基于小波计算的平面物体射影不变性的识别

利用非均匀有理B样条从单幅图像中提取平面物体在射影变换下的几何不变特征,建立经典框架.讨论用小波计算框架的一维和二维特征不变矩的方法以减少计算量.实验结果表明所选取的特征不变量和算法有较小的计算量,对二维平面物体的识别有很好的性能,特别是允许景物中有部分的遮挡物存在.     

基于小波计算的平面物体射影不变性的识别

利用非均匀有理B样条从单幅图像中提取平面物体在射影变换下的几何不变特征,建立经典框架.讨论用小波计算框架的一维和二维特征不变矩的方法以减少计算量.实验结果表明所选取的特征不变量和算法有较小的计算量,对二维平面物体的识别有很好的性能,特别是允许景物中有部分的遮挡物存在.     

基于线性不变矩和角度向量的立体匹配算法

针对室内场景双目立体匹配有别于一般场景立体匹配的特殊性,提出了一种计算简便、准确度高的立体图像匹配算法.该算法首先利用canny算子检测物体的边缘,根据边缘的线性不变矩寻找出目标物体,然后提取出目标物体轮廓的特征点,利用角度直方图计算出左右图像的旋转角度,最后利用角度向量实现左右图像的对应像素点的匹配.线性不变矩有效地将计算复杂度由二维降低到一维,大大降低了计算量.角度向量的提出降低了特征点匹配的复杂度,而且计算简便,准确率高.实验表明,该算法对图像的缩放、旋转、平移均免疫,具有较高的识别精度和良好的抗干扰性,计算效率高于传统方法,有着较高的应用价值.     

光电测量二维微角位移的新方法研究与实现

为了解决对旋转台定位精度校准的同时又能对其台面俯仰角度进行测量的问题,提出了一种光电测量二维微角位移的新方法。在激光自准直光路中,采用数字信号处理器（DSP）获取准直物镜焦平面上V字形分划板在自扫描光电二极管列阵（SSPA）上两次成像之间的位移量,根据光学成像原理与像的外形特征,建立了像的平面位移量与被测物体二维角位移量的关系式,计算出角位移量。此方法测量精度达到0．9”,满足校准的需要。     

一种基于视图和支持向量机的三维物体识别方法

为提高三维物体识别系统性能并减少计算复杂性,本文提出了一种基于视图的方法.首先从三维物体的二维视图中提取颜色矩、纹理特征和仿射不变矩.颜色矩对于物体的大小和姿态不敏感且性能稳健.纹理特征可区别形状相似但外观不同的物体.仿射不变矩在物体发生仿射形变下具有不变性.本文将上述各种特征组合为23个分量的特征向量,送入支持向量机进行训练并识别.基于两种公开的三维物体数据库COIL-100和ALOI测试了本文方法性能.当每物体训练视角为36个(视角间隔10°)时,在两个数据库上的实验都达到了100%的识别率.进一步减少训练视角数量也达到较满意的识别性能,优于文献中的方法.     

基于结构照明和BP神经网络的三维物体识别

提出一种具有旋转不变性的三维物体识别的新方法,该方法通过结构光照明的方法,使物体的高度分布以变形条纹的形式编码于二维强度图中,由于条纹图包含有物体的高度分布信息,因此对条纹的相关识别具有本征三维识别的特点.旋转不变性是通过BP神经网络实现的.计算机模拟结果表明,用二维强度像的基频分量做训练样本设计BP神经网络,选择训练样本和隐藏层神经元的数目,基于结构光编码的BP神经网络对三维物体具有良好的旋转不变识别效果.     

块匹配与光流技术在微结构动态特性检测中的应用

在微结构动态特性的检测中引入块匹配技术和光流技术.结合微结构的运动特性,分别提出了基于块匹配技术和基于光流技术的两种微结构二维运动估计算法.利用这两种二维运动估计算法对平面微谐振器做动态特性分析,均取得了较好的测量结果.对这两种二维运动估计算法在微谐振器动态特性中的测量应用做比较,结果表明:块匹配算法适合简单的平移运动测量,计算量小,但它要求被测器件在结构上有较为特殊或显著的运动特征区域;光流算法通过分析单个像素的灰度分布特性来测量被测器件的运动特性.与块匹配法相比,它并不要求被测器件有较为特殊的运动特征结构.由于光流法以单个像素为计算单位,且其中包含有较多的差分计算,计算过程较为复杂,因此测量分析所需的时间长.     

平面P波作用下半空间中三维洞室的动力响应

采用一种高精度间接边界积分方程法求解了平面P波入射下弹性半空间中三维洞室的动力响应。在精度检验基础上,以圆球形洞室为例,对地表位移动力响应和洞周动应力集中特征进行计算分析,并将三维模型解答同相应二维模型解答进行比较。研究表明,半空间中三维洞室周围波动特征十分复杂,位移和应力峰值及其空间分布特征强烈依赖于入射波频率、方向和洞室埋深。总体上,三维洞室上方地表位移放大效应要弱于二维情况,除个别频率外,标准化位移幅值在3.0以下;三维球形洞室x-z截面内周向应力空间分布特征接近于二维情况,动应力集中幅度略大于后者。三维地下洞室对波的散射应该建立三维模型进行计算分析。     

一种新型非接触位姿检测系统研究

提出一种基于位置敏感探测器(简称PSD)多光束同步检测技术的位姿检测新方法,采用两个二维PSD从不同角度同时测量出目标物体上多个特征点的二维坐标,然后通过坐标变换计算得到各点的三维空间坐标;根据位姿变化前后的各点空间坐标,采用比较简单的算法计算出目标物体的位姿参数.通过算例验证该方法是可行的;由初步的实验获得目标物体的位姿参数,最大平移误差为1.0574mm,最大旋转误差为0.2851rad,需要通过多种途径进一步提高系统的检测精度.     

组合体读图技巧探讨

读图是一种从二维平面图形,通过思维、构思、在想像中还原成三维空间物体的过程。本文通过组合体的形成结构特点,举例说明读图的具体步骤和借助画图切割过程读图的方法技巧。     

一种视频图像序列中运动对象的分割与跟踪算法

本文提出了一种视频图像序列中运动对象的分割与跟踪算法。该算法通过Canny算子检测出差帧图像的边缘信息,并结合当前帧与背景帧的边缘图像,提取出运动对象。在后续帧中通过建立前帧感兴趣运动对象与当前帧中各运动对象的帧间向量来跟踪当前帧中感兴趣的视频对象。实验结果表明,该算法可行,而且由于该算法简单、计算复杂度小,能很好地满足实时监控系统中对感兴趣运动对象的提取与跟踪。     

Modified Factorial-Free Direct Methods for Zernike and Pseudo-Zernike Moment Computation

Modified direct methods for the computation of Zernike and pseudo-Zernike moments are presented in this paper. The presence of many factorial terms in direct methods for computing Zernike-type moments makes their computation process a very time consuming task. Although the computational power of the modern computers is impressively increasing, the calculation of the factorial of a big number is still an inaccurate numerical procedure. The main concept of this paper is that, using Stirling's approximation for the factorial and applying some suitable mathematical properties, novel factorial-free direct methods can be developed. The resulting moments are not equal to those computed using the original direct methods, but they are a sufficiently accurate approximation of them. In addition, their variability does not affect their ability to uniquely describe and distinguish the objects that they represent. This is verified by appropriate pattern recognition experiments.      

Recognition of motion-blurred images by use of the method of moments

Image motion causes a blur that changes features of objects and therefore complicates the task of automatic recognition. In this work we develop two recognition methods for motion-blurred images. For the first method we assume that the motion function and direction during the exposure are given. We develop the relation between the blurred-image moments and the original-image moments based on the motion function only. The recognition is carried out by comparing the moments of the restored image against the moments of the image database. In the second method the motion function is not known. In this case image moments that are invariant with respect to the motion blur are identified, and only these moments are used for recognition. The advantage of the suggested methods is that no time-consuming image restoration is required prior to recognition.     

框架结构智能主动控制的研究

针对框架系统所具有的复杂性、不确定性、大柔度等特点。建立在系统可控性格拉姆矩阵与可观性格拉姆矩阵相等的基础上,为保证系统控制的顺利实现,本文采用了基于内平衡理论的模型降阶方法,取可控度较大的几阶模态进行控制。首先对系统进行离散进而进行控制。本文中,控制方案采用了能够控制不精确模型及未知外力情况下可以达到较好控制效果的广义预测控制。本文最后给出了计算实例。算例表明这一方法是有效的。     

Block‐based recursive motion filtering for preserving true motion vectors in time‐varying texture objects

True motion estimation is necessary for deinterlacing, frame‐rate conversion, and film judder compensation. There have been several block‐based approaches to find true motion vectors by tracing minimum sum‐of‐absolute‐difference values by considering spatial and temporal consistency. However, the algorithms cannot find robust motion vectors when the texture of objects is changed. To find the robust motion vectors in the region, a recursive vector selection scheme and an adaptive weighting parameter are proposed. Previous frame vectors are recursively averaged to be utilized for motion error region. The weighting parameter controls fidelity to input vectors and the recursively averaged ones, where the input vectors come from the conventional estimators. If the input vectors are not reliable, then the mean vectors of the previous frame are used for temporal consistency. Experimental results show more robust motion vectors than those of the conventional methods in time‐varying texture objects. © 2008 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 18, 265–275, 2008; Published online in Wiley InterScience (www.interscience.wiley.com).     

一种快速运动目标检测与跟踪算法

运动目标的检测与跟踪是计算机视觉和图象编码研究的主要内容,有着广泛的应用领域。基于光流场的检测与跟踪是其常用的方法之一,但其计算结果和效率是不能令人满意的。提出了基于帧间差阈值法和光流场相结合的快速运动目标检测与跟踪算法,诸结果表明该算法简单实用、运算速度快,克服了单纯光流场方法的不足。     

单一方法评价结论漂移性的测度研究

主要研究单一方法评价结论的漂移性，以便为组合评价提供实验依据。把现实评价问题分为3大类分别进行仿真实验，利用MATLAB生成待评价对象，利用不同方法进行评价，把评价结果进行标准化处理，并判别评价方法的相容性，以相容方法的平均值为参照系，求各评价方法的漂移度。     

Frontal gait recognition based on spatio-temporal interest points

This paper proposes a spatio-temporal interest point (STIP)-based human gait recognition method for frontal gait videos. The proposed method extracts histograms of oriented gradient (HOG) for STIPs directly from gait videos without involving a gait cycle segment, which is required by silhouette analysis as used by most gait recognition methods. Moreover, silhouette extraction may be affected by noise and carried objects leading to consequent recognition error. Matches of STIPs between two gait videos are then found in terms of HOG to measure the similarity of two videos so as to achieve the goal of gait recognition. The experimental results offer evidence that our method outperforms the existing methods on the Carnegie-Mellon University Motion of Body database and the Chinese Academy of Sciences Institute of Automation data-set B.     

Moving/motionless foreground object detection using fast statistical background updating

Abstract

In video surveillance, the detection of foreground objects in an image sequence from a still camera is very important for object tracking, activity recognition and behaviour understanding. The conventional background subtraction cannot respond promptly to dynamic changes in the background, and temporal difference cannot accurately extract the object shapes and detect motionless objects. In this paper, we propose a fast statistical process control scheme for foreground segmentation. The proposed method can promptly calculate the exact grey-level mean and standard deviation of individual pixels in both short- and long-term image sequences by simply deleting the earliest one among the set of images and adding the current image scene in the image sequence. A short-term updating process can be highly responsive to dynamic changes of the environment, and a long-term updating process can well extract the shape of a moving object. The detection results from both the short- and long-term processes are incorporated to detect motionless objects and eliminate non-stationary background objects. Experimental results have shown that the proposed scheme can be well applied to both indoor and outdoor environments. It can effectively extract foreground objects with various moving speeds or without motion at a high process frame rate.     

An Elevator Button Recognition Method Combining YOLOv5 and OCR

Fast recognition of elevator buttons is a key step for service robots to ride elevators automatically. Although there are some studies in this field, none of them can achieve real-time application due to problems such as recognition speed and algorithm complexity. Elevator button recognition is a comprehensive problem. Not only does it need to detect the position of multiple buttons at the same time, but also needs to accurately identify the characters on each button. The latest version 5 of you only look once algorithm (YOLOv5) has the fastest reasoning speed and can be used for detecting multiple objects in real-time. The advantages of YOLOv5 make it an ideal choice for detecting the position of multiple buttons in an elevator, but it’s not good at specific word recognition. Optical character recognition (OCR) is a well-known technique for character recognition. This paper innovatively improved the YOLOv5 network, integrated OCR technology, and applied them to the elevator button recognition process. First, we changed the detection scale in the YOLOv5 network and only maintained the detection scales of 40 * 40 and 80 * 80, thus improving the overall object detection speed. Then, we put a modified OCR branch after the YOLOv5 network to identify the numbers on the buttons. Finally, we verified this method on different datasets and compared it with other typical methods. The results show that the average recall and precision of this method are 81.2% and 92.4%. Compared with others, the accuracy of this method has reached a very high level, but the recognition speed has reached 0.056 s, which is far higher than other methods.