基于加权相似性度量的特征匹配方法

针对图像特征匹配过程中采集图像易受噪声、光照、尺度等因素影响使产生的匹配结果鲁棒性差、误匹配率高等问题,提出一种基于加权相似性度量（WSM）的特征匹配方法。该方法首先采用基于网格多密度聚类的特征匹配（FM_GMC）算法对原始图像进行特征聚类块划分;其次在每一特征聚类块中,采用Canny提取边缘特征点并使用尺度不变特征变换（SIFT） 进行描述;然后采用加权的方式对特征聚类块之间的空间上下文信息间的Hausdorff距离、图像特征点外观描述子间的欧氏距离以及图像特征点的局部几何灰度信息的归一化互相关度量（NCC）进行相似性度量;最后依据最近邻距离比值（NNDR）对相似性度量结果进一步优化,从而确定特征匹配结果。以古建筑图像为数据集的实验结果表明WSM方法的平均匹配精确率达到92%,在匹配数量和精确率上优于常用的特征匹配方法,验证了该方法的有效性和鲁棒性。     

基于灰度空间特征的模糊C均值聚类图像分割

模糊C均值(FCM)聚类算法广泛用于图像的自动分割,但是该算法没有考虑像素的灰度和空间特征,对噪声十分敏感.因此提出一种改进的算法,在传统的FCM聚类的基础上,运用邻域像素的灰度相似度和聚类分布统计来构造新的隶属函数,对图像进行聚类分割.该方法不仅有效地抑制了噪声的干扰,而且把错分类的像素很容易的纠正过来.对两种类型的含噪图像的实验结果表明该方法对噪声具有很强的鲁棒性和对像素聚类的正确性.     

基于动态拓展的特征匹配方法

针对图像间结构复杂、纹理重复的对象中存在特征匹配鲁棒性差、误匹配率高的问题,结合最近邻思想,提出一种基于动态拓展的特征匹配方法。输入待匹配的两幅图像,采用SIFT算子提取图像初始特征点几何位置信息,构建基础数据集;依据基础数据集,采用核心点周围邻域逐层约束的动态拓展聚类方法,划分图像聚类簇;设计度量函数确定图像对应聚类簇,生成对应簇内特征点的描述子;采用最近邻距离比准则进行特征匹配。以Oxford VGG标准数据集和古建筑图像为对象,验证了该算法的精确性与鲁棒性。     

融合Tamura纹理特征的改进FCM脑MRI图像分割算法

针对FCM算法在分割脑MRI图像时存在噪声敏感性和初始聚类中心随机性的问题,提出一种融合图像Tamura纹理特征的改进FCM图像分割算法.首先提取图像的Tamura纹理特征,将其与灰度特征线性加权构成融合特征.然后使用模糊邻域关系计算像素点的密度,将其与距离关系结合自适应选取初始聚类中心.最后使用融合特征作为更新隶属度和聚类中心的特征约束.实验利用该方法与FCM,D-FCM,WKFCM方法对Brain Web脑MRI数据集中的图像进行分割,并在抗噪性、准确性和运行效率方面进行了比较.实验结果表明,所提算法能获取更优的初始聚类中心,在处理噪声和灰度不均匀图像上表现出更好的鲁棒性,能够快速有效地分割脑MRI图像.     

基于PDCRAN-SIFT的特征匹配技术研究

研究复杂背景下图像的目标识别,提高复杂背景下识别攻击目标的准确性和快速性。针对SIFT特征具有冗余度高、可分性差的缺点,提出PDCRAN-SIFT的方法对SIFT特征进行聚类精选,首先引入概率距离聚类的方法对SIFT特征进行聚类,选取每一类的代表特征向量作为粗聚类特征向量,然后运用随机采样一致算法剔除粗聚类特征向量中与目标图像误匹配的特征向量,从而得到精聚类匹配特征向量。实验表明,经过处理之后的PDCRAN-SIFT特征向量,冗余度大大减小,匹配时间缩短了50%左右,特征可分性和对光照、视角、噪声的鲁棒性也比SIFT特征明显增强。     

基于变分水平集的图像模糊聚类分割

结合变分水平集方法和模糊聚类,提出了一个基于变分水平集的图像聚类分割模型.该模型引入了一个基于图像局部信息的外部模糊聚类能量和一个新的关于零水平集的正则化能量,使得该模型对噪声图像的聚类分割更具鲁棒性.通过在能量泛函中加入一个内部约束能量约束水平集函数为符号距离函数,可以使水平集演化过程无需重新初始化.进一步提出了一种变分形式的聚类中心更新方法,实现了半监督的图像聚类分割.实验中采用不同类型的图像与FCM聚类模型、CV模型、Samson模型进行了对比实验,实验结果显示,该模型能够克服图像中噪声的影响,取得较满意的聚类分割效果.     

烟花算法优化的软子空间MR图像聚类算法

现有的软子空间聚类算法在分割MR图像时易受随机噪声的影响,而且算法因依赖于初始聚类中心的选择而容易陷入局部最优,导致分割效果不理想.针对这一问题,提出一种基于烟花算法的软子空间MR图像聚类算法.算法首先设计一个结合界约束与噪声聚类的目标函数,弥补现有算法对噪声数据敏感的缺陷,并提出一种隶属度计算方法,快速、准确地寻找簇类所在子空间;然后,在聚类过程中引入自适应烟花算法,有效地平衡局部与全局搜索,弥补现有算法容易陷入局部最优的不足.EWKM,FWKM,FSC,LAC算法在UCI数据集、人工合成图像、Berkeley图像数据集以及临床乳腺MR图像、脑部MR图像上的聚类结果表明,所提出的算法不仅在UCI数据集上能够取得较好的结果,而且对图像聚类也具有较好的抗噪性能,尤其是对MR图像的聚类具有较高的精度和鲁棒性,能够较为有效地实现MR图像的分割.     

基于免疫K-means聚类的无监督SAR图像分割

利用图像纹理的信息熵特征,并结合空间矩阵的概念,提出一种基于免疫K-means聚类的无监督SAR图像分割算法.免疫规划的K-means聚类克服收敛结果易陷于局部极值的缺点,且保持K-means算法快速收敛的特点.信息熵的应用可有效抑制相干斑噪声的影响,空间矩阵的引入实现聚类过程中类别的自动合并.该算法执行复杂度不高,对噪声的影响有较强的鲁棒性,分割结果较好,是一种实用的SAR图像分割算法.     

FCM和LS-SVM相结合的粮虫图像分割算法

模糊C均值(FCM)聚类算法广泛用于图像的自动分割,但是该算法对噪声十分敏感。因此提出一种改进的算法,在传统的FCM聚类的基础上,结合LS-SVM对图像进行分类。该方法对噪声有很强的鲁棒性和对像素聚类的正确性。     

多特征-谱聚类的SAR图像溢油分割

经典的K聚类算法,并不适合实现任意形状的聚类,而且有容易陷入局部最小值的不足.提出基于多个纹理特征的谱聚类算法,该方法用灰度共生矩阵(GLCM)提取合成孔径雷达 (SAR)图像的多个特征值,构建谱聚类的特征矩阵,并依据规范切准则,用K均值聚类的方法对拉普拉斯矩阵的第二小的特征值对应的特征向量进行聚类,实现基于SAR图像的溢油的分割.新方法与传统的K聚类方法比较,可以减少相干斑噪声对分割结果的影响,较好的保持图像边缘.仿真结果显示,该算法对于相干斑噪声影响较大的图像具有较强的鲁棒性.     

基于分层曲线简化的运动捕获数据关键帧提取

引入骨骼夹角作为运动特征,并以此确定候选关键帧;采用分层曲线简化算法精选候选关键帧获得最终关键帧集合;通过对算法参数自适应调节,以满足不同压缩率要求.实验结果表明：该算法具有良好的数据压缩效果及较强的运动概括能力,并在一定程度上保证了同类相似运动间关键帧集合的一致性.     

Automatic matching of high‐resolution SAR images

Based on high‐resolution SAR data, in this paper, a novel automatic matching model is proposed. The model, which employs a coarse to fine strategy as a whole, consists of three steps. In the first step, edge features are extracted on different levels of pyramid images and an efficient Hausdorff distance‐based method is used to yield a coarse global feature match. Due to bi‐tree searching, the bottleneck of Hausdorff distance's matching is well resolved. Secondly, SSDA (Sequence Similarity Detection Algorithm) is employed to acquire tie‐points using a cross‐searching approach which treats features extracted from master and slave images equally. Finally, local‐adaptive splitting algorithm with MMSE (Minimum Mean Square Error) is used to achieve a fine matching; local‐adaptive splitting algorithm is the essential process to achieve sub‐pixel matching accuracy, which enhances the process's flexibility and robustness.

Airborne SAR images with high resolution are provided by the Institute of Electronics, CAS and used for experiments—the results of the experiments demonstrate that the model proposed in this paper is robust, with high accuracy (up to a fraction of a pixel), and can be successfully applied to automatic matching of high‐resolution SAR images.     

Stereo matching using weighted dynamic programming on a single-direction four-connected tree

In recent years, stereo matching based on dynamic programming (DP) has been widely studied and various tree structures are proposed to improve the matching accuracy. However, previous DP-based algorithms do not incorporate all the smoothness functions determined by the edges between the adjacent pixels in the image, which will usually lead to lower matching accuracies. In this paper, we propose a novel stereo matching algorithm based on weighted dynamic programming on a single-direction four-connected (SDFC) tree. The SDFC tree structure is a new tree structure which includes all the edges in the image and the disparity of a pixel can be affected by all the edges in the image. However, in the SDFC tree, conventional DP-based algorithms will make the pixels that are far away from the root node provide higher energy than the nearby pixels, which will decrease the matching accuracy. So, the weighted dynamic programming approach is proposed to optimize the energy function on the new tree structure, and all the pixels in the SDFC tree are treated equivalently. Dynamic programming in the SDFC tree of every pixel in the image separately is very time-consuming, so a fast DP optimization method is designed for the SDFC tree, which reduces the computational complexity of the proposed weighted DP algorithm to 12 times of conventional DP based algorithm. Experiments show that our algorithm not only produces quite smooth and reasonable disparity maps which are close to the state-of-the-art results, but also can be implemented quite efficiently. Performance evaluations on the Middlebury data set show that our method ranks top in all the DP-based stereo matching algorithms, even better than the algorithms that apply segmentation techniques. Experimental results in an unmanned ground vehicle (UGV) test bed show that our algorithm gets very good matching results in different outdoor conditions, even on the asphaltic road which is considered to be textureless. This illustrates the robustness of our algorithm.     

基于BP神经网络的景象匹配算法

针对以可见光图像为基准、红外图像为实测的景象匹配问题,提出了一种基于BP神经网络的景象匹配方法。该方法首先抽取满足平移、旋转、尺度不变性的Krawtchouk不变矩作为红外目标的特征描述并构造特征向量,随后,基于主分量分析法消除特征向量各个分量间的相关性,去掉多个对目标匹配识别贡献不大的特征的影响,形成描述目标的有效特征向量。接下来,构造三层BP神经网络,以有效匹配特征作为输入,匹配位置作为输出,按照经验公式确定隐层节点个数,基于样本集对红外目标在可见光基准图像中的匹配过程进行训练,最终形成智能化景象匹配识别器。实验结果表明,与常用的景象匹配算法相比,提出方法不仅具有更高的匹配精度和速度,而且鲁棒性好,能抵抗实测图像的旋转几何畸变。     

基于相对距离的密度聚类算法*

首先介绍传统距离计算方法在聚类应用中的不足,并针对这点提出一种基于权重向量的相对距离计算方法。在应用DBSCAN算法的基础上,融入相对距离的计算及k-d树的范围查找的应用。该算法不仅能得到很好的聚类效果,而且消除了数据的度量单位对聚类结果的影响。     

Efficient near-duplicate image detection with a local-based binary representation

Efficient near-duplicate image detection is important for several applications that feature extraction and matching need to be taken online. Most image representations targeting at conventional image retrieval problems are either computationally expensive to extract and match, or limited in robustness. Aiming at this problem, in this paper, we propose an effective and efficient local-based representation method to encode an image as a binary vector, which is called Local-based Binary Representation (LBR). Local regions are extracted densely from the image, and each region is converted to a simple and effective feature describing its texture. A statistical histogram can be calculated over all the local features, and then it is encoded to a binary vector as the holistic image representation. The proposed binary representation jointly utilizes the local region texture and global visual distribution of the image, based on which a similarity measure can be applied to detect near-duplicate image effectively. The binary encoding scheme can not only greatly speed up the online computation, but also reduce memory cost in real applications. In experiments the precision and recall, as well as computational time of the proposed method are compared with other state-of-the-art image representations and LBR shows clear advantages on online near-duplicate image detection and video keyframe detection tasks.     

一种融合多传感器信息的移动图像识别方法

多传感器数据融合作为一种特殊的数据处理手段在图像识别领域得到了较大的重视和发展, 本文提出了一种融合多传感器信息的移动图像识别方法. 首先通过在智能手机端提取带传感器信息的图像局部特征,增强局部特征的辨别能力; 其次改进了随机聚类森林的建立算法,减少了样本图像训练时间;最后使用快 速几何一致性校验对匹配结果进行检查, 保证算法的识别精度.实验结果表明,本文提出的方法能够快速 有效地识别移动图像,并具有较好的鲁棒性,同时与传统的Vocabulary tree 方法进行比较,本文方法的识别速度和精度较优,训练代价较低.     

基于多尺度残差网络的单应估计方法

单应估计是许多计算机视觉任务中的一个基础且重要的步骤。传统单应估计方法基于特征点匹配,难以在弱纹理图像中工作。深度学习已经应用于单应估计以提高其鲁棒性,但现有方法均未考虑到由于物体尺度差异导致的多尺度问题,所以精度受限。针对上述问题,提出了一种用于单应估计的多尺度残差网络。该网络能够提取图像的多尺度特征信息,并使用多尺度特征融合模块对特征进行有效融合,此外还通过估计四角点归一化偏移进一步降低了网络优化难度。实验表明,在MS-COCO数据集上,该方法平均角点误差仅为0.788个像素,达到了亚像素级的精度,并且在99%情况下能够保持较高的精度。由于综合利用了多尺度特征信息且更容易优化,该方法精度显著提高,并具有更强的鲁棒性。     

基于深度语义匹配的法律条文推荐方法

法律条文(简称法条)是司法量刑的主要依据,法律条文的精准推荐,能够辅助提高法律智能判决的质量.目前,主流的法条推荐模型是将有限数量的法条当作类别标签,采用分类的思想,根据法律文书的案例描述将其归类到相关的法条.但是法条作为法律规范的文字表述形式,现有的分类方法简单将其作为类别标签的索引编号,导致对其语义信息利用不足,影响了推荐质量.针对此问题,研究将主流的法条推荐方法从分类模型转化为语义匹配模型,提出了基于深度语义匹配的法条推荐方法(DeepLawRec).该方法包含局部语义匹配模块和全局语义推荐模块,分别设计双向Transformer卷积网络模型和基于回归树的推荐模型,在理解文本序列的同时,关注与法条匹配学习相关的局部语义特征,增强法条推荐的准确率和可解释性.在公开数据集上的实验结果表明,DeepLawRec方法在推荐质量上优于传统的文本分类以及经典的语义匹配方法,并进一步探讨了如何分析和判读推荐结果.     

Image Interpolation Using Enhanced Multiresolution Critical-Point Filters

There are increasing demands for image interpolation in various fields such as virtual reality, computer animation, and video transmission. The critical-point filters (CPF) we have proposed previously enable completely automatic matching of two images. In our previous method, however, it has taken a long time to compute the optimal parameter values by iterative searches. This paper proposes an improved algorithm called the enhanced critical-point filters (ECPF) where the parameters are stably computed using together the inverse mapping from the destination to the source. It takes only a second to match images whose size is 64 × 64. The algorithm is also improved in its precision by directly handling color images while the previous algorithm has taken only the intensity value into account. We apply ECPF to keyframe interpolation of video sequences. We also apply ECPF to interpolating two different views of an object to generate any intermediate views. In this case, there are many constraints that can be used to determine the mapping between the images. We propose a method to use such constraints to improve the accuracy of the mappings. As the results, image-based pseudo-3D models are easily created from a set of views without any special devices.