A network that learns to recognize three-dimensional objects

The visual recognition of three-dimensional (3-D) objects on the basis of their shape poses at least two difficult problems. First, there is the problem of variable illumination, which can be addressed by working with relatively stable features such as intensity edges rather than the raw intensity images. Second, there is the problem of the initially unknown pose of the object relative to the viewer. In one approach to this problem, a hypothesis is first made about the viewpoint, then the appearance of a model object from such a viewpoint is computed and compared with the actual image. Such recognition schemes generally employ 3-D models of objects, but the automatic learning of 3-D models is itself a difficult problem. To address this problem in computational vision, we have developed a scheme, based on the theory of approximation of multivariate functions, that learns from a small set of perspective views a function mapping any viewpoint to a standard view. A network equivalent to this scheme will thus 'recognize' the object on which it was trained from any viewpoint.     

Spotlight: Hot Target Discovery and Localization with Crowdsourced Photos

Camera-equipped mobile devices are encouraging people to take more photos and the development and growth of social networks is making it increasingly popular to share photos online. When objects appear in overlapping Fields Of View(FOV), this means that they are drawing much attention and thus indicates their popularity. Successfully discovering and locating these objects can be very useful for many applications, such as criminal investigations, event summaries, and crowdsourcing-based Geographical Information Systems(GIS).Existing methods require either prior knowledge of the environment or intentional photographing. In this paper, we propose a seamless approach called "Spotlight", which performs passive localization using crowdsourced photos.Using a graph-based model, we combine object images across multiple camera views. Within each set of combined object images, a photographing map is built on which object localization is performed using plane geometry. We evaluate the system's localization accuracy using photos taken in various scenarios, with the results showing our approach to be effective for passive object localization and to achieve a high level of accuracy.     

基于空间短文本对象的检索策略

针对传统空间文本检索策略中的效率和有效性问题, 对如何从给定的空间文本对象集合中快速有效地检索出top-k个近似结果进行研究。基于一个空间检索的通用框架, 提出一种基于空间文本对象的快速策略, 用于满足用户对效率与有效性的要求。实验结果证明该策略优于现有方法。     

基于几何元素的城市三维地理信息系统空间数据模型

空间数据模型是建立城市三维地理信息系统的核心问题。基于不同的城市空间对象划分方法 ,提出了城市三维地理信息系统的 18个空间数据模型 :①把空间现象抽象为点对象、线对象、规则面对象、规则体对象、不规则对象和DTM等 6类对象 ,基于 6种几何元素 (点、直线段、多边形、函数构造面、TIN面片、规则体素 )的不同组合 ,提出了城市三维地理信息系统的 9个空间数据模型 ;②把空间现象抽象为 5类对象 ,分别为零维空间对象 (点对象 )、一维空间对象 (线对象 )、二维空间对象 (面对象 )、三维空间对象 (体对象 )和DTM ,基于 6种几何元素的不同组合提出了 8个空间数据模型 ;③把空间现象抽象为零维对象、一维对象、二维对象、三维对象和DTM ,基于点、直线段、TIN面片、多边形、规则面、不规则面等 6类几何元素提出了一个空间数据模型。通过对一个数据模型的实验验证和各模型的理论分析 ,说明了这些模型的特点及可行性。     

基于仿射不变性特征的视点空间划分

为了简化三维物体的识别过程,提高三维物体识别的识别率,该文利用Multi-scale autoconvolution、Trace变换、Zernike矩3种仿射不变性特征,对飞机、汽车、人等三维物体进行视点空间划分,用尽可能少的不等间隔的三维物体的二维投影图像来表达三维物体,并以此为依据进行三维物体识别。在此基础上提出一种针对不同类型物体的仿射不变性特征提取策略,并建立一个实现三维物体任意姿态识别的软件系统平台,应用Princeton形状标准库中的部分模型对该平台进行测试。结果表明,该方法能够取得较好的识别效果,识别率在90%以上。     

Numerical simulation of tomographic-SAR imaging and object reconstruction using compressive sensing with L 1/2-norm regularization

By making use of multiple acquisitions of synthetic aperture radar （SAR） observations over the same area, tomographic-SAR （tomo-SAR） technology can achieve three-dimensional （3-D） imaging of the objects of interest. The compressive sensing （CS） approach has been applied to deal with the sparseness of the elevation signals. Due to its sparsity and convexity, the L1-norm regulariza- tion, as an approximated Lo-norm with an exact solution, has been employed in CS to reconstruct the reflectivity profile of the objects. In this paper, based on our studies on polarimetric scattering and SAR imaging simulations, we produce numerical multi-pass tomo-SAR observations of the terrain object. Then, we present the CS with novel L1/2- norm regularization to realize 3-D reconstruction. As a non-convex optimization problem, the L1/2-norm regularization is solved by an iterative algorithm. This numerical simulation of tomo-SAR imaging and 3-D reconstruction of the object modeling can be of great help for parameterized analysis of tomo-SAR imagery. As an example, a tomo-SAR image and 3-D reconstruction of the Beijing National Stadium model are presented.     

基于仿射不变性特征的视点空间划分

为了简化三维物体的识别过程,提高三维物体识别的识别率,该文利用Multi-scale autoconvolution、Trace变换、Zernike矩3种仿射不变性特征,对飞机、汽车、人等三维物体进行视点空间划分,用尽可能少的不等间隔的三维物体的二维投影图像来表达三维物体,并以此为依据进行三维物体识别。在此基础上提出一种针对不同类型物体的仿射不变性特征提取策略,并建立一个实现三维物体任意姿态识别的软件系统平台,应用Princeton形状标准库中的部分模型对该平台进行测试。结果表明,该方法能够取得较好的识别效果,识别率在90%以上。     

基于机器学习的裂纹识别研究现状及发展趋势

裂纹识别一直是机器视觉领域的重要研究内容,尤其是与之相关的自动检测算法在近年来备受关注.深度学习作为机器学习的一个分支,其在裂纹识别方面已显现出强大的功能和灵活性.本文对基于机器学习的裂纹识别技术的发展情况、研究现状以及典型方法进行详细介绍:首先介绍了多种机器学习方法在裂纹识别领域的应用,并从特征提取算法和应用对象等方...     

KBMDS：基于知识的地图数据库系统

ＫＢＭＤＳ是在ＭＤＳ地图数据库的基础上，结合ＡＩ技术，面向地图的基于知识的地图数据库系统，它具有四种功能：编辑功能、查询功能、学习功能和人机交互功能。系统采用了框架语义网络知识表达方法及其揄机制实现对图象物体的描述和检索，基于查询实例的学习功能提高了系统的查询效率，基于规则的图象物体描述的自动知识获取的途径。ＫＢＭＤＳ提供了两种查询语言，一种是具有二维文法的ＱＰＥ；一种是具有线性文法的ＬＩＰＬ语言     

基于Markov随机场的三维物体识别算法

为准确识别出三维物体,提出了一种新的物体特征框架,采用密集采样的多分辨率网格来描述物体观测图像的局部特征,引入Markov随机场模型对网格节点之间的几何关系进行建模。不同图像之间的匹配通过最高置信度优先算法实现,以获取两图像各个节点之间的准确匹配关系以及全局相似度。在Coil-100(columbiaobjectimagelibrary)图像数据库上,以100个物体的4、8、18、36个视角的样本为模板,用其他68、64、54和36个视角的样本进行测试,该算法识别率分别为95.75%、99.30%、100.0%和100.0%,识别准确率明显高于文献中的方法,这说明算法在基于观测图像的物体识别领域有着非常好的应用前景。     

Learnability of multi-instance multi-label learning

Multi-instance multi-label learning(MIML) is a new machine learning framework where one data object is described by multiple instances and associated with multiple class labels.During the past few years,many MIML algorithms have been developed and many applications have been described.However,there lacks theoretical exploration to the learnability of MIML.In this paper,through proving a generalization bound for multi-instance single-label learner and viewing MIML as a number of multi-instance single-label learning subtasks with the correlation among the labels,we show that the MIML hypothesis class constructed from a multi-instance single-label hypothesis class is PAC-learnable.     

面向不平衡数据的逻辑回归偏标记学习算法

偏标记学习是近几年提出的新机器学习框架,已有的逻辑回归偏标记算法尚不能解决数据不平衡问题.建立了一种可以解决数据不平衡的逻辑回归模型偏标记学习算法.基本思想是在多元逻辑回归模型中定义新的似然函数以达到处理不平衡数据的目的.算法先根据训练集中各个类别样本所占比例定义了一个新的似然函数,之后通过逼近和求导等数学手段推导得到了能够求解的光滑的逻辑回归偏标记学习模型.在UCI数据集和真实数据集上的仿真实验表明,所提算法在数据存在不平衡问题时提高了样本的平均分类精度.     

对象存储系统中的柔性对象分布策略

为了使对象存储系统在处理不同大小文件时兼顾并行性和负载均衡,提出一种能够结合哈希算法和分片算法的优点,同时尽量避免其缺点的柔性分布算法.柔性分布算法将大小文件的边界值界定为512Kbyte,小文件直接映射成一个对象并使用哈希策略映射到一个设备中;大文件分割成多个对象,分别放置在不同的设备里.实验结果显示:柔性分布算法在不同规模的系统中开销最小,且性能受设备数增加的影响较小.     

Web3D Learning Framework for 3D Shape Retrieval Based on Hybrid Convolutional Neural Networks

With the rapid development of Web3 D technologies, sketch-based model retrieval has become an increasingly important challenge, while the application of Virtual Reality and 3 D technologies has made shape retrieval of furniture over a web browser feasible. In this paper, we propose a learning framework for shape retrieval based on two Siamese VGG-16 Convolutional Neural Networks(CNNs), and a CNN-based hybrid learning algorithm to select the best view for a shape. In this algorithm, the AlexNet and VGG-16 CNN architectures are used to perform classification tasks and to extract features, respectively. In addition, a feature fusion method is used to measure the similarity relation of the output features from the two Siamese networks. The proposed framework can provide new alternatives for furniture retrieval in the Web3 D environment. The primary innovation is in the employment of deep learning methods to solve the challenge of obtaining the best view of 3 D furniture,and to address cross-domain feature learning problems. We conduct an experiment to verify the feasibility of the framework and the results show our approach to be superior in comparison to many mainstream state-of-the-art approaches.     

多视角多行人目标检测、定位与对应算法

为解决多视角视频监控中多行人情况下的遮挡问题,引入三维重建的思想,提出一种基于空间场的多视角多行人检测、定位和对应算法。该算法首先提取各视角运动前景,融合各视角二值前景图像,利用空间场实施重建,然后根据信息融合结果在空间中检测目标并定位,并由空间检测和定位结果确定各视角中目标的对应关系。在3dsM ax合成数据和实际采集数据上进行的实验测试表明,该算法对图像遮挡的处理能力强,计算复杂度低,基本满足实时要求。     

士兵和装甲车目标多尺度检测方法

针对士兵和装甲车目标的尺度差异大以及目标距离远近造成的目标多尺度问题,以YOLOv4深度学习算法为基础,提出了一种多尺度目标检测方法. 通过针对性的数据增强方法丰富小目标样本的多样性,对输入图像进行分割预处理以提高网络输入小目标的分辨率,并基于特征金字塔网络实现大、中、小目标的分离检测,最后匹配检测结果并进行NMS处理去除冗余检测框,从而实现多尺度目标检测. 实验结果表明,本文方法在保持大目标检测效果的情况下,中、小目标的平均检测精度分别提升了1.20%和5.54%,有效提高了中、小目标的检测效果.      

Online Metric Learning for Relevance Feedback in E-Commerce Image Retrieval Applications

Relevance feedback plays a key role in multiple feature-based image retrieval applications. This paper describes an online metric learning approach for a set of ranking functions. In the feedback round, the most relevant and most nonrelevant images related to the target image are selected to construct a relative comparison triplet. The weighting parameters of the multiple ranking functions are updated by minimizing a quadratic objective function constrained by the triplet. The approach unifies the learning algorithm for the most commonly used ranking functions. Thus, multiple features with their own ranking function can easily be employed in the ranking module without feature reconstruction. The method is computationally inexpensive and appropriate for large-scale e-commerce image retrieval applications. Customized ranking functions are well supported. Practically, simplified ranking functions yield better results when the number of query rounds is relatively small. Experiments with an image dataset from a real e-commerce platform show the superiority of the proposed approach.     

一种动态对象/关系映射框架及其实现

针对现有的对象/关系映射组件不能实现动态改变数据类型的不足,构造了一个动态O/R映射框架DORM.该框架包含了一个对象模型UDO和一种查询语言UQL.DORM不仅提供类型、对象、类型间的继承和组合关系的存储和处理,而且实现了类型的动态更新.最后给出了DORM的实现和应用.     

一种基于贝叶斯理论的目标验证方法

提出了一种验证图像中候选目标的新方案.综合目标的先验知识,将目标验证转化为给定候选目标的条件下,图像特征观察集合的条件概率问题.同时,提取图像的短线段特征,利用短线段特征的观察,给出了一种目标验证方法排除复杂场景图像中虚假建筑物目标.在自然图像集上进行的实验表明,所提出的方法能有效排除虚假目标,满足处理自然图像的要求.     

基于多图像组信息的人脸识别研究

提出利用多图像组信息构建二部学习框架进行人脸识别。首先,利用两种不同的多图像组信息源分别学习两个相应的度量空间模型;其次,将得到的模型合并为一个统一的判别距离度量空间;最后,对所构建的二部学习框架进行范化,使得框架中的多图像组信息能够用来进行子空间学习和距离度量学习。通过在多个标准通用数据集上得到实验结果验证了所提出方法的有效性。