足球视频镜头分类方法

根据视频语义分析和视频摘要等应用对于视频数据结构化的需求,提出了一种针对足球视频的镜头分类方法.通过logo模板匹配检测并定位出视频中的慢镜头,对其余的正常比赛部分做镜头边界检测完成视频切分.基于分块的思想,对正常比赛镜头帧计算其各块的场地像素比率值作为特征,利用SVM分类器将正常比赛镜头分为远镜头、中镜头、球员特写或场外镜头3类.至此,整个视频流可以表示为结构化的四类镜头类型标示序列.实验结果表明,该方法在视频切分和镜头类型识别的准确性方面具有良好的效果.     

基于透视点和区块特征的多车粘连分割方法

针对交通视频车辆检测与跟踪中经常出现的车辆前景粘连问题,提出了一种利用透视点在图像轮廓上搜索车辆分割点并通过区块特征识别车辆的粘连车辆分割方法.根据路面场景中车道线信息提取出透视点和车道区域,结合混合高斯模型与形态学梯度轮廓算法提取出车辆前景团块.利用透视点原理从前往后顺序分割粘连的车辆前景团块.对分割开的待检定区域,利用车辆区块特征进行检验识别,修正错误分割,将粘连的多辆车逐一分割.实验结果表明,该方法对直线道路上多辆车粘连的分割有较好的准确度和适应性.     

机场跑道异物定位算法研究

跑道异物的存在给机场安全造成了很大的危险隐患,为了减小异物存在导致的经济和人身安全的影响,利用三维坐标定位技术,将Hough变换改进后用于确定跑道位置,归一化算法实现以机场跑道灯为特征点的匹配,同时为了提高匹配效果,采用特征点旋转整合,有效获取特征点位置。结合传感器逐区域扫描时的方位改变与像机初始坐标系换算关系,确定跑道异物的相对空间位置,实现对跑道异物的定位。该算法利用坐标变换方法,理论基础完善,实际应用时,所需传感器数目较少,成本较低,定位准确,有很好的应用价值。     

基于支持向量机和灰度共生矩阵的纹理图像分割方法

图像分割是计算机视觉领域的研究热点。灰度共生矩阵是图像灰度的二阶统计度量,反映了纹理图像灰度在方向、局部邻域和变化幅度的综合信息,以能量、对比度、熵、差方差和差熵作为纹理图像的特征,利用支持向量机(SVM)对这些特征进行训练和分类,以达到纹理图像分割的目的。详细说明了纹理图像的分割过程,同时分析了不同参数情况下对分割精度的影响。针对Brodatz纹理库的实验结果表明:该方法具有较好的分割效果。     

基于图像的异型墙地砖在线分道系统

提出了一种异型墙地砖在线提取和纹理比较的方法。将CCD采集的墙地砖图像进行灰度阈值分割后,经过以矩形模式为特征的轮廓搜索,对原始图像中的砖面区域做标准化仿射变换处理,实现墙地砖的在线提取。依据计算共生矩阵特征向量,采用最小距离分类原理进行纹理比较分类决策,并采用自适应阈值参数加强算法的鲁棒性。基于所提取算法研制了异型墙地砖在线分道系统,并在实际现场运行成功,取得良好的分道效果。     

灰度不均的弱边缘血管影像的水平集分割方法

针对血管影像中灰度不均和弱边缘情况下已有水平集模型不能正确分割血管问题,提出一种耦合了血管影像的几何信息、边缘信息和区域信息的水平集分割方法.首先,采用Hessian矩阵的各向异性性对血管状目标进行识别,对原始影像数据进行多尺度滤波;然后采用拉普拉斯算子零交叉点的快速边缘积分方法将边缘信息嵌入能量泛函中,构建一种基于结构、边缘和区域信息的水平集分割方法.相比于单一依靠影像边缘信息或区域信息模型及其改进模型,该方法在分割严重灰度不均匀的血管造影影像上能够准确提取血管,并精确定位血管边缘.     

一种基于Kalman滤波器的自适应背景建模改进算法

基于传统Kalman滤波器理论的背景建模方法,不能很好地解决目标缓慢运动导致背景模型出现拖影的问题。针对该问题,提出了一种结合Kalman滤波器理论与动态区域重构的自适应背景建模改进算法,介绍了Kalman滤波器理论主要思想和改进算法的方法流程与效果。与传统的Kalman背景建模相比,该方法在增加少量计算复杂度的前提下,较好地解决了目标缓慢运动导致背景模型出现拖影的问题,同时也能较好地消除背景噪声。通过对图像序列的仿真实验证明:该方法具有很好的实用性与鲁棒性。     

基于声纳图像的水平集分割算法研究

针对现有的图像分割方法无法准确地分割声纳图像的问题,提出了一种改进的水平集声纳图像分割方法。介绍了LBF能量模型,借鉴其无重初始化的水平集演化思想。为克服声纳图像中复杂背景带来的负面效应,利用形态学顶帽—底帽变换对声纳图像进行预处理,并在此基础上进行无需初始化的水平集分割。进行仿真对比实验,实验结果显示:与LBF能量模型相比,改进的水平集分割方法更加适应于背景不均匀的声纳图像分割。     

Segmenting customers by transaction data with concept hierarchy

The segmentation of customers is crucial for an organization wishing to develop appropriate promotion strategies for different clusters. Clustering customers provides an in-depth understanding of their behavior. However, previous studies have paid little attention to the similarity of different items in transaction. Lack of categories and concept levels of items, results from item-based segmentation methods are not as good as expected. Through employing a concept hierarchy of items, this study proposes a segmentation methodology to identify similarities between customers. First, the dissimilarity between transaction sequences is defined. Second, we adopt hierarchical clustering method to segment customers by their transaction data with concept hierarchy of consumed items. After segmentation, three cluster validation indices are used for optimizing the number of clusters of customers. Through the compassion of normalized index, the segmentation method proposed by this study rendered better results than other traditional methods.     

An automated vision system for container-code recognition

Automatic container-code recognition is of great importance to the modern container management system. Similar techniques have been proposed for vehicle license plate recognition in past decades. Compared with license plate recognition, automatic container-code recognition faces more challenges due to the severity of nonuniform illumination and invalidation of color information. In this paper, a computer vision based container-code recognition technique is proposed. The system consists of three function modules, namely location, isolation, and character recognition. In location module, we propose a text-line region location algorithm, which takes into account the characteristics of single character as well as the spatial relationship between successive characters. This module locates the text-line regions by using a horizontal high-pass filter and scanline analysis. To resolve nonuniform illumination, a two-step procedure is applied to segment container-code characters, and a projection process is adopted to isolate characters in the isolation module. In character recognition module, the character recognition is achieved by classifying the extracted features, which represent the character image, with trained support vector machines (SVMs). The experimental results demonstrate the efficiency and effectiveness of the proposed technique for practical usage.