基于深度网络的多形态人脸识别

在实际的自动人脸识别系统中,输入的识别图像往往在表情、分辨率大小以及姿态方面呈现出多种变化。现在很多方法尝试通过线性或局部线性的映射来寻找由这些变化共享的统一的特征空间。利用由受限玻尔兹曼机(RBM)堆叠成的深度神经网络来发掘这些变化内在的非线性表达。深度网络能够学习高维数据到低维数据的映射关系,并且有助于提高图像分类和识别的性能。同时,为了实现在一个统一的深度框架下同时进行特征提取和识别,在网络的顶层增加了一个监督的回归层。在预训练阶段,通过训练集中不同姿态、不同表情以及不同分辨率的图像对网络进行初始化。在微调阶段,通过网络的输出与标签之间的差 并利用标准反向传播的方法 对模型的参数空间进行调整。在测试阶段,从测试库中随机选择一幅图像,获得统一空间下的特征向量。通过与参考图像库中的所有特征向量进行对比,利用最近邻域的方法识别人脸身份。在具有丰富表情以及大姿态变化的CMU-PIE人脸数据库上进行了全面的实验,结果表明,提出的方法取得了比最新的局域线性映射(或局部线性)的人脸识别方法更高的识别率。     

基于视觉标记的增强现实系统建模及配准误差问题研究

针对当前基于视觉的增强现实系统标记大、成本高的缺陷,提出一套结合简单组合标记的增强现实系统的实现方案.基于弱透视模型,以色彩和形状作为融合特征,研究了利用冗余特征实现标记物体的3D位姿还原方法.为使配准方案能够适应普通的视觉采集设备,提出根据摄像机标定时得到的扭曲参数对渲染后的模型进行再扭曲的配准修正方法.结合OpenCV和OpenGL中的空间变换的概念,给出了虚拟和物理世界配准融合的解决方案,并基于统计学数学模型,提出降低配准时的误差以及提升配准稳定性的方案.实验结果表明,提出的算法能有效地将静态标签的抖动降低至亚像素级.     

A coarse-to-fine IP-driven registration for pose estimation from single ultrasound image

A fast registration making use of implicit polynomial (IP) models is helpful for the real-time pose estimation from single clinical free-hand Ultrasound (US) image, because it is superior in the areas such as robustness against image noise, fast registration without enquiring correspondences, and fast IP coefficient transformation. However it might lead to the lack of accuracy or failure registration.In this paper, we present a novel registration method based on a coarse-to-fine IP representation. The approach starts from a high-speed and reliable registration with a coarse (of low degree) IP model and stops when the desired accuracy is achieved by a fine (of high degree) IP model. Over the previous IP-to-point based methods our contributions are: (i) keeping the efficiency without requiring pair-wised correspondences, (ii) enhancing the robustness, and (iii) improving the accuracy. The experimental result demonstrates the good performance of our registration method and its capabilities of overcoming the limitations of unconstrained freehand ultrasound data, resulting in fast, robust and accurate registration.     

Excavator 3D pose estimation using deep learning and hybrid datasets

Earthwork operations are crucial parts of most construction projects. Heavy construction equipment and workers are often required to work in limited workspaces simultaneously. Struck-by accidents resulting from poor worker and equipment interactions account for a large proportion of accidents and fatalities on construction sites. The emerging technologies based on computer vision and artificial intelligence offer an opportunity to enhance construction safety through advanced monitoring utilizing site cameras. A crucial pre-requisite to the development of safety monitoring applications is the ability to identify accurately and localize the position of the equipment and its critical components in 3D space. This study proposes a workflow for excavator 3D pose estimation based on deep learning using RGB images. In the proposed workflow, an articulated 3D digital twin of an excavator is used to generate the necessary data for training a 3D pose estimation model. In addition, a method for generating hybrid datasets (simulation and laboratory) for adapting the 3D pose estimation model for various scenarios with different camera parameters is proposed. Evaluations prove the capability of the workflow in estimating the 3D pose of excavators. The study concludes by discussing the limitations and future research opportunities.     

Amortized constant time state estimation in Pose SLAM and hierarchical SLAM using a mixed Kalman-information filter

The computational bottleneck in all information-based algorithms for simultaneous localization and mapping (SLAM) is the recovery of the state mean and covariance. The mean is needed to evaluate model Jacobians and the covariance is needed to generate data association hypotheses. In general, recovering the state mean and covariance requires the inversion of a matrix with the size of the state, which is computationally too expensive in time and memory for large problems. Exactly sparse state representations, such as that of Pose SLAM, alleviate the cost of state recovery either in time or in memory, but not in both. In this paper, we present an approach to state estimation that is linear both in execution time and in memory footprint at loop closure, and constant otherwise. The method relies on a state representation that combines the Kalman and the information-based approaches. The strategy is valid for any SLAM system that maintains constraints between marginal states at different time slices. This includes both Pose SLAM, the variant of SLAM where only the robot trajectory is estimated, and hierarchical techniques in which submaps are registered with a network of relative geometric constraints.     

Face recognition based on 2D images under illumination and pose variations

We propose a novel 2D image-based approach that can simultaneously handle illumination and pose variations to enhance face recognition rate. It is much simpler, requires much less computational effort than the methods based on 3D models, and provides a comparable or better recognition rate.     

MODEEP: a motion-based object detection and pose estimation method for airborne FLIR sequences

In this paper, we present a method called MODEEP (Motion-based Object DEtection and Estimation of Pose) to detect independently moving objects (IMOs) in forward-looking infrared (FLIR) image sequences taken from an airborne, moving platform. Ego-motion effects are removed through a robust multi-scale affine image registration process. Thereafter, areas with residual motion indicate potential object activity. These areas are detected, refined and selected using a Bayesian classifier. The resulting regions are clustered into pairs such that each pair represents one object's front and rear end. Using motion and scene knowledge, we estimate object pose and establish a region of interest (ROI) for each pair. Edge elements within each ROI are used to segment the convex cover containing the IMO. We show detailed results on real, complex, cluttered and noisy sequences. Moreover, we outline the integration of our fast and robust system into a comprehensive automatic target recognition (ATR) and action classification system.     

View-based virtual learning and recognition of 3D object using view model obtained by motion-stereo

View-based approach for learning and recognition of 3D object and its pose detection was proved to be affective and efficient, except its high learning cost. In this research, we propose a virtual learning approach which generates learning samples of views of an object from its 3D view model obtained by motion-stereo method. From the generated learning sample views, features of high-order autocorrelation are extracted, and discriminant feature spaces for object recognition and pose detection are built. Recognition experiments on real objects are carried out to show the effectiveness of the proposed method. Caihua Wang, Ph.D.: He received his B.S. in mathematics and M.E. in electronic engineering from Renmin University of China, Beijing, China in 1983 and 1986, and his Ph. D. from Shizuoka University, Hamamatsu, Japan in 1996. He is a JST domestic fellow and is doing his post doctoral research at Electrotechnical Laboratory. His research interests are computer vision and image processing. He is a member of IEICE and IPSJ. Katsuhiko Sakaue, Ph.D.: He received the B.E., M.E., and Ph.D. degrees all in electronic engineering from University of Tokyo, in 1976, 1978 and 1981, respectively. In 1981, he joined the Electrotechnical Laboratory, Ministry of International Trade and Industry, and engaged in researches in image processing and computer vision. He received the Encouragement Prize in 1979 from IEICE, and the Paper Award in 1985 from Information.     

定向完备偏序集上的模糊Scott拓扑

设L是一个完备剩余格,定义了定向完备偏序集上的Hohle意义下的满层L-拓扑,称之为模糊Scott拓扑;基于满层的L-滤子,建立了相应的Scott收敛理论。证明了一个定向完备偏序集是连续的当且仅当对于每一个满层的L-滤子,其Scott收敛等价于按模糊Scott拓扑收敛。