Learning of perceptual grouping for object segmentation on RGB-D data

Object segmentation of unknown objects with arbitrary shape in cluttered scenes is an ambitious goal in computer vision and became a great impulse with the introduction of cheap and powerful RGB-D sensors. We introduce a framework for segmenting RGB-D images where data is processed in a hierarchical fashion. After pre-clustering on pixel level parametric surface patches are estimated. Different relations between patch-pairs are calculated, which we derive from perceptual grouping principles, and support vector machine classification is employed to learn Perceptual Grouping. Finally, we show that object hypotheses generation with Graph-Cut finds a globally optimal solution and prevents wrong grouping. Our framework is able to segment objects, even if they are stacked or jumbled in cluttered scenes. We also tackle the problem of segmenting objects when they are partially occluded. The work is evaluated on publicly available object segmentation databases and also compared with state-of-the-art work of object segmentation.     

基于纹元森林和显著性先验的弱监督图像语义分割方法

弱监督语义分割任务常利用训练集中全体图像的超像素及其相似度建立图模型,使用图像级别标记的监督关系进行约束求解。全局建模缺少单幅图像结构信息,同时此类参数方法受到复杂度限制,无法使用大规模的弱监督训练数据。针对以上问题,该文提出一种基于纹元森林和显著性先验的弱监督图像语义分割方法。算法使用弱监督数据和图像显著性训练随机森林分类器用于语义纹元森林特征(Semantic Texton Forest, STF)的提取。测试时,先将图像进行过分割,然后提取超像素语义纹元特征,利用朴素贝叶斯法进行超像素标记的概率估计,最后在条件随机场(CRF)框架下结合图像显著性信息定义了新的能量函数表达式,将图像的标注(labeling)问题转换为能量最小化问题求解。在MSRC-21类数据库上进行了验证,完成了语义分割任务。结果表明,在并未对整个训练集建立图模型的情况下,仅利用单幅图像的显著性信息也可以得到较好的分割结果,同时非参模型有利于规模数据分析。     

基于一种动态颜色模型的模糊C-均值图像分割

不同颜色的可见光本质上是具有不同波长范围的电磁波.本文试探性地提出了一种动态颜色模型,它模拟了成像曝光时间内图像平面所接收到的电磁波的动态变化.离散化之后,彩色图像的颜色特征能够被表示成一个K维矢量,称为彩色图像的动态颜色空间表示.然后建立了模糊C-均值分割算法,分别在动态颜色空间和RGB空间分割彩色图像,实验结果表明动态颜色空间的分割结果优于RGB空间的分割,从而验证了动态颜色空间的性能.笔者相信本文所提出的动态颜色模型也能够被用于纹理分析或其它的图像处理领域.     

A New Real Time Object Segmentation and Tracking Algorithm and its Parallel Hardware Architecture

Most of the emerging content-based multimedia technologies are based on efficient methods to solve machine early vision tasks. Among other tasks, object segmentation is perhaps the most important problem in single image processing. The solution of this problem is the key technology of the development of the majority of leading-edge interactive video communication technology and telepresence systems. The aim of this paper is to present a robust framework for real-time object segmentation and tracking in video sequences taken simultaneously from different perspectives. The other contribution of the paper is to present a new dedicated parallel hardware architecture. Its composed of a mixture of Digital Signal Processing (DSP) and Field Programmable Gate Array (FPGA) technologies and uses the Content Addressable Memory (CAM) as a main processing unit. Experimental results indicate that small amount of hardware can deliver real-time performance and high accuracy. This is an improvement over previous systems, where execution time of the second-order using a greater amount of hardware has been proposed.Mahmoud Meribout (BS85, M91). Received the PhD degree in Electronic Engineering from the University of Technology of Compiegne (France), in January 3rd 1995. He worked one year as an Associate Researcher, where he has been involved in some industrial projects related to hardware board design and video processing. FromNovember 1995 to October 2000, he has beenworking in Japan, with NTT and NEC corporations respectively, where he has been involved in several projects related to hardware & software design of next generation multimedia and networking equipments. He is holding several Japanese and American patents related to this particular topic. In 1998, he has received the NTT best award for his research and development records.Maamouru Nakanishi received the B.S. and M.S. in Electronics in 1985 and 1987 respectively, all from Kyoto University. Since 1987, he has been engaged in R&D of parallel processing architectures, memory-processor integration technology at NTT Corporation (Tokyo, Japan), when he has been involved in research on Advanced Content Addressable Memory applied to video processing. In 1998, he has received the NTT best award for his research and development records.     

SAR图像理解与解译研究进展

本文主要评述SAR图像理解与解译的研究现状，并指出了该领域的相关问题及今后的发展趋势．论文从SAR图像相干斑抑制、图像分割、特征提取、识别与分类等六个方面较深入地分析了SAR图像理解与解译关键技术的最新进展，指出了进一步研究的方向．     

结合肤色分割的支持向量机人脸检测方法

人脸检测是人脸识别与图像及视频检索的一项重要任务。本文提出了一种结合分割与模板匹配-支持向理机(Support Vector Machines,SVM)的正面人脸检测算法。算法在肤色区域分割基础上，输入图像采用基于Harr小波描述的特征提取方法，然后结合平均脸模板匹配利用 SVM分类器进行分类。实验结果表明该算法十分有效。     

一种快速红外图像分割方法

为准确地实现目标识别,提出了将二维最大熵图像分割方法应用于红外图像实行分割．利用图像的二维直方图,二维最大熵分割方法不仅考虑了象素的灰度信息,而且还充／矿利用了象素的空间领域信息,能取得较为理想的分割结果．然而该方法所需的巨大运算量限制了其实际应用．运用PSO算法代替穷尽搜索获得阈值向量,求解速度可提高300～400倍,提高了分割效率．通过对实际的红外图像分割表明,这种方法简单、有效．     

基于Contourlet域HMT模型的多尺度图像分割

基于Contourlet系数分布统计特性,结合隐马尔可夫树（HMT）模型和贝叶斯准则提出一种新的图像分割算法.为了更有效保持Contourlet域不同尺度间的信息,提出一种新的加权邻域背景模型,给出了基于高斯混合模型的象素级分割算法和基于新的背景模型的多尺度融合算法.分别选择合成纹理图像、航拍图像和SAR图像进行实验,并与小波域HMTseg方法进行比较以说明算法的有效性.对合成纹理图像给出错分概率作为评价参数.实验结果表明本文方法不但在边缘信息和方向信息保持上有明显改进,而且错分概率明显降低,对真实图像得到了理想的分割效果.     

一种序列图像中运动点目标的检测方法

序列图像中运动点目标的检测按图像的成像系统不同,可分为红外图像中运动点目标的检测和可见光图像中运动点目标的检测,而现有检测算法多是针对前者。为寻找一种适用于两种类型图像的运动点目标的检测方法,以多云天空为研究背景,提出了一种新的运动点目标检测算法。采用高通滤波和形态学滤波相结合的方法进行背景抑制,基于检测前跟踪（TBD）的基本思想,根据相邻三帧进行目标分割,利用轨迹能量累积方法完成目标检测。理论分析和仿真结果表明,该算法简单易行．既适用于红外图像又适用于可见光图像的运动点目标,而且对目标的运动速度和方向无任何限制。     

SAR图像舰船目标分割算法研究

针对SAR图像固有的乘性斑噪，把概率竞争网络用于SAR图像分割和水上目标检测，充分利用了图像像素间的空间邻接关系，提高了分割、检测的准确性和有效性，取得较好的结果。