一种基于分割的两步立体匹配算法

提出了一种基于分割的两步立体匹配算法。第1步,利用动态规划和左右一致性约束计算地面控制点,将原始图像分割为许多小区域,假定每一个区域为同一个视差,并通过地面控制点计算部分区域的视差值;第2步将剩余未匹配的区域编码为一条染色体,用遗传算法进行优化,得到最终的稠密视差图。选用了一些标准图像对进行测试,实验结果表明,该算法具有很好的性能。     

自适应分层粒子群稠密视差图估计

分析了遗传算法的缺陷,提出了自适应分层粒子群(PSO)立体匹配算法计算稠密视差图。首先采用SIFT(scale invariant feature transform)特征检测和匹配算法准确地确定视差范围;其次根据图像和视差范围的大小分层,建立由粗及细的自适应分层图像金字塔结构,加快搜索速度、减少错误匹配;然后在优化函数中引入能根据匹配窗口大小自动变化的因子来调整灰度项和平滑项数据的权重,并用改进的带变异算子的整数形式的PSO进行优化,避免了遗传算法搜索的盲目性以及容易陷入局部最优的缺陷,更快、更好地找到最优解。最后合成图像以及真实图像的实验结果表明该方法精度较高,速度较快。     

基于多目立体匹配的深度获取方法

提出一种深度获取方法,利用基于颜色分割的多目立体匹配算法,从多个视点图像中提取深度信息。利用mean-shift算法,根据颜色信息分割参考图像,提取图像中的颜色一致性区域,通过局部窗口匹配算法进行多目立体匹配得到多幅初始视差图,根据融合准则将多幅视差图合成为一幅视差图以提高视差图的精度并对视差图进行优化后处理,按照视差与深度的关系,将视差图转化为深度图。该算法能有效处理匹配过程中的遮挡区域,提高匹配精度和视差图的准确度。     

基于网络最小割的分层立体视觉匹配方法

立体匹配是计算机视觉领域中的一个难点问题.为了得到准确的高密度的视差图,本文提出一种基于网络最小割的分层匹配方法.该方法综合运用区域灰度相关法和最小割全局最优搜索策略.首先对原图像对进行两层金字塔分解,在低分辨率的图像中运用网络最小割方法求得全局最优匹配.然后在低分辨率的图像中匹配的像素对的约束下,在原图像对中采用区域灰度相关法进行匹配,得到高密度视差图.这样既缩小匹配时的搜索空间,又保证匹配的可靠性.实验表明,该方法是有效可行的.     

一种基于双向互匹配的视差图算法

提出一种视差图生成算法,利用左图像和右图像双向互匹配的的方法,从图像中获取左-右和右-左的视差图和梯度场,接着采用Winner-Take-All策略,对两幅视差图像进行匹配,得到初始视差图。最后,对视差图上存在的误匹配点进行优化。通过实验验证,该算法能够有效地提高匹配视差图的准确度。     

快速视差范围估计算法及其应用

视差范围估计在立体匹配中非常重要,准确的视差范围能提高立体匹配的精度和速度.为此提出一种基于匹配代价搜索和图像细分的快速视差范围估计算法.该算法将输入图像均匀分成多个图像块,采用匹配代价搜索计算每一图像块的视差,找到视差最大(最小)的图像块,并利用迭代细分规则将该图像块继续分成更小的子块,直至得到稳定的最大(最小)视差;利用匹配代价图对图像块进行可靠性检测,以解决弱纹理块容易误匹配的问题.实验结果表明,文中算法在保持97.3%的平均命中率的同时将立体匹配的平均搜索空间降低了27.7%,比采用传统算法可以得到更准确的视差范围;将该算法应用于立体匹配算法中降低了其平均误匹配率,并将计算时间缩短了20%~45%.     

一种利用动态规划和左右一致性的立体匹配算法

立体匹配是计算机视觉领域研究的一个重要课题,为了得到准确、稠密的视差图,提出了一种利用动态规划和左右一致性的立体匹配算法。该算法首先分别以左、右图像为基元图像,计算各自的视差空间图像,在视差空间图像上利用动态规划,计算得到左视差图和右视差图;然后通过使用左右视差图之间的一致性关系,消除误匹配点,得到较为准确的部分视差图;最后利用视差图的顺序约束关系,给出未匹配视差点的搜索空间计算方法,并利用一种简单有效的方法来计算这些点的视差值。在一些标准立体图像对上所做的实验结果表明,该算法效果良好。     

一种基于改进小生境遗传算法的图像分割方法

利用经典的Otsu算法和基本遗传算法相结合进行图像分割存在有算法效率低、容易提前形成伪解的问题,对于上述问题,提出一种基于改进小生境遗传算法的图像分割算法(IVNGAMS)。算法全局优化了二维Otsu图像分割函数,可以按照个体适应度大小自动控制遗传参数。并通过引入模拟退火算法,进一步提升算法的局部搜索能力。实验结果表明,改进的图像分割方法能更好提升算法的全局搜索能力,能够更加稳定快速的收敛到最佳的分割阈值,并且得到了更好的图像分割效果。     

垂直交叉双向搜索策略的自适应窗口匹配算法

目的 基于区域的局部匹配算法是一种简单高效的立体匹配方法.针对局部算法中窗口的抉择问题,提出了基于垂直交叉双向搜索的自适应窗口匹配算法.方法 该算法考虑到局部区域内灰度值与视差值的相关性,通过垂直交叉双向搜索策略自适应地调节窗口的形状和大小,并获得相应掩码窗口;再利用积分图像计算掩码窗口的匹配代价,获取视差图;最后采用米字投票和双边滤波器两个步骤对视差图进行修复.结果 针对不同图像采用提出的自适应窗口算法,得到了适用于各种图像结构的匹配窗口,相较于原始垂直交叉算法的匹配精度提高了约30％ (Teddy),同时两步骤视差后处理较好地保持了图像边缘.结论 实验结果表明,该算法改善了规则窗口产生的视差边缘扩充问题,在提高视差精度的同时提高了算法鲁棒性.     

基于图像区域分割和置信传播的立体匹配算法

传统基于像素的立体匹配算法误匹配率较高.为解决该问题,提出一种基于图像区域分割和置信传播的匹配算法.采用均值偏移对参考图像进行区域分割,通过自适应权值匹配计算初始视差图,对各分割区域的初始视差用平面模型拟合得到视差平面参数,使用基于区域的改进置信传播算法求得各区域的最优视差平面,从而得到最终视差图.与全局优化的经典置信传播算法和图割算法的对比实验结果表明,该算法能降低低纹理区域和遮挡区域的误匹配率.     

GA-based intelligent digital redesign of fuzzy-model-based controllers

Intelligent digital redesign involves converting a continuous-time fuzzy-model-based controller into an equivalent discrete-time counterpart for the digital control of continuous-time nonlinear systems by using the Takagi-Sugeno (TS) fuzzy models. In this paper, the authors present a new global state-matching intelligent digital redesign method for nonlinear systems by using genetic algorithms (GAs). More precisely, the intelligent digital redesign problem is converted to an equivalent optimization problem, and then GAs are adopted to find a solution. The search space, in which each problem variable is defined for GAs, are systematically obtained by the interval arithmetic operations. The proposed method results in global matching of the states of the analogously controlled system with those of the digitally controlled system while the conventional intelligent digital redesign method does not. The Chen's chaotic system is used as an illustrative example to show the effectiveness and the feasibility of the developed method. The proposed method provides a new approach for the digital redesign of a class of fuzzy-model-based controllers.     

Hybrid symbiotic genetic optimisation for robust edge-based stereo correspondence.

This work proposes a novel algorithm for performing robust feature-based stereo-matching, without the ordering constraint. The calculation of the disparity map is decomposed to a set of disjoint intra-row subproblems, each one having two objectives: the search for a high confidence intra-row matching and the enforcement of figural continuity at the inter-row level. A separate genetic algorithm (GA) is allocated at each epipolar to search the feasible solution space. All GAs evolve parallely in a symbiotic fashion and continuously exchange currently available solution information to enable optimisation of figural continuity. To accelerate the search, we adapt a deterministic solver to seed the GAs and design problem-specific genetic operators for greater efficiency.     

Fast spatio-temporal stereo for intelligent transportation systems

This paper presents a fast approach for matching stereoscopic images acquired by stereo cameras mounted aboard a moving car. The proposed approach exploits the spatio-temporal consistency between consecutive frames in stereo sequences to improve matching results. This means that the matching process at current frame uses the matching results obtained at its preceding one. The preceding frame allows to compute an Initial Disparity Map for the current frame. The initial disparity map is used to derive disparity ranges for each scanline as well as what we call Matching Control Edge Points. Dynamic programming is performed for matching edge points in stereo pairs. The matching control edge points are used to drive the search for an optimal solution in the search plane. This is accomplished by dividing the dynamic programming search space into a number of subspaces depending on the number of the matching control edge points. The proposed approach has been tested both on virtual and real stereo images sequences demonstrating satisfactory performance.     

Linkage Discovery through Data Mining [Research Frontier]

Genetic algorithms (GAs) are extensively adopted in various aspects of data mining, e.g., association rules, clustering, and classification [1, 2, 3]. Instead of applying GAs for data mining, this study addresses linkage discovery, an essential topic in GAs, by using data mining methods. Inspired by natural evolution, GAs utilize selection, crossover, and mutation operations to evolve candidate solutions into global optima [4]. This evolutionary scheme can effectively resolve many search and optimization problems. As the most salient feature of GAs, crossover enables the recombination of good parts of two selected chromosomes, yet, in doing so, may disrupt the collected promising segments.     

基于区域间协同优化的立体匹配算法

提出了一种基于分割区域间协同优化的立体匹配算法. 该算法以图像区域为匹配基元, 利用区域的彩色特征以及相邻区域间应满足的平滑和遮挡关系定义了区域的匹配能量函数, 并引入区域之间的合作竞争机制, 通过协同优化使所定义的匹配能量极小化, 从而得到比较理想的视差结果. 算法首先对参考图像进行分割, 利用相关法得到各分割区域的初始匹配; 然后用平面模型对各区域的视差进行拟合, 得到各区域的视差平面参数; 最后, 基于协同优化的思想, 采用局部优化的方法对各区域的视差平面参数进行迭代优化, 直至得到比较合理的视差图为止. 采用Middlebury test set进行的实验结果表明, 该方法在性能上可以和目前最好的立体匹配算法相媲美, 得到的视差结果接近于真实视差.     

基于协作Hopfield网络的迭代立体匹配算法

针对立体匹配算法中求解能量函数全局最小问题,提出一种基于协作Hopfield网络的迭代立体匹配算法.它采用两个具有相似结构的Hopfield神经网络协作求解匹配问题,两个网络的不同之处是匹配过程中所采用的基准图不同.然后根据左右一致性约束实现两个Hopfield网络之间的协作,从而避免落入局部最小.为加快收敛速度,该算法将视差图的最优搜索问题转换为二值神经网络的迭代收敛过程.利用局部匹配算法的结果预标记初始视差,以设定神经网络初始权重.并根据局部匹配算法中隐含的假定条件,提出了局部匹配算法视差结果的评估准则,以确定各像素的视差搜索范围,从而减少各次迭代过程中状态待确定的神经元个数.实验表明该方法在性能和收敛速度上都要优于传统的Boltzmann机方法.     

基于特征约束及区域相关的体视匹配方法

立体匹配是计算机视觉领域的一个关键问题,同时也是难点问题。为了得到准确的高密度视差图,通过对基于区域和基于特征的体视方法的讨论,综合两种方法的优点,提出了基于边缘特征约束及区域相关的立体匹配算法。该方法首先利用基于特征技术来得到边缘特征点,对边缘特征点再做灰度等区域相关匹配处理,然后在匹配的边缘特征点约束下,对非边缘特征点采用区域相关算法进行匹配,得到整体高密度视差图。这样既缩小了匹配搜索空间,又保证了匹配的可靠性。实验结果表明,该算法具有良好的效果和实用价值。     

基于自适应匹配窗及多特征融合的立体匹配*

针对局部立体匹配方法中存在的匹配窗口大小选择困难、边缘处视差模糊及弱纹理区域、斜面或曲面匹配精度较低等问题,提出基于CIELAB空间下色度分割的自适应窗选取及多特征融合的局部立体匹配算法.首先,在CIELAB空间上对立体图像对进行色度分割,依据同质区域的分布获取初始匹配支持域,同时估计遮挡区域,更新匹配支持域.然后,基于更新后的匹配支持域,采用自适应权值的线性加权多特征融合匹配方法得到初始视差图.最后,利用左右视差一致性检测方法进行误匹配检验,利用基于分割的均值滤波器进行视差优化及细化,得到稠密匹配视差结果.实验表明文中算法有效,匹配精度较高,尤其在弱纹理区域及斜面等情况下匹配效果较好.     

A distributed hierarchical genetic algorithm for efficient optimization and pattern matching

In this paper we propose a new approach in genetic algorithm called distributed hierarchical genetic algorithm (DHGA) for optimization and pattern matching. It is eventually a hybrid technique combining the advantages of both distributed and hierarchical processes in exploring the search space. The search is initially distributed over the space and then in each subspace the algorithm works in a hierarchical way. The entire space is essentially partitioned into a number of subspaces depending on the dimensionality of the space. This is done in order to spread the search process more evenly over the whole space. In each subspace the genetic algorithm is employed for searching and the search process advances from one hypercube to a neighboring hypercube hierarchically depending on the convergence status of the population and the solution obtained so far. The dimension of the hypercube and the resolution of the search space are altered with iterations. Thus the search process passes through variable resolution (coarse-to-fine) search space. Both analytical and empirical studies have been carried out to evaluate the performance between DHGA and distributed conventional GA (DCGA) for different function optimization problems. Further, the performance of the algorithms is demonstrated on problems like pattern matching and object matching with edge map.     

A portable stereo vision system for whole body surface imaging

This paper presents a whole body surface imaging system based on stereo vision technology. We have adopted a compact and economical configuration which involves only four stereo units to image the frontal and rear sides of the body. The success of the system depends on a stereo matching process that can effectively segment the body from the background in addition to recovering sufficient geometric details. For this purpose, we have developed a novel sub-pixel, dense stereo matching algorithm which includes two major phases. In the first phase, the foreground is accurately segmented with the help of a predefined virtual interface in the disparity space image, and a coarse disparity map is generated with block matching. In the second phase, local least squares matching is performed in combination with global optimization within a regularization framework, so as to ensure both accuracy and reliability. Our experimental results show that the system can realistically capture smooth and complete whole body shapes with high accuracy.