素描图像指导的换装行人重识别

目的 传统行人重识别方法提取到的特征中包含大量的衣物信息,在换装行人重识别问题中,依靠衣物相关的信息难以准确判别行人身份,使模型性能显著下降;虽然一些方法从轮廓图像中提取行人的体型信息以增强行人特征,但轮廓图像的质量参差不齐,鲁棒性差。针对这些问题,本文提出一种素描图像指导的换装行人重识别方法。方法 首先,本文认为相对于轮廓图像,素描图像能够提供更鲁棒、更精准的行人体型信息,因此本文使用素描图像提取行人的体型信息,并将其融入表观特征以获取完备的行人特征。然后,提出一个基于素描图像的衣物无关权重指导模块,进一步使用素描图像中的衣物位置信息指导表观特征的提取过程,从而减少表观特征中的衣物信息,增强表观特征的判别力。结果 在LTCC(long-term cloth changing)和PRCC(person re-identification under moderate clothing change)两个常用换装行人数据集上,本文方法与最先进的方法进行了对比。相较于先进方法,在LTCC和PRCC数据集上,本文方法在Rank-1性能指标上分别提高了6.5%和3.9%。实验结果表明,素描图像...     

换装行人重识别研究进展

行人重识别旨在建立目标行人在多个无交叉覆盖监控区域间的身份联系,在智慧城市、司法侦查和监控安全等领域具有重要应用价值。传统行人重识别方法针对短时间跨度场景,依赖行人外观特征的稳定不变性,旨在克服光照差异、视角变化和物体遮挡等挑战。与之不同,换装行人重识别针对长时间跨度场景,除受限于上述挑战还面临换装带来的外观变化问题,是近几年的一个研究难点和热点。围绕换装行人重识别,本文从数据集和解决方法两个方面综述国内外研究进展,探讨面临的挑战和难点。首先,梳理并比较了当前针对换装行人重识别的数据集,从采集方式、行人及样本数量等方面分析其挑战性和面临的局限性。然后,在简单回顾换装行人重识别发展历史的基础上,将其归纳为基于非视觉传感器的方法和基于视觉相机的方法两类。针对基于非视觉传感器的方法,介绍了深度传感器、射频信号等在换装行人重识别中的应用。针对基于视觉相机的方法,详细阐述了基于显式特征设计与提取的方法、基于特征解耦的方法和基于隐式数据驱动自适应学习的方法。在此基础上,探讨了当前换装行人重识别面临的问题并展望未来的发展趋势,旨在为相关研究提供参考。     

基于相机感知距离矩阵的无监督行人重识别

在跨场景、跨设备的行人重识别中虽然增加了可利用的行人数据,但由于行人姿态不同、部分遮挡现象,难以避免引入样本噪声,在聚类过程中易生成错误的伪标签,造成标签噪声,影响模型的优化。为减弱噪声影响,应用相机感知的距离矩阵对抗相机偏移引起的样本噪声问题,利用对噪声鲁棒的动态对称对比损失减少标签噪声,提出基于相机感知距离矩阵的无监督行人重识别算法。在聚类前通过更改度量行人特征相似度的距离矩阵,利用相机感知距离矩阵来增强类内距离度量准确性,减少由于拍摄视角不同对聚类效果造成的负面影响。同时,结合噪声标签学习方法,进行损失设计,提出动态对称对比损失函数,联合损失训练,不断精炼伪标签。在DukeMTMC-reID和Market-1501两个数据集上进行实验,验证了提出方法的有效性。     

红外与可见光图像渐进融合深度网络

目的 红外与可见光图像融合的目标是获得具有完整场景表达能力的高质量融合图像。由于深度特征具有良好的泛化性、鲁棒性和发展潜力,很多基于深度学习的融合方法被提出,在深度特征空间进行图像融合,并取得了良好的效果。此外,受传统基于多尺度分解的融合方法的启发,不同尺度的特征有利于保留源图像的更多信息。基于此,提出了一种新颖的渐进式红外与可见光图像融合框架(progressive fusion, ProFuse)。方法 该框架以U-Net为骨干提取多尺度特征,然后逐渐融合多尺度特征,既对包含全局信息的高层特征和包含更多细节的低层特征进行融合,也在原始尺寸特征(保持更多细节)和其他更小尺寸特征(保持语义信息)上进行融合,最终逐层重建融合图像。结果 实验在TNO(Toegepast Natuurwetenschappelijk Onderzoek)和INO(Institut National D’optique)数据集上与其他6种方法进行比较,在选择的6项客观指标上,本文方法在互信息(mutual Information, MI)上相比FusionGAN(generative adversarial ...     

以双目图像互为监督的相容解迭代优选算法

在计算机视觉领域的双目立体匹配方向, 基于神经网络的深度学习算法需要场景数据集进行训练, 泛化能力差. 针对这两个问题, 根据神经网络能够模拟函数的特点, 提出一种无需在数据集上训练, 以双目图像互为监督的深度场景相容解迭代优选算法. 该算法使用场景位置猜测网络模拟关于当前双目图像的深度场景相容位置空间, 用与该网络匹配的互监督损失函数通过梯度下降法指导该网络在输入双目图像上迭代学习, 搜索深度场景相容位置空间中的可行解, 整个算法过程无需在数据集上训练. 与CREStereo、PCW-Net、CFNet等算法在Middlebury标准数据集图像上的对比实验表明, 该算法在非遮挡区域的平均误匹配率为2.52%, 在所有区域的平均误匹配率为7.26%, 比对比实验中的其他算法有更低的平均误匹配率.     

基于多对多生成对抗网络的非对称跨域迁移行人再识别

无监督跨域迁移学习是行人再识别中一个非常重要的任务. 给定一个有标注的源域和一个没有标注的目标域, 无监督跨域迁移的关键点在于尽可能地把源域的知识迁移到目标域. 然而, 目前的跨域迁移方法忽略了域内各视角分布的差异性, 导致迁移效果不好. 针对这个缺陷, 本文提出了一个基于多视角的非对称跨域迁移学习的新问题. 为了实现这种非对称跨域迁移, 提出了一种基于多对多生成对抗网络(Many-to-many generative adversarial network, M2M-GAN)的迁移方法. 该方法嵌入了指定的源域视角标记和目标域视角标记作为引导信息, 并增加了视角分类器用于鉴别不同的视角分布, 从而使模型能自动针对不同的源域视角和目标域视角组合采取不同的迁移方式. 在行人再识别基准数据集Market1501、DukeMTMC-reID和MSMT17上, 实验验证了本文的方法能有效提升迁移效果, 达到更高的无监督跨域行人再识别准确率.     

Visible–infrared person re-identification based on key-point feature extraction and optimization

Feature extraction for visible–infrared person re-identification (VI-ReID) is challenging because of the cross-modality discrepancy in the images taken by different spectral cameras. Most of the existing VI-ReID methods often ignore the potential relationship between features. In this paper, we intend to transform low-order person features into high-order graph features, and make full use of the hidden information between person features. Therefore, we propose a multi-hop attention graph convolution network (MAGC) to extract robust person joint feature information using residual attention mechanism while reducing the impact of environmental noise. The transfer of higher order graph features within MAGC enables the network to learn the hidden relationship between features. We also introduce the self-attention semantic perception layer (SSPL) which can adaptively select more discriminant features to further promote the transmission of useful information. The experiments on VI-ReID datasets demonstrate its effectiveness.     

Fine-grained-based multi-feature fusion for occluded person re-identification

Many previous occluded person re-identification(re-ID) methods try to use additional clues (pose estimation or semantic parsing models) to focus on non-occluded regions. However, these methods extremely rely on the performance of additional clues and often capture pedestrian features by designing complex modules. In this work, we propose a simple Fine-Grained Multi-Feature Fusion Network (FGMFN) to extract discriminative features, which is a dual-branch structure consisting of global feature branch and partial feature branch. Firstly, we utilize a chunking strategy to extract multi-granularity features to make the pedestrian information contained in it more comprehensive. Secondly, a spatial transformer network is introduced to localize the pedestrian’s upper body, and then introduce a relation-aware attention module to explore the fine-grained information. Finally, we fuse the features obtained from the two branches to obtain a more robust pedestrian representation. Extensive experiments verify the effectiveness of our method under the occlusion scenario.     

Unsupervised tissue classification of brain MR images for voxel‐based morphometry analysis

In this article, a fully unsupervised method for brain tissue segmentation of T1‐weighted MRI 3D volumes is proposed. The method uses the Fuzzy C‐Means (FCM) clustering algorithm and a Fully Connected Cascade Neural Network (FCCNN) classifier. Traditional manual segmentation methods require neuro‐radiological expertise and significant time while semiautomatic methods depend on parameter's setup and trial‐and‐error methodologies that may lead to high intraoperator/interoperator variability. The proposed method selects the most useful MRI data according to FCM fuzziness values and trains the FCCNN to learn to classify brain’ tissues into White Matter, Gray Matter, and Cerebro‐Spinal Fluid in an unsupervised way. The method has been tested on the IBSR dataset, on the BrainWeb Phantom, on the BrainWeb SBD dataset, and on the real dataset “University of Palermo Policlinico Hospital” (UPPH), Italy. Sensitivity, Specificity, Dice and F‐Factor scores have been calculated on the IBSR and BrainWeb datasets segmented using the proposed method, the FCM algorithm, and two state‐of‐the‐art brain segmentation software packages (FSL and SPM) to prove the effectiveness of the proposed approach. A qualitative evaluation involving a group of five expert radiologists has been performed segmenting the real dataset using the proposed approach and the comparison algorithms. Finally, a usability analysis on the proposed method and reference methods has been carried out from the same group of expert radiologists. The achieved results show that the segmentations of the proposed method are comparable or better than the reference methods with a better usability and degree of acceptance.     

基于负熵的人员脚步震动信号检测识别算法

提出一种基于震动信号的人员脚步检测识别算法,该算法根据信息论中的负熵概念,采用高阶累积量的负熵近似计算方法.仿真与实测结果证明,与一般的模式识别算法相比,该算法具有三个重要的优点,包括环境适应性强、识别准确率高和运算量小.这些优点使得该算法更适用于能量受限、随机自组的无线传感器网络,能够在野外环境下准确、简单的检测识别人员脚步震动信号.