基于包装材料数字图像的RGB颜色空间色差评价方法研究

目的解决包装材料复杂颜色区域色差评价,目测检验方法存在判定标准复杂、结果一致性差等问题。方法提出一种基于包装材料高清数字图像的RGB颜色空间色差评价方法。通过搭建高清数字图像采集装置,获取包装材料标准样(上限、中限、下限)的全幅数字图像。通过采用RGB颜色空间降维的方法,对标准样测试区域的颜色空间进行离散边界拟合,构建标准样品测试区域的三限样RGB色差模型,与测试样对应测试区域的颜色空间进行对比;根据颜色空间模型的一致性判定测试样色差是否合格。结果对比多种离散边界拟合模型,采用滚球法边缘拟合模型的包含率和多出率分别达到100%、78.9%,模型拟合效果最佳。通过对包装材料色差缺陷样验证测试,采用上述色差评价方法准确率达到100%。结论本文提出的滚球法颜色空间边缘拟合模型和色差评价方法,可实现对包装材料复杂颜色区域的色差合格性判定,有利于提高企业生产的智能化检测水平。

基于神经辐射场的RGB图像点云重建多肉植物及尺寸测量研究

以多肉植物盆栽为研究对象,使用手持式RGB相机采集11个多肉植物盆栽的视频数据,通过将视频转换为图像帧、选取优质清晰图像帧、计算相机位姿得到含丰富信息的RGB图像数据。提出一种改进神经辐射场的多肉植物三维重建方法,根据实际场景提出新的射线采样策略,同时引入改进的图像修复模块与隐式模型重建点云方法,并根据点云重建结果提取多肉植株的叶片数、株高、冠围、凸包体积、叶长、叶宽和叶色共7个表型参数。最后选取具有代表性、易测量的叶片数、株高、冠围、叶长和叶宽5个表型参数进行精度评估与误差原因分析,平均绝对百分比误差(MAPE)分别为2.32%、3.95%、4.95%、5.59%和9.55%,均方根误差(RMSE)分别为0.86片和1.95、17.54、1.87、1.27 mm,决定系数(R^(2))分别为0.99、0.99、0.86、0.91和0.89。精度评估结果表明,所提取的表型参数能够准确、高效地反映多肉植株生长状态,充分发挥RGB图像新视角合成技术、图像处理技术与三维点云重建技术的优势,实现多肉植株盆栽的表型参数高精度、非破坏性提取,能够为多肉植物的种植和养育以及为非固定、多视角的RGB数据获取研究提供重要的技术支持。

基于激光雷达与RGB相机融合的玉米作物行检测算法研究

针对单一传感器在面对复杂田间环境适应性差的问题,本文提出了一种基于固态激光雷达(LiDAR)与RGB相机融合的玉米作物行检测方法。首先,研究了固态激光雷达和RGB相机联合标定方法,同步获取玉米作物行图像和点云数据并进行数据预处理。然后,将预处理后的图像数据和点云数据融合,实现点云“着色”,基于点云“着色”提出聚类感兴趣密度区域算法。利用“着色”点云完成聚类,并结合作物种植农艺标准(行距),分别验证点云信息和颜色信息的可用性,能够选择最优信息完成作物行感兴趣区域聚类。最后,通过划分点云水平条带的方式确定目标点云的特征点聚类区域,取作物行特征点,并利用最小二乘法拟合作物行检测线。仅需调整行距参数,算法可实现全生命周期的作物行检测,利用正常工况下玉米苗期、前期、中期和后期数据开展算法验证,作物行中心线平均误差不大于1.781°,准确率不小于92.69%,平均耗时不超过102.7 ms。此外,为验证算法鲁棒性,开展了复杂农田背景环境,如高杂草背景、断行、苗期杂草高度与玉米高度相近以及玉米完全封行4种工况作物行检测,算法平均误差不大于1.935°,准确率不小于91.94%,平均耗时不超过108.3 ms。通过讨论阐述了基于点云“着色”开展作物行中心线提取的优越性,本文算法可为作物行中心线可靠检测提供参考。

基于RGB模型的草莓叶片光合作用指标估测

为了研究基于图像红(R)、绿(G)、蓝(B)颜色参数和叶片SPAD值预测光合作用指标的可行性,以草莓叶片为试验材料,构建多元线性回归模型和反向传播(BP)神经网络模型,对叶片蒸腾速率、气孔导度、净光合速率、胞间CO_(2)浓度进行估测,并对其精度进行评价和验证。结果表明,基于BP神经网络模型,使用图像RGB颜色参数和SPAD值对叶片蒸腾速率进行预测的效果较好,其次是气孔导度。BP神经网络模型的估测精度高于多元线性回归模型,蒸腾速率、气孔导度、净光合速率和胞间CO_(2)浓度的模型预测准确率分别达到91.5%、83.3%、74.4%和71.5%。BP神经网络的蒸腾速率模型、气孔导度模型的决定系数(R2)分别为0.9222、0.8423,均方根误差(RMSE)分别为0.0002、0.0259,平均绝对误差(MAE)分别为0.0001、0.0006。由结果可知,通过数码相机采集图像,并构建RGB模型,可简易快速估测草莓叶片蒸腾速率、气孔导度,能用于生产中草莓光合指标的估测。

用于RGBT跟踪的孪生混合信息融合算法

可见光与热红外跟踪(又称RGBT(RGB-Thermal)跟踪)的核心是有效地利用不同模态的信息,针对决策级融合中单分支产生低质结果影响算法判定目标的问题,提出一个用于RGBT跟踪的孪生混合信息融合算法SiamMIF。首先,使用孪生主干网络(SBN)进行多模态特征提取;其次,从信噪比的角度分析低质图像对双分支并行决策产生的影响,进而设计了一个信噪比驱动的信息交互模块(IIM)对低信噪比特征进行信息互补;再次,利用双流无锚跟踪头(ADH)对补偿后的特征进行分类回归;最后,采用自适应轻量决策模块(ALDM)对跟踪结果进行融合,并快速判定目标位置。在4个RGBT基准数据集GTOT、RGBT234、VOT-RGBT2019和LasHeR上的实验结果表明,所提算法在LasHeR数据集上的成功率和精确度分别为0.396和0.518,相较于APFNet(Attribute-based Progressive Fusion Network)提升9.4%和3.6%,在其他3个数据集上也能取得较好结果,且在GPU上的帧率能达到40 frame/s。

OLED屏下RGB图像优化算法

全面屏的流行对智能手机前置摄像头提出了屏下高质量拍摄的要求。目前用于屏下拍摄方案的有机发光二极管(Organic Light-Emitting Diode,OLED)透明屏存在光衍射、折射等现象,导致拍摄的RGB图像易产生模糊和细节丢失等问题。针对上述问题,提出了一种OLED屏下RGB图像优化算法。针对目前屏下RGB图像优化数据集较少的问题,设计实现了一种基于智能手机的OLED透明屏屏下图像数据采集装置,采集并制作了由10000多组典型场景构成的屏下图像数据集。其次,提出了一种基于生成对抗网络(Generative Adversarial Nets,GAN)的屏下RGB图像优化算法,其中生成器采用残差网络学习屏下图像细节信息,所设计的感知损失函数是颜色损失、对抗损失和内容损失三者的结合。实验结果表明,基于主观视觉和峰值信噪比(Peak Signal-to-Noise Ratio,PSNR)、结构相似性(Structural Similarity,SSIM)等定量评价指标,本文算法在自建数据集上的图像优化效果优于当前的DPED等方法的效果。

基于RGB模型的汽车指针仪表示数的识别

本文对汽车指针仪表示数的识别,提出了一种基于RGB彩色空间图像处理的识别方案。首先,根据仪表盘图像的R、G、B分量的直方图,利用该分量的欧几里德距离法分割仪表盘的圆心和指针;然后,提取圆心的骨架和圆心的坐标,建立新的坐标系;最后,将指针的质心和仪表盘的圆心连成一条直线,通过直线位置识别指针读数。实验证明:该方法可用于指针式汽车仪表的自动化测试。

基于多模态双向信息增强的RGBT跟踪网络

可见光-热红外(RGBT)目标跟踪旨在挖掘可见光和热红外数据的互补优势,实现鲁棒的目标跟踪。目前主流方法通常引入模态权重来实现多模态信息融合,但简单地为各个模态分配权重无法充分挖掘可见光和热红外模态的互补优势。基于此,提出了一种多模态双向信息增强的RGBT跟踪网络(MBIENet)。设计了一种特征聚合模块,用于聚合模态共享特征和模态特定特征以建模目标外观信息;提出了一种新的多模态双向调制融合模块,可有效融合模态互补信息,减少冗余特征和无用特征对跟踪器的影响;提出了一个轻量化的通道空间注意力模块,可自适应调整不同环境下不同模态的贡献。在GTOT、RGBT234和LasHeR数据集上的实验结果表明:所提跟踪算法的准确率和成功率优于当前主流的跟踪算法。

液晶显示器sRGB特性文件精度分析研究

当把一幅彩色图像的RGB值分别输入到不同的颜色设备上时,所输出彩色图像的Lab值将有很大的不同,使彩色图像失真。由于不同设备系统表征颜色的特性不同,所以同一组RGB数据在不同显示器上所呈现出的颜色不同,不同显示器所能表现的颜色范围也不同。

基于多尺度模态融合的RGB-T目标跟踪网络

可见光-热红外(RGB-T)目标跟踪因受光照条件限制较小受到关注。针对不同尺度特征的分辨率与语义信息存在差异、可见光与热红外两种模态信息不一致的特点,以及现有网络在多模态融合策略上的不足,提出一种RGB-T目标跟踪网络。网络采用孪生结构,首先将主干特征提取网络输出的模板图像特征与搜索图像特征从单尺度拓展到多尺度,并对可见光与热红外模态在不同尺度上分别进行模态融合,然后将得到的融合特征通过注意力机制增强特征表示,最后通过区域建议网络得到预测结果。在GTOT与RGBT-234两个公开RGB-T数据集上的实验结果表明,该网络跟踪精度和成功率较高,可以应对复杂的跟踪场景,相比于其他网络具有更高的跟踪性能。

基于多模态RGB-T的显著性目标检测算法

针对RGB(Red Green Blue)模态与热度模态信息表征形式不一致,特征信息无法有效挖掘、融合问题,提出了一种新的联合注意力强化网络-FCNet(Feature Sharpening and Cross-modal Feature Fusion Net)。首先,通过双维度注意力机制提升图像特征映射能力;然后,利用跨模态特征融合机制捕获目标区域;最后,利用逐层解码结构消除背景干扰,优化检测目标。实验结果表明,该优化改进算法运算参数更少、运算时间更短,且模型整体检测性能均优于现有多模态检测模型性能。

结合视锥变换和RGB体素图的半监督三维目标检测

基于LiDAR、可见光等多模态传感器的高精度三维目标检测是自动驾驶领域的关键技术。为了提高目标检测的精度和方位感知能力,降低模型对于标注数据的依赖,结合视锥变换方法优化了三维点云方向特征提取策略,提出了一种基于视锥变换和半监督学习架构的三维目标检测技术。具体而言,基于通道注意力模块优化视锥体对远距离目标的感知能力,提出了RGB体素模块提升遮挡目标的识别精度。首先通过深度网络从RGB图像中提取纹理信息,将其与激光雷达的距离信息融合,以保持三维空间特征的完整性。其次,通过特征融合模块提取体素空间特征的权重。最后,采用自适应伪标签方法降低对标注样本的依赖,并基于群体投票方法进一步降低误报率。实验结果表明,该方法在KITTI数据集上取得了令人满意的成果,行人和车辆目标检测的准确率分别达到了56.30%和75.88%。该研究为未来在复杂的场景中实现高效的三维目标检测提供了思路,并为进一步优化自动驾驶的多模态数据融合技术奠定了基础。

基于RGB图像的设施番茄冠层覆盖度估算方法研究

以番茄为试材,采用设施实验箱育苗的方式,通过设施环境下桁架搭载可见光(RGB)相机系统获取图像数据,研究了不同植被指数算法分割番茄冠层图像的精度,并实现设施番茄冠层覆盖度提取方法的评估,以期为其他设施作物的冠层覆盖度估算提供方法指导。结果表明:EXG算法、EXGR算法和CIVE算法均可用于估算设施番茄的冠层覆盖度,与真值之间的均方根误差(RMSE)分别为0.049、0.078、0.088,决定系数(R^(2))分别为0.911、0.845、0.841,不同植被指数分割算法估算的设施番茄冠层覆盖度之间存在差异,与真值图像相比,EXGR算法在定植10 d的图像分割时,分割精度较低,冠层覆盖度估算值偏大,CIVE算法在定植66 d的图像分割时,由于分割过度,导致冠层覆盖度估算值偏小,而EXG算法在各时期的分割精度较高,冠层覆盖度的估算值与真值之间的吻合度最好。这表明EXG算法能够更有效的实现设施番茄的植土分割且估算精度更高。

基于RGB图像的三维人手姿态估计技术综述

鉴于RGB相机在虚拟现实头盔等移动计算设备中的普遍性,基于RGB图像的三维人手姿态估计技术具有广阔的应用前景和研究价值,近年来已成为计算机视觉领域的一个研究热点.得益于深度学习技术的快速发展,与之相关的三维人手姿态估计算法层出不穷.文中回顾和总结了三维人手姿态估计技术.首先简述了三维人手姿态估计的相关工作,指出了其当前面临的挑战;然后梳理了基于RGB图像的三维人手姿态估计算法,对现有的基于参数模型方法和非参数模型方法进行了讨论,分析了每类算法包含的技术方法以及优缺点;之后总结了相关的三维手数据集与评价标准,并比较了每类算法在常用数据集上的表现;最后探讨了该技术的发展前景.

Mechanism of internal thermal runaway propagation in blade batteries

Blade batteries are extensively used in electric vehicles,but unavoidable thermal runaway is an inherent threat to their safe use.This study experimentally investigated the mechanism underlying thermal runaway propagation within a blade battery by using a nail to trigger thermal runaway and thermocouples to track its propagation inside a cell.The results showed that the internal thermal runaway could propagate for up to 272 s,which is comparable to that of a traditional battery module.The velocity of the thermal runaway propagation fluctuated between 1 and 8 mm s^(-1),depending on both the electrolyte content and high-temperature gas diffusion.In the early stages of thermal runaway,the electrolyte participated in the reaction,which intensified the thermal runaway and accelerated its propagation.As the battery temperature increased,the electrolyte evaporated,which attenuated the acceleration effect.Gas diffusion affected thermal runaway propagation through both heat transfer and mass transfer.The experimental results indicated that gas diffusion accelerated the velocity of thermal runaway propagation by 36.84%.We used a 1D mathematical model and confirmed that convective heat transfer induced by gas diffusion increased the velocity of thermal runaway propagation by 5.46%-17.06%.Finally,the temperature rate curve was analyzed,and a three-stage mechanism for internal thermal runaway propagation was proposed.In Stage I,convective heat transfer from electrolyte evaporation locally increased the temperature to 100℃.In Stage II,solid heat transfer locally increases the temperature to trigger thermal runaway.In StageⅢ,thermal runaway sharply increases the local temperature.The proposed mechanism sheds light on the internal thermal runaway propagation of blade batteries and offers valuable insights into safety considerations for future design.

Inherent thermal-responsive strategies for safe lithium batteries

Safe batteries are the basis for next-generation application scenarios such as portable energy storage devices and electric vehicles,which are crucial to achieving carbon neutralization.Electrolytes,separators,and electrodes as main components of lithium batteries strongly affect the occurrence of safety accidents.Responsive materials,which can respond to external stimuli or environmental change,have triggered extensive attentions recently,holding great promise in facilitating safe and smart batteries.This review thoroughly discusses recent advances regarding the construction of high-safety lithium batteries based on internal thermal-responsive strategies,together with the corresponding changes in electrochemical performance under external stimulus.Furthermore,the existing challenges and outlook for the design of safe batteries are presented,creating valuable insights and proposing directions for the practical implementation of safe lithium batteries.

Converting TREx-RGB green-channel data to 557.7 nm auroral intensity:Methodology and initial results

The recently deployed Transition Region Explorer(TREx)-RGB(red-green-blue)all-sky imager(ASI)is designed to capture“true color”images of the aurora and airglow.Because the 557.7 nm green line is usually the brightest emission line in visible auroras,the green channel of a TREx-RGB camera is usually dominated by the 557.7 nm emission.Under this rationale,the TREx mission does not include a specific 557.7 nm imager and is designed to use the RGB green-channel data as a proxy for the 557.7 nm aurora.In this study,we present an initial effort to establish the conversion ratio or formula linking the RGB green-channel data to the absolute intensity of 557.7 nm auroras,which is crucial for quantitative uses of the RGB data.We illustrate two approaches:(1)through a comparison with the collocated measurement of green-line auroras from the TREx spectrograph,and(2)through a comparison with the modeled green-line intensity according to realistic electron precipitation flux measurements from low-Earth-orbit satellites,with the aid of an auroral transport model.We demonstrate the procedures and provide initial results for the TREx-RGB ASIs at the Rabbit Lake and Lucky Lake stations.The RGB response is found to be nonlinear.Empirical conversion ratios or formulas between RGB green-channel data and the green-line auroral intensity are given and can be applied immediately by TREx-RGB data users.The methodology established in this study will also be applicable to the upcoming SMILE ASI mission,which will adopt a similar RGB camera system in its deployment.

Effect of safety valve types on the gas venting behavior and thermal runaway hazard severity of large-format prismatic lithium iron phosphate batteries

The safety valve is an important component to ensure the safe operation of lithium-ion batteries(LIBs).However,the effect of safety valve type on the thermal runaway(TR)and gas venting behavior of LIBs,as well as the TR hazard severity of LIBs,are not known.In this paper,the TR and gas venting behavior of three 100 A h lithium iron phosphate(LFP)batteries with different safety valves are investigated under overheating.Compared to previous studies,the main contribution of this work is in studying and evaluating the effect of gas venting behavior and TR hazard severity of LFP batteries with three safety valve types.Two significant results are obtained:(Ⅰ)the safety valve type dominates over gas venting pressure of battery during safety venting,the maximum gas venting pressure of LFP batteries with a round safety valve is 3320 Pa,which is one order of magnitude higher than other batteries with oval or cavity safety valve;(Ⅱ)the LFP battery with oval safety valve has the lowest TR hazard as shown by the TR hazard assessment model based on gray-fuzzy analytic hierarchy process.This study reveals the effect of safety valve type on TR and gas venting,providing a clear direction for the safety valve design.

Personal Thermal Management by Radiative Cooling and Heating

Maintaining thermal comfort within the human body is crucial for optimal health and overall well-being.By merely broadening the setpoint of indoor temperatures,we could significantly slash energy usage in building heating,ventilation,and air-conditioning systems.In recent years,there has been a surge in advancements in personal thermal management(PTM),aiming to regulate heat and moisture transfer within our immediate surroundings,clothing,and skin.The advent of PTM is driven by the rapid development in nano/micro-materials and energy science and engineering.An emerging research area in PTM is personal radiative thermal management(PRTM),which demonstrates immense potential with its high radiative heat transfer efficiency and ease of regulation.However,it is less taken into account in traditional textiles,and there currently lies a gap in our knowledge and understanding of PRTM.In this review,we aim to present a thorough analysis of advanced textile materials and technologies for PRTM.Specifically,we will introduce and discuss the underlying radiation heat transfer mechanisms,fabrication methods of textiles,and various indoor/outdoor applications in light of their different regulation functionalities,including radiative cooling,radiative heating,and dual-mode thermoregulation.Furthermore,we will shine a light on the current hurdles,propose potential strategies,and delve into future technology trends for PRTM with an emphasis on functionalities and applications.

Heat transfer enhanced inorganic phase change material compositing carbon nanotubes for battery thermal management and thermal runaway propagation mitigation

Developing technologies that can be applied simultaneously in battery thermal management(BTM)and thermal runaway(TR)mitigation is significant to improving the safety of lithium-ion battery systems.Inorganic phase change material(PCM)with nonflammability has the potential to achieve this dual function.This study proposed an encapsulated inorganic phase change material(EPCM)with a heat transfer enhancement for battery systems,where Na_(2)HPO_(4)·12H_(2)O was used as the core PCM encapsulated by silica and the additive of carbon nanotube(CNT)was applied to enhance the thermal conductivity.The microstructure and thermal properties of the EPCM/CNT were analyzed by a series of characterization tests.Two different incorporating methods of CNT were compared and the proper CNT adding amount was also studied.After preparation,the battery thermal management performance and TR propagation mitigation effects of EPCM/CNT were further investigated on the battery modules.The experimental results of thermal management tests showed that EPCM/CNT not only slowed down the temperature rising of the module but also improved the temperature uniformity during normal operation.The peak battery temperature decreased from 76℃to 61.2℃at 2 C discharge rate and the temperature difference was controlled below 3℃.Moreover,the results of TR propagation tests demonstrated that nonflammable EPCM/CNT with good heat absorption could work as a TR barrier,which exhibited effective mitigation on TR and TR propagation.The trigger time of three cells was successfully delayed by 129,474 and 551 s,respectively and the propagation intervals were greatly extended as well.