Journal of Raman Spectroscopy  Volume:   Issue:   DOI: 10.1002/jrs.6657  Published: 2024  
  Abstract:The examination and identification of silicate minerals are critical for advancing our understanding of the evolutionary journey of Earth-like bodies. To facilitate an efficient and productive process, it is imperative that these minerals be detected swiftly and accurately. This study is designed to explore the relationship between varying concentrations of cations and their corresponding Raman shifts. The focus is on primary silicate minerals in Earth-like bodies, specifically olivine, pyroxene, and feldspar, utilizing data from the RRUFF database. Employing a fitting formula, we identify distinct Raman peak ranges associated with different silicate minerals. Our research covers a wide array of mineral types, including five varieties of olivine (forsterite [Mg2SiO4], fayalite [Fe2+2SiO4], tephroite [Mn2+2SiO4], monticellite [CaMgSiO4], and kirschsteinite [CaFe2+SiO4]), four types of pyroxene (ferrosilite [Fe2+2Si2O6], enstatite [Mg2Si2O6], hedenbergite [CaFe2+Si2O6], and diopside [CaMgSi2O6]), and three varieties of feldspar (alkali feldspar [KAlSi3O8], albite [NaAlSi3O8], and anorthite [CaAl2Si2O8]). The accuracy of matching Raman characteristics is exceptionally high for all olivine and pyroxene types (100%) and an impressive 86% for feldspar. The findings from this study highlight the crucial role of Raman spectroscopy in the field of silicate mineralogy and suggest significant implications for enhancing future exploration missions to Earth-like bodies. © 2024 John Wiley & Sons Ltd.
  Accession Number: 20240615494176
• Record 32 of
  
  Title:Redundant-Coded Masked Grid Pattern for Full-sky Star Identification
  
  Author(s):Liao, Jiawen(1); Wei, Xin(2); Niu, Axi(3); Zhang, Yanning(3); Kweon, Inso(4); Qi, Chun(5)
  Source: IEEE Transactions on Aerospace and Electronic Systems  Volume:   Issue:   DOI: 10.1109/TAES.2024.3374714  Published: 2024  
  Abstract:Full-sky autonomous star identification is one of the key technologies in the research of star sensors. As one of the classical pattern-based star identification methods, the Grid algorithm has shown promising performance. Na further modified it to improve robustness to position noise. However, the inherent alignment star mismatch and pattern inconsistency are still not solved. To address these problems, we propose a novel star identification method. Specifically, we design distance-guided redundant-coded patterns for different alignment stars to alleviate the problem of alignment star mismatch. Then, we create a masked grid pattern to address the inconsistency between the sensor pattern and the catalog pattern. Distances of the reference stars to their corresponding alignment stars are adopted to assist in choosing the correct alignment star, as well as reducing the number of catalog patterns that need to be evaluated. Experimental results on both synthesized and night sky images show that the proposed algorithm is quite robust to false stars, position noise, and magnitude noise. The identification accuracy of this algorithm is 98.43% with standard deviations of position noises is 2.0 pixels and 98.52% with standard deviations of magnitude noises is 0.5 Mv. Moreover, the algorithm obtains an average identification accuracy of 99.6% from night sky images. IEEE
  Accession Number: 20241215762514
• Record 33 of
  
  Title:Static spectroscopic ellipsometer based on division-of-amplitude polarization demodulation
  
  Author(s):Li, Siyuan(1,2); Deng, Zhongxun(3); Quan, Naicheng(4); Zhang, Chunmin(5)
  Source: Optics Communications  Volume: 552  Issue:   DOI: 10.1016/j.optcom.2023.130115  Published: February 1, 2024  
  Abstract:Theoretical and experimental demonstrations of a static spectroscopic ellipsometer are presented. It uses a linear polarizer for generating polarization states to interact with the sample, and three non-polarization beam splitters incorporating four achromatic quarter waveplate/linear analyzer pairs for analyzing the polarization states after the interaction. Compared to previous instruments, the most significant advantage of the described model is that it can obtain the spectral ellipsometric parameters with the same spectral resolution as the spectrometer in the system by a single snapshot. © 2023
  Accession Number: 20234515014937
• Record 34 of
  
  Title:Optical alignment technology for 1-meter accurate infrared magnetic system telescope
  
  Author(s):Fu, Xing(1); Lei, Yu(1,2); Li, Hua(1); Kewei, E.(1); Wang, Peng(1); Liu, Junpeng(1); Shen, Yuliang(3); Wang, Dongguang(3)
  Source: Journal of Astronomical Telescopes, Instruments, and Systems  Volume: 10  Issue: 1  DOI: 10.1117/1.JATIS.10.1.014004  Published: January 1, 2024  
  Abstract:Accurate infrared magnetic system (AIMS) is a ground-based solar telescope with the effective aperture of 1 m. The system has complex optical path and contains multiple aspherical mirrors. Since some mirrors are anisotropic in space, parallel light undergoes complex spatial reflection after passing through the optical pupil. It is also required that part of the optical axis coincides with the mechanical rotation axis. The system is difficult to align. This article proposes two innovative alignment methods. First, a modularized alignment method is presented. Each module is individually assembled with optical reference reserved. System integration can be completed through optical reference of each module. Second, computer-aided alignment technology is adopted to achieve perfect wavefront. By perturbing the secondary mirror (M2), the influence of M2 position on the wavefront is measured and the mathematical relationship is obtained. Based on the measured wavefront data, the least squares method is used to calculate the M2 alignment and multiple adjustments have been made to M2. The final system wavefront has reached RMS=0.12 λ@632.8 nm. Through observations of stars and sunspots, it has been demonstrated that the optical system has good wavefront quality. The observed sunspot is clear with the penumbral and umbra discernible. The proposed method has been verified and provides an effective alignment solution for complex off-axis telescope with large aperture. © 2024 Society of Photo-Optical Instrumentation Engineers (SPIE).
  Accession Number: 20241515878917
• Record 35 of
  
  Title:HQ-I2IT: Redesign the optimization scheme to improve image quality in CycleGAN-based image translation systems
  
  Author(s):Zhang, Yipeng(1,2,3); Hu, Bingliang(1,2); Huang, Yingying(1,2,3); Gao, Chi(1,2,3); Yin, Jianfu(1,2,3); Wang, Quang(1,2)
  Source: IET Image Processing  Volume: 18  Issue: 2  DOI: 10.1049/ipr2.12965  Published: February 7, 2024  
  Abstract:The image-to-image translation (I2IT) task aims to transform images from the source domain into the specified target domain. State-of-the-art CycleGAN-based translation algorithms typically use cycle consistency loss and latent regression loss to constrain translation. In this work, it is demonstrated that the model parameters constrained by the cycle consistency loss and the latent regression loss are equivalent to optimizing the medians of the data distribution and the generative distribution. In addition, there is a style bias in the translation. This bias interacts between the generator and the style encoder and visually exhibits translation errors, e.g. the style of the generated image is not equal to the style of the reference image. To address these issues, a new I2IT model termed high-quality-I2IT (HQ-I2IT) is proposed. The optimization scheme is redesigned to prevent the model from optimizing the median of the data distribution. In addition, by separating the optimization of the generator and the latent code estimator, the redesigned model avoids error interactions and gradually corrects errors during training, thereby avoiding learning the median of the generated distribution. The experimental results demonstrate that the visual quality of the images produced by HQ-I2IT is significantly improved without changing the generator structure, especially when guided by the reference images. Specifically, the Fréchet inception distance on the AFHQ and CelebA-HQ datasets are reduced from 19.8 to 10.2 and from 23.8 to 17.0, respectively. © 2023 The Authors. IET Image Processing published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology.
  Accession Number: 20234314951511
• Record 36 of
  
  Title:A Lightweight Remote Sensing Aircraft Object Detection Network Based on Improved YOLOv5n
  
  Author(s):Wang, Jiale(1,2); Bai, Zhe(1); Zhang, Ximing(1); Qiu, Yuehong(1)
  Source: Remote Sensing  Volume: 16  Issue: 5  DOI: 10.3390/rs16050857  Published: March 2024  
  Abstract:Due to the issues of remote sensing object detection algorithms based on deep learning, such as a high number of network parameters, large model size, and high computational requirements, it is challenging to deploy them on small mobile devices. This paper proposes an extremely lightweight remote sensing aircraft object detection network based on the improved YOLOv5n. This network combines Shufflenet v2 and YOLOv5n, significantly reducing the network size while ensuring high detection accuracy. It substitutes the original CIoU and convolution with EIoU and deformable convolution, optimizing for the small-scale characteristics of aircraft objects and further accelerating convergence and improving regression accuracy. Additionally, a coordinate attention (CA) mechanism is introduced at the end of the backbone to focus on orientation perception and positional information. We conducted a series of experiments, comparing our method with networks like GhostNet, PP-LCNet, MobileNetV3, and MobileNetV3s, and performed detailed ablation studies. The experimental results on the Mar20 public dataset indicate that, compared to the original YOLOv5n network, our lightweight network has only about one-fifth of its parameter count, with only a slight decrease of 2.7% in mAP@0.5. At the same time, compared with other lightweight networks of the same magnitude, our network achieves an effective balance between detection accuracy and resource consumption such as memory and computing power, providing a novel solution for the implementation and hardware deployment of lightweight remote sensing object detection networks. © 2024 by the authors.
  Accession Number: 20241115749023