Evaluation of Stereo Matching Costs on Images with Radiometric Differences

Stereo correspondence methods rely on matching costs for computing the similarity of image locations. We evaluate the insensitivity of different costs for passive binocular stereo methods with respect to radiometric variations of the input images. We consider both pixel-based and window-based variants like the absolute difference, the sampling-insensitive absolute difference, and normalized cross correlation, as well as their zero-mean versions. We also consider filters like LoG, mean, and bilateral background subtraction (BilSub) and nonparametric measures like Rank, SoftRank, Census, and Ordinal. Finally, hierarchical mutual information (HMI) is considered as pixelwise cost. Using stereo data sets with ground-truth disparities taken under controlled changes of exposure and lighting, we evaluate the costs with a local, a semiglobal, and a global stereo method. We measure the performance of all costs in the presence of simulated and real radiometric differences, including exposure differences, vignetting, varying lighting, and noise. Overall, the ranking of methods across all data sets and experiments appears to be consistent. Among the best costs are BilSub, which performs consistently very well for low radiometric differences; HMI, which is slightly better as pixelwise matching cost in some cases and for strong image noise; and Census, which showed the best and most robust overall performance.     

Similarity Analysis of Video Sequences Using an Artificial Neural Network

Comparison of video sequences is an important operation in many multimedia information systems. The similarity measure for comparison is typically based on some measure of correlation with the perceptual similarity (or difference) amongst the video sequences or with the similarity (or difference) in some measure of semantics associated with the video sequences. In content-based similarity analysis, the video data are expressed in terms of different features. Similarity matching is then performed by quantifying the feature relationships between the target video and query video shots, with either an individual feature or with a feature combination. In this study, two approaches are proposed for the similarity analysis of video shots. In the first approach, mosaic images are created from video shots, and the similarity analysis is done by determining the similarities amongst the mosaic images. In the second approach, key frames are extracted for each video shot and the similarity amongst video shots is determined by comparing the key frames of the video shots. The features extracted include image histograms, slopes, edges, and wavelets. Both individual features and feature combinations are used in similarity matching using an artificial neural network. The similarity rank of the query video shots is determined based on the values of the coefficients of determination and the mean absolute error. The study reported in this paper shows that the mosaic-based similarity analysis can be expected to yield a more reliable result, whereas the key frame-based similarity analysis could be potentially applied to a wider range of applications. The weighted non-linear feature combination is shown to yield better results than a single feature for video similarity analysis. The coefficient of determination is shown to be a better criterion than the mean absolute error in similarity matching analysis.     

A two-stage cross correlation approach to template matching

Two-stage template matching with sum of absolute differences as the similarity measure has been developed by Vanderburg and Rosenfeld [1], [2]. This correspondence shows the development of two-stage template matching with cross correlation as the similarity measure. The threshold value of the first-stage is derived analytically and its validity is verified experimentally. Considerable speed-up over the one-stage process can be obtained by introducing only a small false dismissal probability.     

Posterior probability measure for image matching

Template matching is one of the principle techniques in visual tracking. Various similarity measures have been developed to find the target in an acquired image by matching with a template. However, mismatching or misidentification may sporadically occur due to the influence of the background pixels included in the designated target model. Taking into account the statistical features of a search region, a novel similarity measure is proposed, which can decrease the interference of the background pixels enclosed in the model. It highlights the significant target features and at the same time reduces the influence of the features shared by both the target and the background. It exhibits an excellent monotonic property and a distinct peak-like distribution. This new measure is also demonstrated to have a direct interpretation of posterior probability and is named as posterior probability measure (PPM). The proposed PPM can be obtained through a pixel-wise computation and exhibits suitability for image matching. The pixel-wise computation also enables a fast measure update after a target region has changed, which results in a new adaptive scaling method for tracking a target with a varying size. Experiments show that it provides a higher precision in the localization and a discriminatory power superior to the existing similarity measures, such as Bhattacharyya coefficient, Kullback–Leibler divergence, and normalized cross correlation. The effectiveness of the adaptive scaling method is demonstrated in experiments.     

基于视频图像Harris角点检测的车辆测速

为了方便快速地进行车速测量，提出了一种用于测量视频图像中车辆速度的方法。该方法取运动角点为特征量，通过选择灰度相关函数为特征匹配函数，实现了车辆在一定帧差内移动距离的测量。与传统方法相比，该方法不仅对设备的限制更小，而且运算速度更快。     

角点测度在月球表面多光谱图像融合中的应用

根据Harris角点检测原理,提出角点测度的概念,并以角点测度响应值作为高频图像融合系数的选择依据,进而提出基于图像冗余小波域的角点测度重要中心系数算法。算法首先利用冗余小波变换把多光谱图像分解成小波平面和相似平面,然后利用角点测度响应函数来估计小波平面的角点测度,用基于角点测度响应值的重要中心系数融合规则融合小波平面。对相似平面则采取加权平均的融合规则,最后通过冗余小波逆变换得到融合图像。在实验中,用Clementine月球表面多光谱数据和SPOT5多光谱数据验证了算法的有效性,并和其他方法做了比较,除了基于视觉的主观比较以外,还引入了标准差、熵、清晰度和相关系数等客观评价指标对融合结果进行评价,结果表明,算法有效地保持了原图像的细节特征,如边缘、角点等。     

基于空间纹理相似性的图像角点特征匹配算法

本文针对传统图像角点特征匹配算法的匹配速度慢且准确率低等问题,提出一种基于空间纹理相似性的图像角点特征匹配算法。首先,计算图像目标上角点对应的空间距离矩阵;然后,通过计算图像角点的空间距离矩阵在对应角点邻域LBP特征向量上的瑞利商,将角点在图像灰度特征空间内的度量问题转换为纹理特征空间内幅值的度量问题;最后,根据角点对应的瑞利商的大小,实现不同图像间的角点特征匹配。对不同条件下采集的图像进行角点特征匹配,得到的匹配结果表明本文算法不仅能够很好的适应图像光照、几何变化,得到的匹配正确率较高,同时与传统算法相比本文算法在运行时间上也有大幅度的降低,当处理特征数量较小时平均降低48ms,而匹配特征数量较多时能够降低2408ms。     

Measures Based on Fuzzy Similarity for Stereo Matching of Color Images

In this article, new local matching measures, based on fuzzy similarity, for stereo matching of color images are proposed and evaluated. By formulating the individual similarity between a pair of pixels as a conjunction of the similarities of the respective color components, the field programmable gate array implementation problem becomes more computationally tractable, since the word-length of the numbers carrying the similarity information can be reduced compared to standard techniques (Sum of absolute differences and Sum of squared differences). It is also shown that combining information about color and local image structure (horizontal gradient component) in the matching measure and using a multi-scale measure is advantageous compared to using standard measures, in terms of percentage of correct matches and MSE for the erroneous matches. However, these improvements come at the prize of more complex implementations. Since the techniques are window-based, they share the typical drawbacks associated with other techniques of this type. This is expected, since the focus is on developing new local matching measure without addressing issues like post-processing of the resulting disparity maps. The techniques have been tested on two stereo color image pairs.     

遥感图像配准中相似性测度的比较和分析

相似性测度是图像配准中的重要部分之一。本文从遥感图像配准的相似性测度中选取了互信息、相关系数和差方和等三种具有代表性的相似性测度,从计算时间、锐度、对噪声的容忍性以及对多源图像配准的影响等方面,通过实验对它们的性能进行了比较和分析。实验结果表明,不同测度具有不同的有效性和适用范围。     

基于色彩相似度的自适应立体匹配

提出了一种结合权值矩阵和相似性系数矩阵构造的区域匹配方法。该方法首先运用色彩相似性和距离临近性对窗内的每一点相对于待匹配点的自适应权值进行分配,得到一个权值矩阵,为了提高在视差不连续区域的匹配精度,使用了边界点矩阵来降低相对应点的权值。然后在RGB色彩空间中根据待匹配点和对应点的匹配窗内的每一点的颜色绝对差值和来自适应分配相似性系数矩阵。最后利用上述方法对Middlebury网站上提供的四幅立体图像对Tsukuba、Venus、Teddy和Cones进行了实验,总体正确率分别达到了91.82%、96.19%、76.6%和86.9%。     

基于二维Gabor小波变换的角点匹配算法

图像配准研究的核心问题在于提高配准的速度和精度,而图像配准的结果主要取决于特征的匹配精度。为了提高特征匹配精度,本文提出了一种基于二维Gabor小波变换的角点匹配算法。该算法首先采用改进的Harris角点检测方法提取角点,得到角点位置的坐标,利用多个二维Gabor小波模板对参考图像和待配准图像进行滤波,从滤波图像中提取角点坐标处的复Gabor小波系数,并以此作为角点的特征描述,然后引入两种相似性度量因子对角点进行匹配。通过对不同图像进行大量的实验,该算法在选择合适的参数,同时采用最长公共子序列度量因子的情况下,能成功提取较多的同名点对,并且能够取得较高的匹配率。     

基于抽象匹配流的彩色图像配准模型

在抽象匹配流框架下,构造能够克服大色差问题的彩色图像配准模型.该模型中,数据项采用互相关函数作为2幅图像间的相似性度量,以解决大色差问题;正则项采用各向异性扩散滤波器约束图像演化,从而实现在演化过程中对图像特征的有效保持.扩散滤波器中的扩散系数定义为关于彩色结构张量的函数,以使图像演化能够综合各通道信息,解决了各通道所得位移场不一致而引起的色彩混迭问题.实验结果表明,文中模型对具有大色差的彩色图像能够实现有效配准.     

基于Laplace的LSSIM图像配准算法

提出一种基于Laplace变换的图像配准算法. 首先利用经典的角点检测算法提取待匹配图像的特征点或角点; 其次利用相位相关法估算出两幅图像的重叠区域, 以缩小匹配范围; 然后对角点邻域模板区域施行Laplace变换; 最后利用基于改进的SSIM (结构相似性)作为相似性度量准则建立特征点之间的匹配关系. 实验结果表明, 该方法可以很好的完成特征点匹配, 匹配点对充足且具有很高的准确率, 而且对亮度差异具有一定的鲁棒性, 从而保证图像配准精度.     

基于加权相似性的MCCNN训练集选择方法

为解决MCCNN网络立体匹配的训练数据集选择问题,研究一种基于相关性比较、余弦相似性和结构相似性的加权度量选择方法,通过实验确定三者的加权系数,使用三者的加权值衡量训练集与待匹配图像数据分布的互相似性、训练集本身的自相似性,以互相似性和自相似性加和值最高的对应数据集作为选择的训练集.通过InStereo2k图像和实拍图...     

基于神经网络的边缘强度互相关匹配可信度分析

匹配可信度是分析图象匹配质量的主要指标,针对归一化边缘强度互相关匹配算法,研究了基于神经网络的匹配可信度判别方法,即以参考图与若干实时图的匹配实验结果作为训练用样本数据,然后利用ＢＰ网络进行训练,再将训练后的网络用于匹配可信度的判别。通过实际卫片与航片图象对的匹配实验,证实了该方法的有效性。     

基于互相关边界特性和图像积分的快速模板匹配算法

基于归一化算法求解相似度原理,提出了综合利用互相关的边界条件和图像积分计算相似度的快速算法,在不降低匹配精度的前提下较大地提高了匹配速度。计算相似度时,归一化算法需要计算各位置的自相关值和互相关值,本算法先只计算自相关值,再利用Holder不等式原理,结合给定的边界阈值,剔除不满足条件的位置,减少其对应的互相关值计算。应用图像的积分进行匹配在于整个图像的积分可以在匹配之前进行计算,而在匹配过程中每一个子区域的自相关可以通过图像积分快速求得。本算法已在焊线机芯片识别系统中应用,结果表明该算法匹配的速度快而又不降低匹配精度,具有实际应用价值。     

基于多传感器图像融合技术的鲁棒景象匹配算法

将多个传感器获取的具有不同大小、不同分辨力和不同信噪比的实时图基于Kalm an滤波的方法进行融合,以提高实时图的性能;搜索融合后的实时图在基准图中的位置达到目标定位之目的,即进行景象匹配。在景象匹配过程中,选用归一化互相关系数作为相似性度量。多组实验与分析表明:所介绍的基于多传感器图像融合技术的景象匹配算法可以有效地解决实时图存在部分遮挡、灰度与对比度变化以及复杂噪声干扰等影响下的景象匹配问题。     

结合双向相似性变换的重定向图像质量评价

目的 显示设备的多样化使得图像重定向的作用日益凸显。不同的重定向方法产生不同视觉感受的重定向图像,而如何评价重定向图像的质量,优化重定向算法是当前研究的热点与难点,为此,提出一种结合双向相似性变换的重定向图像质量评价方法。方法 首先对原始图像和重定向图像进行像素点双向匹配,利用网格顶点坐标对计算前向变换矩阵和后向变换矩阵。然后由相似性变换矩阵与标准变换矩阵间的距离得到重定向图像的几何失真。由网格面积缺失得到重定向图像的信息损失。最后结合网格的显著性,融合前向匹配与后向匹配的几何失真和信息损失得到重定向图像的质量。结果 该方法在RetargetMe和CUHK数据库上的KRCC（Kendall rank correlation coefficient）和SROCC（Spearman rank-order correlation coefficient）性能分别达到了0.46和0.71,较现有方法有较大提升。在前向匹配与后向匹配测试中,双向匹配的测试结果优于单向匹配。结论 本文方法将图像的重定向处理看做相似性变换过程。实验结果表明,从相似性变换矩阵中提取的相关特征能够较精确度量重定向图像的几何失真,而由此引发的网格面积缺失也能准确反映出重定向图像的信息损失。另外,采用双向匹配机制一定程度上减少了像素匹配误差对实验结果的影响,有效提升了重定向图像质量预测的准确性。该方法对重定向图像的质量评价效果好,适用于重定向图像的质量预测及算法优化。     

基于Harris角点和SIFT描述符的高分辨率遥感影像匹配算法

影像匹配是诸多遥感影像处理和影像分析的一个关键环节。传统基于角点的灰度相关匹配算法由于不具备旋转不变性而需要人工干预进行粗匹配,无法实现自动化。SIFT(scale invariant feature transform)算法能很好地解决图像旋转、缩放等问题,但是对于几何结构特征更加清晰、纹理信息更加丰富的高分辨率遥感影像而言,该算法消耗内存多、运算速度慢的问题非常突出。将两者结合,提出基于Harris角点和SIFT描述符的影像匹配算法。实验结果表明,相比SIFT算法,该算法大量缩减了运算时间,同时保留了SIFT描述符的旋转不变性和对光照变化的适应性,克服了灰度相关算法无法实现全自动的缺点,在高分辨率遥感影像匹配上效果较好。