Numerically Invariant Signature Curves

Corrected versions of the numerically invariant expressions for the affine and Euclidean signature of a planar curve introduced by Calabi et al. in (Int. J. Comput. Vision, 26: 107–135, 1998) are presented. The new formulas are valid for fine but otherwise arbitrary partitions of the curve. We also give numerically invariant expressions for the four differential invariants parameterizing the three dimensional version of the Euclidean signature curve, namely the curvature, the torsion and their derivatives with respect to arc length.     

基于隐含多项式曲线欧氏不变量物体识别

目标物体的识别和匹配在计算机视觉、图像视频压缩与传输中都有重要应用。隐含多项式曲线对物体有良好的描述能力,用它识别和匹配目标物体是很有效的。文章给出了任意次隐含多项式曲线欧氏几何不变量的计算方法,理论证明这些不变量是完全独立的并且是完备的。文中的实验证明基于这些欧氏不变量能较好地识别发生欧氏变换的目标物体。     

Ring to Line Mapping and Orientation Invariant Transform for Object Recognition

This research uses the ring to line mapping technique to map the object image to the straight-line signals. The ‘vector magnitude invariant transform’ technique is used to transfer the object signal to an invariant vector magnitude quantity for object-identification. The ‘vector magnitude invariant transform’ technique can solve the image rotation problem. Various vertical magnitude quantity strips are generated to cope with the image-shifting problem. In this research, 105 comparisons are conducted to find the accuracy-rate of the developed algorithm. Within those 105 comparisons, 15 comparisons are conducted for self-comparison. The other 90 comparisons are conducted for comparisons between two different object images. The algorithm developed in this research can precisely classify the object image.     

基于微分不变量和区域增长法的深度图分割

用深度图(点云)直接对曲面物体进行识别,将会加大曲面物体的数学表示的难度。该文提出一种对深度图进行自动分割的方法,基于微分不变量进行初始分割,得到初始的核区域,用区域增长法进行曲面片增长,将深度图分割成多个区域。该算法原理简单、易于理解和编程。通过一个米老鼠头部深度图实例证明了该方法的有效性。     

基于Radon和解析Fourier-Mellin变换的尺度与旋转不变目标识别算法

由于正交矩对噪声鲁棒性强、重建效果好,因此被广泛应用于目标识别与分类中,但是正交矩本质上缺乏尺度变换不变性,而且必要的图像二值化与规一化过程会引入重采样与重量化误差。为此,在研究现有正交矩的基础上,提出了一种基于Radon变换和解析Fourier-Mellin变换的尺度与旋转不变的目标识别算法。该算法首先直接对目标灰度图像进行Radon变换,然后对Radon变换结果进行进一步解析,通过Fourier-Mellin变换将原图像的旋转变化转化为相位变化,将原图像的尺度变化转化为幅度变化;最后,通过定义一旋转与尺度不变函数,同时利用不变函数的4种特征,再应用k-近邻法实现分类。理论与实验结果表明,由于避免了正交矩方法存在的重采样与重量化误差,该算法的分类精度高于基于正交矩的分类方法,而且对白噪声的鲁棒性也显著高于基于正交矩的识别与分类方法。     

Higher-Order Neural Networks Applied to 2D and 3D Object Recognition

A higher-order neural network (HONN) can be designed to be invariant to geometric transformations such as scale, translation, and in-plane rotation. Invariances are built directly into the architecture of a HONN and do not need to be learned. Thus, for 2D object recognition, the network needs to be trained on just one view of each object class, not numerous scaled, translated, and rotated views. Because the 2D object recognition task is a component of the 3D object recognition task, built-in 2D invariance also decreases the size of the training set required for 3D object recognition. We present results for 2D object recognition both in simulation and within a robotic vision experiment and for 3D object recognition in simulation. We also compare our method to other approaches and show that HONNs have distinct advantages for position, scale, and rotation-invariant object recognition.The major drawback of HONNs is that the size of the input field is limited due to the memory required for the large number of interconnections in a fully connected network. We present partial connectivity strategies and a coarse-coding technique for overcoming this limitation and increasing the input field to that required by practical object recognition problems.     

基于帧间差分法和不变矩特征的运动目标检测与识别

针对视频监视的复杂背景,提出了一种基于帧间差分法和不变矩特征的运动目标检测与识别方法.在运动目标检测算法中,首先对定义的像素区域进行数据分析,然后通过帧间的数据差异产生运动信号,捕捉到场景中的运动目标.在目标识剐方法中,首先提取运动目标区域的不变矩特征,并对其矢量进行标准化,然后利用遗传小波神经网络作为模式识别器,实现运动目标的自动识别.为验证该方法及网络模型的泛化能力,在Matalb7.0中对模型进行了仿真.     

几何不变性及其在3D物体识别中的应用

三维物体识别是计算机视觉研究的重要内容之一，它要求从3D物体的2D图象中识别和定位物体．由于物体成像时会受到观察视角、摄像机参数的影响，因此使得同一物体在不同观察视角、不同摄像机参数等条件下所得到的图象存在差异．但由于几何不变性方法可以有效地消除这种差异带给3D物体识别的不利影响，所以，近20年来这种方法受到了广泛的关注和研究．为使人们了解该领域的研究现状，以对该领域的研究有所启发，首先讨论了基于几何不变性的3D物体识别方法的研究内容，包括研究的几何框架和其不变性以及几何不变性在3D物体识别中的主要应用；其次，总结性地评述了该领域的研究现状；最后，提出了研究的发展方向．     

Invariants of Families of Coplanar Conics and Their Applications to Object Recognition

This paper presents mutual invariants of families ofcoplanar conics. These invariants are compared with the use ofinvariants of two conics and a case is presented where the proposedinvariants have a greater discriminating power than the previouslyused invariants. The use of invariants for two conics is extended toany number of coplanar conics. A lambda-matrix is associated witheach family of coplanar conics. The use of lambda-matrices isextended from the single variable polynomial to multi-variablepolynomials. The Segre characteristic and other invariants of thelambda-matrix are used as invariants of the family of conics.     

Kinect 下深度信息获取技术及其在三维目标识别中的应用综述

微软公司 2010 年推出的 Kinect 深度传感器能够同步提供场景深度和彩色信息,其应用的一个关键领域就是目标 识别。传统的目标识别大多限制在特殊的情形,如：手势识别、人脸识别,而大规模的目标识别是近年来的研究趋势。 通过 Kinect 得到的 RGB-D 数据集多为室内和办公环境下获取的多场景、多视角、分目标类型的数据集,为大规模的目标 识别算法设计提供了学习基础。同时,Kinect 获取的深度信息为目标识别提供了强有力的线索,利用深度信息的识别方法 较以前的方法具有无法比拟的优势,大大地提高了识别的精度。文章首先对 Kinect 的深度获取技术做了详细介绍;其次 对现有的 3D 目标识别方法进行综述,接着对已有的 3D 测试数据集进行分析和比较;最后对文章进行小结以及对未来 3D 目标识别算法和 3D 测试数据集的发展趋势作了简单的阐述。