Preserving Topology by a Digitization Process

The main task of digital image processing is to recognize properties of real objects based on their digital images. These images are obtained by some sampling device, like a CCD camera, and represented as finite sets of points that are assigned some value in a gray-level or color scale. Based on technical properties of sampling devices, these points are usually assumed to form a square grid and are modeled as finite subsets of Z². Therefore, a fundamental question in digital image processing is which features in the digital image correspond, under certain conditions, to properties of the underlying objects. In practical applications this question is mostly answered by visually judging the obtained digital images. In this paper we present a comprehensive answer to this question with respect to topological properties. In particular, we derive conditions relating properties of real objects to the grid size of the sampling device which guarantee that a real object and its digital image are topologically equivalent. These conditions also imply that two digital images of a given object are topologically equivalent. This means, for example, that shifting or rotating an object or the camera cannot lead to topologically different images, i.e., topological properties of obtained digital images are invariant under shifting and rotation.     

基于时空注意力机制的目标跟踪算法

目标跟踪技术在智能监控、人机交互、无人驾驶等诸多领域得到了广泛的应用。近年来,学者们提出了许多高效的算法。然而,随着跟踪环境越来越复杂,目标跟踪算法在遮挡、光照变化、背景干扰等复杂环境下仍然面临着巨大的挑战,从而导致目标跟踪失败。针对上述问题,提出了一种基于时空注意力机制的目标跟踪算法。首先,采用孪生网络架构来提高对特征的判别能力;然后,引入改进的通道注意力机制和空间注意力机制,对不同通道和空间位置的特征施加不同的权重,并着重关注空间位置和通道位置上对目标跟踪有利的特征。此外,还提出了一种高效的目标模板在线更新机制,将第一帧图像特征与后续跟踪图像帧中置信度较高的图像特征进行融合,以降低发生目标漂移的风险。最后,在OTB2013和OTB2015数据集上对所提跟踪算法进行了测试。实验结果表明,所提算法的性能相比当前主流的跟踪算法提高了6.3%。     

可见光–红外特征交互与融合的YOLOv5目标检测算法

目标检测是自动驾驶系统的关键技术, 普通RGB目标检测算法在夜间和恶劣气候等场景往往表现一般, 融合可见光和红外信息的目标检测算法因而受到诸多研究关注. 现有方法通常融合结构复杂, 且忽视了模态间信息交流的重要性. 对此, 本文以YOLOv5为基本框架, 提出一种可见光–红外特征交互与融合的目标检测算法, 使用一种新的主干网络跨阶段局部(CSPDarknet53-F), 采用双分支结构分别提取可见光和红外特征. 然后, 通过特征交互模块重构各模态的信息成分和比例, 提升模态间信息交流, 使可见光和红外特征进行更充分的融合. 在FLIR-aligned和M$^3$FD数据集上的大量实验证明, 本文算法使用的CSPDarknet53-F在协同利用可见光和红外信息方面更加出色, 提升了模型精度, 同时, 拥有对抗光照强度骤变的鲁棒性     

基于分布式事件触发的交互系统研究

针对在多目标交互系统中,因查询方式致使交互节点资源受限的问题,提出了由事件触发的分布式控制方案。设计了分布式多目标虚拟交互系统,并采用Phantom Omni设备作为交互主端进行了仿真实验。在事件触发机制的情况下,系统每次单个对象交互视为完成一次事件触发,多个对象交互产生的触发事件采用分布式处理。在交互过程中,交互对象依据自身节点信息向控制系统发送交互过程中的采样值,并在Phantom Omni设备完成再现,操作者经由Phantom 0mni设备对交互状态进行感知。实验结果表明,该控制方案能够充分利用系统的节点资源,系统的动态特性能满足交互控制需求。     

Feature-based design for heterogeneous objects

Heterogeneous objects are objects composed of different constituent materials. In these objects, multiple desirable properties from different constituent materials can be synthesized into one part. In order to obtain mass applications of such heterogeneous objects, efficient and effective design methodologies for heterogeneous objects are crucial.In this paper, we present a feature based design methodology to facilitate heterogeneous object design. Under this methodology, designers design heterogeneous objects using high-level design components that have engineering significance. These high level components are form features and material features. In this paper, we first examine the relationships between form features and material features in heterogeneous objects. We then propose three synthesized material features in accordance with our examination of these features. Based on these proposed features, we develop a feature based design methodology for heterogeneous objects. Two enabling methods for this design methodology, material heterogeneity specification within each feature and combination of these material features, are developed. A physics (diffusion) based B-spline method is developed to (1) allow design intent of material variation be explicitly captured by boundary conditions, (2) ensure smooth material variation across the feature volume. A novel method, direct face neighborhood alteration, is developed to increase the efficiency of combining heterogeneous material features.Examples of using this feature based design methodology for heterogeneous object design, such as a prosthesis design, are presented.     

Tracking people in video sequences using multiple models

This paper presents a multiple model real-time tracking technique for video sequences, based on the mean-shift algorithm. The proposed approach incorporates spatial information from several connected regions into the histogram-based representation model of the target, and enables multiple models to be used to represent the same object. The use of several regions to capture the color spatial information into a single combined model, allow us to increase the object tracking efficiency. By using multiple models, we can make the tracking scheme more robust in order to work with sequences with illumination and pose changes. We define a model selection function that takes into account both the similarity of the model with the information present in the image, and the target dynamics. In the tracking experiments presented, our method successfully coped with lighting changes, occlusion, and clutter.     

FIRST: Fractal Indexing and Retrieval SysTem for Image Databases

We present an image indexing method and a system to perform content-based retrieval in heterogeneous image databases (IDB). The method is based upon the fractal framework of the iterated function systems (IFS) widely used for image compression. The image index is represented through a vector of numeric features, corresponding to contractive functions (CF) of the IFS framework. The construction of the index vector requires a preliminary processing of the images to select an appropriate set of indexing features (i.e. contractive functions). The latter will be successively used to fill in the vector components, computed as frequencies by which the selected contractive functions appear inside the images. In order to manipulate the index vectors efficiently we use discrete Fourier transform (DFT) to reduce their cardinalities and use a spatial access method (SAM), like R*-tree, to improve search performances. The sound theoretical framework underlying the method enabled us to formally prove some properties of the index. However, for a complete validation of the indexing method, also in terms of effectiveness and efficacy, we performed several experiments on a large collection of images from different domains, which revealed good system performances with a low percentage of false alarms and false dismissals.     

基于模糊类别共生矩阵的纹理疵点检测方法

纹理图像中的不规则部分通常称为疵点。纹理通常由空间分布和灰度分布共同描述,由于灰度共生矩阵能兼顾二者,因此具有很好的描述纹理的能力,不过其对纹理的正常部分与不正常部分的区分能力仍然有限,且计算效率低。为克服灰度共生矩阵以上的不足,提出了一种用模糊类别共生矩阵的方法来检测不规则纹理。该方法首先通过学习纹理的基本特征,如纹理的灰度概率密度分布、纹理主方向和周期等来确定模糊类别共生矩阵的一些关键参数,并将灰度级划分为几个纹理色调类别;然后根据后验概率函数得出各个灰度级对每个色调类别的模糊隶属度,同时计算模糊类别共生矩阵,并提取一些更为简单的特征;最后利用异常点检测的方法,即可很好地区分正常纹理和疵点。实践证明,该方法不仅比已有的灰度共生矩阵方法更简单,计算效率更高,而且能更好地表示不规则纹理。     

Performance optimization of object comparison

Comparing objects can be considered as a hierarchical process. Separate aspects of objects are compared to each other, and the results of these comparisons are combined into a single result in one or more steps by aggregation operators. The set of operators used to compare the objects and the way these operators are related with each other is called the comparison scheme. If a threshold is applied to the final result of the object comparison, the mathematical properties of the operators in the comparison scheme can be used to derive thresholds on the intermediate results. These derived threshold can be used to break of a comparison early, thus offering a reduction of the comparison cost. Using this information, we show that the order in which the operators are evaluated has an influence on the average cost of comparing two objects. Next, we proceed with a study of the properties that allow us to find an optimal order, such that this average cost is minimized. Finally, we provide an algorithm that calculates an optimal order efficiently. Although specifically developed for object comparison, the algorithm can be applied to all kinds of selection processes that involve the combination of several test results. © 2009 Wiley Periodicals, Inc.     

Context-based object-class recognition and retrieval by generalized correlograms

We present a novel approach for retrieval of object categories based on a novel type of image representation: the Generalized Correlogram (GC). In our image representation, the object is described as a constellation of GCs where each one encodes information about some local part and the spatial relations from this part to others (i.e., the part's context). We show how such a representation can be used with fast procedures that learn the object category with weak supervision and efficiently match the model of the object against large collections of images. In the learning stage, we show that by integrating our representation with Boosting the system is able to obtain a compact model that is represented by very few features, where each feature conveys key properties about the object's parts and their spatial arrangement. In the matching step, we propose direct procedures that exploit our representation for efficiently considering spatial coherence between the matching of local parts. Combined with an appropriate data organization such as Inverted Files, we show that thousands of images can be evaluated efficiently. The framework has been applied to different standard databases and we show that our results are favorably compared against state-of-the-art methods in both computational cost and accuracy.     

Key-components: detection of salient regions on 3D meshes

In this paper, we present a method to detect stable components on 3D meshes. A component is a salient region on the mesh which contains discriminative local features. Our goal is to represent a 3D mesh with a set of regions, which we called key-components, that characterize the represented object and therefore, they could be used for effective matching and recognition. As key-components are features in coarse scales, they are less sensitive to mesh deformations such as noise. In addition, the number of key-components is low compared to other local representations such as keypoints, allowing us to use them in efficient subsequent tasks. A desirable characteristic of a decomposition is that the components should be repeatable regardless shape transformations. We show in the experiments that the key-components are repeatable and robust under several transformations using the SHREC’2010 feature detection benchmark. In addition, we discover the connection between the theory of saliency of visual parts from the cognitive science and the results obtained with our technique.     

一种基于对象序列图的组件交互协议设计方法

基于组件的软件开发(component-basedsoftwaredevelopment,简称CBSD)现已成为软件开发的主流范型之一,其关心的核心问题是组件标准化与组件间的互操作性.它在互操作方面被广泛采用的思想是,分离组件的功能与交互特征,使用独立部件-交互协议来协调组件之间的交互.基于这种思想,探讨运用UML(unifiedmodelinglanguage)的对象序列图(objectsequencediagram,简称OSD)方法进行组件交互协议设计的多个方面的研究,其中包括OSD规范的形式定义以及规范的静态和动态形式分析方法,并为开发组件交互协议提出了一个集成OSD可视化建模和形式分析技术的软件工具框架.     

结合注意力机制的深度学习图像目标检测

利用卷积神经网络进行目标检测时，提取的卷积特征具有很强的平移不变性，这将削弱模型的定位性能。事实上，目标对象通常具有不同的子区域特征和宽高比特性，但在目前流行的两阶段目标检测框架中，很少考虑这些具有平移尺度敏感性的特征成分。为了优化模型的特征表达，将在两阶段目标检测框架中引入与子区域特征和宽高比特性相关的注意力特征库，并生成注意力特征图对原始的ROI池化特征进行优化。另外，在注意力特征图的辅助下，模型特征维度可以有效地进行缩减。实验结果表明，引入注意力模块后，模型的检测精度和检测速度有明显提升。     

Stochastic model utilizing spectral and spatial characteristics

In remote sensing, because of physical properties of targets, sensor pixels in spatial proximity to one another are class conditionally correlated. Our main objective is to exploit this spatial correlation. Therefore, a two-dimensional causal first order Markov model was used to extract the spatial and spectral information and, based upon it, new object classifiers with improved performance were developed. First, the minimum distance (MT) and the maximum likelihood (ML) object classifiers are discussed. Then, based on the proposed model, these two classifiers are modified, and a linear object classifier is introduced. Finally, experimental results are presented.     

Feature‐oriented programming: A new way of object composition

We propose a new model for flexible composition of objects from a set of features. Features are services of an object and are similar to classes in object‐oriented languages. In many cases, features have to be adapted in the presence of other features, which is also called the feature interaction problem. We introduce explicit interaction handlers which can adapt features to other features by overriding methods. When features are composed, the appropriate interaction handling is added in a way which generalizes inheritance and aggregation. For a set of features, an exponential number of different feature combinations is possible, based on a quadratic number of interaction resolutions. We present the feature model as an extension of Java and give two translations to Java, one via inheritance and the other via aggregation. We show that the feature model interacts nicely with several common language extensions such as type parameters, exceptions, and higher‐order functions. Copyright © 2001 John Wiley & Sons, Ltd.     

Spatial color histogram based center voting method for subsequent object tracking and segmentation

In this paper, we introduce an algorithm for object tracking in video sequences. In order to represent the object to be tracked, we propose a new spatial color histogram model which encodes both the color distribution and spatial information. Using this spatial color histogram model, a voting method based on the generalized Hough transform is employed to estimate the object location from frame to frame. The proposed voting based method, called the center voting method, requests every pixel near the previous object center to cast a vote for locating the new object center in the new frame. Once the location of the object is obtained, the back projection method is used to segment the object from the background. Experiment results show successful tracking of the object even when the object being tracked changes in size and shares similar color with the background.     

融合目标特征和空间信息的粒子滤波跟踪

传统的基于颜色直方图的粒子滤波跟踪算法不能很好地利用跟踪对象的空间结构信息,因此在邻域颜色相似或目标模型微小变化时,不能取得良好的跟踪效果。提出一种融合目标特征和目标空间位置信息的粒子滤波跟踪算法,该算法鉴于目标空间位置包含跟踪对象一定的结构信息,可以和目标特征互为补充,利用定义的融合目标特征和目标空间位置的度量函数来进行跟踪对象相似度度量,以提高跟踪算法的稳健性和精确性。同时针对粒子滤波计算粒子相似度时可并行的特点,运用OpenMP共享存储并行计算进行粒子滤波跟踪的加速。实验表明,基于融合目标特征和空间信息的粒子滤波跟踪算法能得到更鲁棒的跟踪效果,可以有效地提高目标跟踪的速度。     

Influence of contextual objects on spatial interactions and viewpoints sharing in virtual environments

Collaborative virtual environments (CVEs) are 3D spaces in which users share virtual objects, communicate, and work together. To collaborate efficiently, users must develop a common representation of their shared virtual space. In this work, we investigated spatial communication in virtual environments. In order to perform an object co-manipulation task, the users must be able to communicate and exchange spatial information, such as object position, in a virtual environment. We conducted an experiment in which we manipulated the contents of the shared virtual space to understand how users verbally construct a common spatial representation of their environment. Forty-four students participated in the experiment to assess the influence of contextual objects on spatial communication and sharing of viewpoints. The participants were asked to perform in dyads an object co-manipulation task. The results show that the presence of a contextual object such as fixed and lateralized visual landmarks in the virtual environment positively influences the way male operators collaborate to perform this task. These results allow us to provide some design recommendations for CVEs for object manipulation tasks.     

Category Level Object Segmentation by Combining Bag-of-Words Models with Dirichlet Processes and Random Fields

This paper addresses the problem of accurately segmenting instances of object classes in images without any human interaction. Our model combines a bag-of-words recognition component with spatial regularization based on a random field and a Dirichlet process mixture. Bag-of-words models successfully predict the presence of an object within an image; however, they can not accurately locate object boundaries. Random Fields take into account the spatial layout of images and provide local spatial regularization. Yet, as they use local coupling between image labels, they fail to capture larger scale structures needed for object recognition. These components are combined with a Dirichlet process mixture. It models images as a composition of regions, each representing a single object instance. Gibbs sampling is used for parameter estimations and object segmentation.