Global attention retinex network for low light image enhancement

Most low-light image enhancement methods only adjust the brightness, contrast and noise reduction of low-light images, making it difficult to recover the lost information in darker areas of the image, and even cause color distortion and blurring. To solve the above problems, a global attention-based Retinex network (GARN) for low-light image enhancement is proposed in this paper. We propose a novel global attention module which computes multiple dimensional information in the channel attention module to help facilitate inference learning. Then the global attention module is embedded into different layers of the network to extract richer shallow texture features and deep semantic features. This means that the rich features are more conducive to learning the mapping relationship between low-light images to normal-light images, so that the detail recovery of dark regions is enhanced in low-light images. We also collected a low/normal light image dataset with multiple scenes, in which the images paired as training set can succeed to be applied to low-light image enhancement under different lighting conditions. Experimental results on publicly available datasets show that our method has better effectiveness and generality than the state-of-the-art methods in terms of evaluations metrics such as PSNR, SSIM, NIQE, Entropy.     

基于宇称-时间对称光子晶体结构的磁场传感器研究

提出了一种基于宇称-时间(parity-time,PT) 对称结构的磁场检测微腔模型,利用水 基Fe₃O₄磁流体介质层的磁光效应和PT对称结构的光放大效应,当磁场强度发生变化时 ,引 起结构禁带中缺陷模位置的改变。因此,可以通过检测缺陷模位置实现磁场强度的检测。本 文利用传递矩阵法(transfer matrix method,TMM)从理论上分析了结构的透射光谱,在此 基础上对结构周期、介质层厚度、PT对称单元中宏观洛伦兹振荡强度等参数进行了数值优化 。优化后的仿真结果表明,该传感器能够实现有效的磁场磁强检测,磁场强度在0—500Oe 的 范围内,传感器的灵敏度可以达到97.14 nm/RIU,优值和检测极限分 别可以达到10⁴和10－7 RIU。同时, 传感器的分辨率数值最小可以达到0.05 Oe和最 大可以达到2Oe。本文所设计的传感器结构可用于高综合性能磁场强度检测设备的设计。     

Multi-scale fusion residual encoder-decoder approach for low illumination image enhancement

The sensing light source of the line scan camera cannot be fully exposed in a low light environment due to the extremely small number of photons and high noise, which leads to a reduction in image quality. A multi-scale fusion residual encoder-decoder (FRED) was proposed to solve the problem. By directly learning the end-to-end mapping between light and dark images, FRED can enhance the image's brightness with the details and colors of the original image fully restored. A residual block (RB) was added to the network structure to increase feature diversity and speed up network training. Moreover, the addition of a dense context feature aggregation module (DCFAM) made up for the deficiency of spatial information in the deep network by aggregating the context's global multi-scale features. The experimental results show that the FRED is superior to most other algorithms in visual effect and quantitative evaluation of peak signa-to-noise ratio (PSNR) and structural similarity index measure (SSIM). For the factor that FRED can restore the brightness of images while representing the edge and color of the image effectively, a satisfactory visual quality is obtained under the enhancement of low-light.     

基于多尺度圆周频率滤波与卷积神经网络的遥感图像飞机目标检测方法研究

针对遥感图像飞机目标检测因目标尺度不一存在漏警、虚警等问题,该文基于遥感图像中飞机目标形状特征和灰度变化特点提出了一种多尺度圆周频率滤波(MSCFF)与卷积神经网络(CNN)相结合的MSCFF+CNN飞机目标自动检测算法.该算法首先采用多尺度圆周频率滤波器滤除遥感图像复杂背景,实现不同尺度飞机目标候选区域提取;然后,通过构建卷积神经网络(CNN)模型实现候选区域有效分类,最终精确确定飞机目标位置.最后,基于获取的真实遥感图像进行目标检测算法实验验证,经统计该算法的飞机目标检测率为94.38%,虚警率为3.76%,实验结果充分验证了该文算法的有效性,该算法可为机场监管、军事侦察等应用提供重要的技术支持.     

基于无人机影像的电网绝缘子自爆识别

电网因其在电能传输方面的关键性作用,在我国民生项目建设领域一直扮演着至关重要的角色。电网杆塔上的绝缘子一旦发生自爆(也称“缺陷”),绝缘子会自动剥落,输电线路就会产生安全隐患,严重时会降低输电线路的运行寿命,甚至会引发供电中断,发生大范围的停电事故,造成巨大的财产损失。目前,主流的巡检方法为人工巡检,该方法不仅耗时耗力,而且也存在一定主观出错率,已不适用于目前电路巡检的实际情况。本设计采用YOLO V5网络模型,对无人机航拍影像中绝缘子串及绝缘子自爆进行自动识别。首先通过平移、翻转、裁剪等,对航拍绝缘子影像数据集进行数据增广,并对增广后的数据集在LabelImg中进行标注,然后利用YOLO V5网络模型对绝缘子串及绝缘子自爆进行识别,最后采用PyQt5框架在PyCharm中设计了绝缘子自爆识别的系统界面,对模型进行调用,实现了绝缘子串及绝缘子自爆识别。本设计采用从网络上下载、国家电网提供、数据增广所得到的500张无人机航拍影像作为数据集,对所得数据集进行人工标注,再使用YOLO V5网络模型进行训练和测试,结果表明YOLO V5网络模型对绝缘子串具有较高的识别精度,最高识别精度为90.2%,对绝缘子自爆的最高检测精度为80.8%。这说明了YOLO V5网络模型在绝缘子串识别方面有较好的表现,但是由于训练集中绝缘子自爆的样本影像数量有限,所以该网络模型对绝缘子的自爆识别存在一定局限性,本实验能够部分代替人力实现电网绝缘子智能巡检,提高了检测效率。     

图像分类问题的几种深度学习策略

针对图像分类学习不够深入的问题,提出图像分类问题的几种深度学习策略研究。通过分析当前主流的主动深度学习图像、多标签图像和多尺度网络图像三种深度学习方法的工作原理和存在的优势与不足,探讨图像分类问题的优化学习策略。随后采用图像分类问题的几种深度学习策略实验的方式加以对比,实验结果表明,参数共享的深度学习图像分类方法不仅提高了预测速度,而且还能确保模型的准确性。     

基于超像素级卷积神经网络的多聚焦图像融合算法

该文提出了基于超像素级卷积神经网络(sp-CNN)的多聚焦图像融合算法。该方法首先对源图像进行多尺度超像素分割,将获取的超像素输入sp-CNN,并对输出的初始分类映射图进行连通域操作得到初始决策图;然后根据多幅初始决策图的异同获得不确定区域,并利用空间频率对其再分类,得到阶段决策图;最后利用形态学对阶段决策图进行后处理,并根据所得的最终决策图融合图像。该文算法直接利用超像素分割块进行图像融合,其相较以往利用重叠块的融合算法可达到降低时间复杂度的目的,同时可获得较好的融合效果。     

A path planning model based on nested regular hexagons using weighted centroid localization

In many wireless sensor network (WSN) applications, the location of a sensor node is crucial for determining where the event or situation of interest occurred. Therefore, localization is one of the critical challenges in WSNs. Mobile anchor node assisted localization (MANAL) is one of the promising solutions for the localization of statically deployed sensors. The main problem in MANAL localization is that the path planning of the mobile anchor (MA) node should be done so that the localization error in the network will be minimal and that all unknown nodes in the network are covered. This paper proposes a new path planning approach called nested hexagons curves (NHexCurves) for MANAL. NHexCurves guarantees that it will receive messages from at least three non-collinear anchors to locate all unknown nodes in the network. The proposed model has compared six different path planning schemes in the literature using weighted centroid localization (WCL). In these comparisons, first of all, localization errors of the models are compared using some statistical concepts. Second, the variation of the localization error according to parameters such as resolution (R) and the standard deviation of noise (σ) is observed. Then, with similar approaches, the standard deviation of errors, localization ratio, scalability performances, and finally, path lengths of the models are examined. The simulation results show that the NHexCurves static path planning model proposed in this study stands out compared to other models with high localization error and localization ratio performance, especially at low resolutions, due to its path design. At the same time, the lowest error values according to σ are obtained with the proposed model among all models considered.     

一种新型的表面声波阻抗传感器及其在液相体系中的应用

一种新型的表面声波阻抗传感器及其在液相体系中的应用姚守拙，陈康，刘德忠，聂利华（湖南大学化学化工系，长沙，４１００８２）关键词表面声波，阻抗传感器，液相应用表面声波传感器（ＳＡＷＳ）具有较高的灵敏度和检测精度，已广泛应用于气体微量组分分析、色谱检测和...     

Polymeric pH indicators immobilized PVA membranes for optical sensors of high basicity based on a kinetic process

Two kinds of polymeric pH indicators PPF (phenolphthalein-formaldehyde product) and CPF (o-cresolphthalein-formaldehyde product) immobilized cross-linked poly(vinyl alcohol) membranes (PPF-PVA and CPF-PVA) for optical intermittent determination of high basicity ([OH⁻] = 1-8 M) based on a kinetic process were developed. In our previous work, we had demonstrated that PPF-PVA and CPF-PVA could perform the determination of high pH values from pH 10.0 to 14.0. Here the discoloring kinetic behaviors of PPF-PVA and CPF-PVA were compared with those of free phenolphthalein, o-cresolphthalein and thymolphthalein. Experimental results and theoretical analysis indicated that the response behaviors of the optodes’ membranes in concentrated NaOH solutions were diffusion-independent and still complied with the pseudo-first-order kinetics. In addition, two data analysis methods for determination were presented. One was directly based on the reduced absorbance; the other was based on the discoloring kinetic constant. It was found that the latter could perform a rapid (60 s) and reliable (relative standard deviation: 2.6%) determination for high basicity. These kinds of optodes can be used repeatedly when they are immersed in low-pH solutions to regain the protonated form after each determination.