基于雨雾分离处理和多尺度网络的图像去雨方法

雨带来的雨条纹和雨雾会降低户外拍摄图像的质量,为了去除雨雾对图像的影响,提出了一种基于雨雾分离处理和多尺度卷积神经网络的图像去雨方法。首先利用导向滤波将雨线和图像细节信息提取到高频层,雨雾和背景信息则分离到低频层;然后构建多尺度卷积神经网络来去除高频层中的雨线,网络中融入多个稠密连接模块以提升特征提取的准确性;其次构建多层特征融合的轻量级去雾网络来去除低频层中的雨雾,采用参数一体化结构避免了估计多个大气散射模型参数导致的次优解;最后再结合处理后的高低频结果还原出清晰图像。在多个合成的雨雾数据集以及真实自然场景图像上进行测试,定性和定量结果表明,提出的方法在去除雨雾影响的同时较好地保留了色彩信息,和近年的算法相比,图像结构相似性约提升了0.02~0.08,图像峰值信噪比约提升了0.2~3.5 dB。     

基于集成经验模态分解和最小二乘双支持向量回归机的风速预测算法研究

为有效解决风电大规模并网过程中面临的并网难和弃风等问题,实现可再生能源大规模平滑并网并保证大电网的安全稳定运行,采用集成经验模态分解(ensemble empirical mode decomposition,EEMD)和最小二乘双支持向量回归机(least square twin support vector regression,LSTSVR)算法进行风电场风速预测。分别介绍了LSTSVR、EEMD及自适应变异粒子群算法原理。给出基于EEMD和LSTSVR的风速预测流程,以安徽女儿岭风电场测风声雷达30、70 m处风速采样数据为例,开展基于EEMD和LSTSVR的风速预测算法验证,预测结果误差分析表明:基于EEMD+LSTSVR+自适应变异粒子群算法可以实现风电场风速的高精度预测。     

Semantic guided knowledge graph for large-scale zero-shot learning

Zero-shot learning has received growing attention, which aims to improve generalization to unseen concepts. The key challenge in zero-shot tasks is to precisely model the relationship between seen and unseen classes. Most existing zero-shot learning methods capture inter-class relationships via a shared embedding space, leading to inadequate use of relationships and poor performance. Recently, knowledge graph-based methods have emerged as a new trend of zero-shot learning. These methods use a knowledge graph to accurately model the inter-class relationships. However, the currently dominant method for zero-shot learning directly extracts the fixed connection from off-the-shelf WordNet, which will inherit the inherent noise in WordNet. In this paper, we propose a novel method that adopts class-level semantic information as a guidance to construct a new semantic guided knowledge graph (SG-KG), which can correct the errors in the existing knowledge graph and accurately model the inter-class relationships. Specifically, our method includes two main steps: noise suppression and semantic enhancement. Noise suppression is used to eliminate noise edges in the knowledge graph, and semantic enhancement is used to connect two classes with strong relations. To promote high efficient information propagation among classes, we develop a novel multi-granularity fusion network (MGFN) that integrates discriminative information from multiple GCN branches. Extensive experiments on the large-scale ImageNet-21K dataset and AWA2 dataset demonstrate that our method consistently surpasses existing methods and achieves a new state-of-the-art result.     

结合隶属度与像素交替引导滤波的鲁棒模糊聚类算法

模糊局部信息C-均值（FLICM）聚类算法是目前应用较广泛的图像分割算法,然而仅适用于处理低噪声图像。FLICM算法与像素引导隶属度滤波的结合在一定程度上提高了噪声抑制能力,但仍无法满足强噪声图像的分割需求。联合引导滤波与基于核度量的加权模糊局部信息C-均值（KWFLICM）聚类算法,提出一种隶属度与像素值交替引导的核模糊聚类算法。将像素引导隶属度滤波模块和隶属度引导像素滤波模块引入KWFLICM算法,构造一种引导滤波约束的多目标核模糊聚类优化模型,采用最小二乘法对该模型进行迭代求解。在迭代过程中,通过像素引导隶属度滤波和隶属度引导像素滤波,分别修正输入图像的隶属度和像素值,进一步提高核模糊聚类算法对含噪图像的鲁棒性。实验结果表明,与同类核模糊聚类算法相比,该算法在莱斯噪声干扰下的误分率、精确度、峰值信噪比、Jaccard相似系数等评价指标上表现突出,具有更好的分割性能和更强的鲁棒性。     

Photoacoustic‐Guided Surgery with Indocyanine Green‐Coated Superparamagnetic Iron Oxide Nanoparticle Clusters

A common cause of local tumor recurrence in brain tumor surgery results from incomplete surgical resection. Adjunctive technologies meant to facilitate gross total resection have had limited efficacy to date. Contrast agents used to delineate tumors preoperatively cannot be easily or accurately used in the real‐time operative setting. Although multimodal imaging contrast agents are developed to help the surgeon discern tumor from normal tissue in the operating room, these contrast agents are not readily translatable. This study has developed a novel contrast agent comprised solely of two Food and Drug Administration approved components, indocyanine green (ICG) and superparamagnetic iron oxide (SPIO) nanoparticles—with no additional amphiphiles or carrier materials, to enable preoperative detection by magnetic resonance (MR) imaging and intraoperative photoacoustic (PA) imaging. The encapsulation efficiency of both ICG and SPIO within the formulated clusters is ≈100%, and the total ICG payload is 20–30% of the total weight (ICG + SPIO). The ICG–SPIO clusters are stable in physiologic conditions; can be taken up within tumors by enhanced permeability and retention; and are detectable by MR. In a preclinical surgical resection model in mice, following injection of ICG–SPIO clusters, animals undergoing PA‐guided surgery demonstrate increased progression‐free survival compared to animals undergoing microscopic surgery.     

超声导波频散曲线反演皮质骨参数的研究

皮质骨参数的变化可以反映骨质的健康状况,因此文章提出了一种频散能量匹配算法来反演皮质骨参数。首先,通过频散特性方程计算出预定义参数下的频散曲线数据库;其次,利用时域有限差分算法建立皮质骨声场模型,将仿真的时域信号通过功率谱估计得到频散能量信息,并与数据库进行匹配,通过分析匹配结果的能量从而得到皮质骨的厚度、纵波速度和横波速度。仿真结果显示：与理论值相比,厚度的平均相对误差为3.8%,纵波速度与横波速度的平均相对误差分别为0.6%、0.9%。对三组离体牛胫骨皮质骨进行反演,离体实验结果显示：与真实值相比,厚度相对误差为4.9%,且实验频散曲线与反演得到的理论频散曲线吻合。因此文中所提出的反演算法可以有效获取皮质骨的厚度,横波速度和纵波速度,从而为评价骨质的健康状况提供可靠的依据。     

基于分组策略的多目标三维装箱算法

目的 针对现有三维装箱算法优化目标单一、优化效率低的问题,提出适用于求解大规模货物装载问题的多目标装箱算法,以提高装箱规划效率,确保货物运输安全。方法 考虑5种现实约束条件,以体积利用率和装载垛型重心偏移量为优化目标,建立多目标货物装载优化模型。采用拟人式装箱对货物进行预分组,减小决策空间,然后结合分组信息与装箱算法生成初始解;引入数据驱动的装箱交叉算子提高算法收敛性;设计多策略变异算子提高算法结果的多样性。结果 以公共数据集和真实航空货物数据作为实验数据进行实验。实验结果表明,在满足多种约束条件下,集装箱装载强异构货物平均体积利用率达到92.0%,重心位置空间偏移从20 cm减少到7.5 cm,并且算法运行时间减少了73.5%。结论 本文所提算法应用于求解大规模多目标三维装箱问题,提高了装箱质量和效率,可为三维装箱算法的工程应用提供参考。     

基于突变微粒群算法的图像融合技术

文中利用突变微粒群算法进行图像融合;将图像融合问题转化为优化问题,通过使目标函数(平均梯度)最大以求得好的融合效果;并以离散平稳小波变换代替传统的离散小波变换,以克服离散小波变换缺乏平移不变性的缺点;先对源图像进行离散平稳小波分解,细节部分利用能量系数矩阵进行融合;对于近似部分,利用突变微粒群算法求出最优权值对近似部分进行加权融合;实验结果表明,该方法的融合效果优于传统的融合算法。