Secrets of adaptive support weight techniques for local stereo matching

In recent years, local stereo matching algorithms have again become very popular in the stereo community. This is mainly due to the introduction of adaptive support weight algorithms that can for the first time produce results that are on par with global stereo methods. The crux in these adaptive support weight methods is to assign an individual weight to each pixel within the support window. Adaptive support weight algorithms differ mainly in the manner in which this weight computation is carried out.In this paper we present an extensive evaluation study. We evaluate the performance of various methods for computing adaptive support weights including the original bilateral filter-based weights, as well as more recent approaches based on geodesic distances or on the guided filter. To obtain reliable findings, we test these different weight functions on a large set of 35 ground truth disparity pairs. We have implemented all approaches on the GPU, which allows for a fair comparison of run time on modern hardware platforms. Apart from the standard local matching using fronto-parallel windows, we also embed the competing weight functions into the recent PatchMatch Stereo approach, which uses slanted sub-pixel windows and represents a state-of-the-art local algorithm. In the final part of the paper, we aim at shedding light on general points of adaptive support weight matching, which, for example, includes a discussion about symmetric versus asymmetric support weight approaches.     

根据灰度值信息自适应窗口的半全局匹配

目的 立体匹配算法是立体视觉研究的关键点,算法的匹配精度和速度直接影响3维重建的效果。对于传统立体匹配算法来说,弱纹理区域、视差深度不连续区域和被遮挡区域的匹配精度依旧不理想,为此选择具有全局匹配算法和局部匹配算法部分优点、性能介于两种算法之间、且鲁棒性强的半全局立体匹配算法作为研究内容,提出自适应窗口与半全局立体匹配算法相结合的改进方向。方法 以通过AD（absolute difference）算法求匹配代价的半全局立体匹配算法为基础,首先改变算法匹配代价的计算方式,研究窗口大小对算法性能的影响,然后加入自适应窗口算法,研究自适应窗口对算法性能的影响,最后对改进算法进行算法性能评价与比较。结果 实验结果表明,匹配窗口的选择能够影响匹配算法性能、提高算法的适用范围,自适应窗口的加入能够提高算法匹配精度特别是深度不连续区域的匹配精度,并有效降低算法运行时间,对Cones测试图像集,改进的算法较改进前误匹配率在3个测试区域平均减少2.29%;对于所有测试图像集,算法运行时间较加入自适应窗口前平均减少28.5%。结论 加入自适应窗口的半全局立体匹配算法具有更优的算法性能,能够根据应用场景调节算法匹配精度和匹配速度。     

带有时变非线性预处理的立体声回波消除方法研究

就立体声回波消除问题提出了两种新的信号非线性预处理方法，并给出了相应的自适应算法。新的预处理方法比Benesty(1997)及Joncour(1998)的方法对信号的非线性处理部分要少，因而对语音信号的质量影响有所下降。仿真结果表明，新的预处理方法与其实现算法相结合应用于立体声回波消除时，比Benesty(1997)及Joncour(1998) 所提方法的效果更好。     

A fast stereo matching algorithm suitable for embedded real-time systems

In this paper, the challenge of fast stereo matching for embedded systems is tackled. Limited resources, e.g. memory and processing power, and most importantly real-time capability on embedded systems for robotic applications, do not permit the use of most sophisticated stereo matching approaches. The strengths and weaknesses of different matching approaches have been analyzed and a well-suited solution has been found in a Census-based stereo matching algorithm. The novelty of the algorithm used is the explicit adaption and optimization of the well-known Census transform in respect to embedded real-time systems in software. The most important change in comparison with the classic Census transform is the usage of a sparse Census mask which halves the processing time with nearly unchanged matching quality. This is due the fact that large sparse Census masks perform better than small dense masks with the same processing effort. The evidence of this assumption is given by the results of experiments with different mask sizes. Another contribution of this work is the presentation of a complete stereo matching system with its correlation-based core algorithm, the detailed analysis and evaluation of the results, and the optimized high speed realization on different embedded and PC platforms. The algorithm handles difficult areas for stereo matching, such as areas with low texture, very well in comparison to state-of-the-art real-time methods. It can successfully eliminate false positives to provide reliable 3D data. The system is robust, easy to parameterize and offers high flexibility. It also achieves high performance on several, including resource-limited, systems without losing the good quality of stereo matching. A detailed performance analysis of the algorithm is given for optimized reference implementations on various commercial of the shelf (COTS) platforms, e.g. a PC, a DSP and a GPU, reaching a frame rate of up to 75 fps for 640 × 480 images and 50 disparities. The matching quality and processing time is compared to other algorithms on the Middlebury stereo evaluation website reaching a middle quality and top performance rank. Additional evaluation is done by comparing the results with a very fast and well-known sum of absolute differences algorithm using several Middlebury datasets and real-world scenarios.     

A Database and Evaluation Methodology for Optical Flow

The quantitative evaluation of optical flow algorithms by Barron et al. (1994) led to significant advances in performance. The challenges for optical flow algorithms today go beyond the datasets and evaluation methods proposed in that paper. Instead, they center on problems associated with complex natural scenes, including nonrigid motion, real sensor noise, and motion discontinuities. We propose a new set of benchmarks and evaluation methods for the next generation of optical flow algorithms. To that end, we contribute four types of data to test different aspects of optical flow algorithms: (1) sequences with nonrigid motion where the ground-truth flow is determined by tracking hidden fluorescent texture, (2) realistic synthetic sequences, (3) high frame-rate video used to study interpolation error, and (4) modified stereo sequences of static scenes. In addition to the average angular error used by Barron et al., we compute the absolute flow endpoint error, measures for frame interpolation error, improved statistics, and results at motion discontinuities and in textureless regions. In October 2007, we published the performance of several well-known methods on a preliminary version of our data to establish the current state of the art. We also made the data freely available on the web at . Subsequently a number of researchers have uploaded their results to our website and published papers using the data. A significant improvement in performance has already been achieved. In this paper we analyze the results obtained to date and draw a large number of conclusions from them.     

基于像元集的置信传递立体匹配

为了提高立体匹配效率和克服处理区域的视差跳跃,提出了一种基于像元集的置信传递立体匹配方法。该方法首先以像素为基元,利用层次置信传递算法得到较为准确的初始视差;然后依次根据颜色和初始视差对参考图像进行分割,再利用分裂合并策略对分割后的像元集进行平面拟合,以消除颜色分割错误对匹配造成的影响;最后在拟合后的像元集空间,利用标准置信传递优化算法得到最终解。采用国际标准图像进行测试的实验结果表明,该方法的匹配效率和精度优于同类方法。     

基于PatchMatch的半全局高效双目立体匹配算法

近年来双目立体匹配技术发展迅速,高精度、高分辨率、大视差的应用需求无疑对该技术的计算效率提出了更高的要求。由于传统立体匹配算法固有的计算复杂度正比于视差范围,已经难以满足高分辨率、大视差的应用场景。因此,从计算复杂度、匹配精度、匹配原理等多方面综合考虑,提出了一种基于PatchMatch的半全局双目立体匹配算法,在路径代价计算过程中使用空间传播机制,将可能的视差由整个视差范围降低为t个候选视差(t远远小于视差范围),显著减少了候选视差的数量,大幅提高了半全局算法的计算效率。对KITTI2015数据集的评估结果表明,该算法以5.81%的错误匹配率和20.2 s的匹配时间实现了准确性和实时性的明显提高。因此,作为传统立体匹配改进算法,该设计可以为大视差双目立体匹配系统提供高效的解决方案。     

Accurate 3D-vision-based obstacle detection for an autonomous train

In this paper we present a 3D-vision based obstacle detection system for an autonomously operating train in open terrain environments. The system produces dense depth data in real-time from a stereo camera system with a baseline of 1.4 m to fulfill accuracy requirements for reliable obstacle detection 80 m ahead. On an existing high speed stereo engine, several modifications have been applied to significantly improve the overall performance of the system. Hierarchical stereo matching and slanted correlation masks increased the quality of the depth data in a way that the obstacle detection rate increased from 89.4% to 97.75% while the false positive detection rate could be kept as low as 0.25%. The evaluation results have been obtained from extensive real-world test data. An additional stereo matching speed-up of factor 2.15 was achieved and the overall latency of obstacle detection is considerably faster than 300 ms.     

Local algorithms on dense two-frame stereo matching

Stereo matching is a fundamental and crucial problem in computer vision. In the last decades, many researchers have been working on it and made great progress. Generally stereo algorithms can be classified into local methods and global methods. In this paper, the challenges of stereo matching are first introduced, and then we focus on local approaches which have simpler structures and higher efficiency than global ones. Local algorithms generally perform four steps: cost computation, cost aggregation, disparity computation and disparity refinement. Every step is deeply investigated, and most work focuses on cost aggregation. We studied most of the classical local methods and divide them into several classes. The classification well illustrates the development history of local stereo correspondence and shows the essence of local matching along with its important and difficult points. At the end we give the future development trend of local methods.     

三维视觉前沿进展

在自动驾驶、机器人、数字城市以及虚拟/混合现实等应用的驱动下,三维视觉得到了广泛的关注。三维视觉研究主要围绕深度图像获取、视觉定位与制图、三维建模及三维理解等任务而展开。本文围绕上述三维视觉任务,对国内外研究进展进行了综合评述和对比分析。首先,针对深度图像获取任务,从非端到端立体匹配、端到端立体匹配及无监督立体匹配3个方面对立体匹配研究进展进行了回顾,从深度回归网络和深度补全网络两个方面对单目深度估计研究进展进行了回顾。其次,针对视觉定位与制图任务,从端到端视觉定位和非端到端视觉定位两个方面对大场景下的视觉定位研究进展进行了回顾,并从视觉同步定位与地图构建和融合其他传感器的同步定位与地图构建两个方面对同步定位与地图构建的研究进展进行了回顾。再次,针对三维建模任务,从深度三维表征学习、深度三维生成模型、结构化表征学习与生成模型以及基于深度学习的三维重建等4个方面对三维几何建模研究进展进行了回顾,并从多视RGB重建、单深度相机和多深度相机方法以及单视图RGB方法等3个方面对人体动态建模研究进展进行了回顾。最后,针对三维理解任务,从点云语义分割和点云实例分割两个方面对点云语义理解研究进展进行了回顾。在此基础上,给出了三维视觉研究的未来发展趋势,旨在为相关研究者提供参考。     

Stereo for Image-Based Rendering using Image Over-Segmentation

In this paper, we propose a stereo method specifically designed for image-based rendering. For effective image-based rendering, the interpolated views need only be visually plausible. The implication is that the extracted depths do not need to be correct, as long as the recovered views appear to be correct. Our stereo algorithm relies on over-segmenting the source images. Computing match values over entire segments rather than single pixels provides robustness to noise and intensity bias. Color-based segmentation also helps to more precisely delineate object boundaries, which is important for reducing boundary artifacts in synthesized views. The depths of the segments for each image are computed using loopy belief propagation within a Markov Random Field framework. Neighboring MRFs are used for occlusion reasoning and ensuring that neighboring depth maps are consistent. We tested our stereo algorithm on several stereo pairs from the Middlebury data set, and show rendering results based on two of these data sets. We also show results for video-based rendering.     

Efficient Dense Stereo with Occlusions for New View-Synthesis by Four-State Dynamic Programming

A new algorithm is proposed for efficient stereo and novel view synthesis. Given the video streams acquired by two synchronized cameras the proposed algorithm synthesises images from a virtual camera in arbitrary position near the physical cameras. The new technique is based on an improved, dynamic-programming, stereo algorithm for efficient novel view generation. The two main contributions of this paper are: (i) a new four state matching graph for dense stereo dynamic programming, that supports accurate occlusion labelling; (ii) a compact geometric derivation for novel view synthesis by direct projection of the minimum cost surface. Furthermore, the paper presents an algorithm for the temporal maintenance of a background model to enhance the rendering of occlusions and reduce temporal artefacts (flicker); and a cost aggregation algorithm that acts directly in the three-dimensional matching cost space. The proposed algorithm has been designed to work with input images with large disparity range, a common practical situation. The enhanced occlusion handling capabilities of the new dynamic programming algorithm are evaluated against those of the most powerful state-of-the-art dynamic programming and graph-cut techniques. Four-state DP is also evaluated against the disparity-based Middlebury error metrics and its performance found to be amongst the best of the efficient algorithms. A number of examples demonstrate the robustness of four-state DP to artefacts in stereo video streams. This includes demonstrations of cyclopean view synthesis in extended conversational sequences, synthesis from a freely translating virtual camera and, finally, basic 3D scene editing.     

An experimental comparison of min-cut/max-flow algorithms for energy minimization in vision

Minimum cut/maximum flow algorithms on graphs have emerged as an increasingly useful tool for exactor approximate energy minimization in low-level vision. The combinatorial optimization literature provides many min-cut/max-flow algorithms with different polynomial time complexity. Their practical efficiency, however, has to date been studied mainly outside the scope of computer vision. The goal of this paper is to provide an experimental comparison of the efficiency of min-cut/max flow algorithms for applications in vision. We compare the running times of several standard algorithms, as well as a new algorithm that we have recently developed. The algorithms we study include both Goldberg-Tarjan style "push -relabel" methods and algorithms based on Ford-Fulkerson style "augmenting paths." We benchmark these algorithms on a number of typical graphs in the contexts of image restoration, stereo, and segmentation. In many cases, our new algorithm works several times faster than any of the other methods, making near real-time performance possible. An implementation of our max-flow/min-cut algorithm is available upon request for research purposes.     

Efficient Belief Propagation for Early Vision

Markov random field models provide a robust and unified framework for early vision problems such as stereo and image restoration. Inference algorithms based on graph cuts and belief propagation have been found to yield accurate results, but despite recent advances are often too slow for practical use. In this paper we present some algorithmic techniques that substantially improve the running time of the loopy belief propagation approach. One of the techniques reduces the complexity of the inference algorithm to be linear rather than quadratic in the number of possible labels for each pixel, which is important for problems such as image restoration that have a large label set. Another technique speeds up and reduces the memory requirements of belief propagation on grid graphs. A third technique is a multi-grid method that makes it possible to obtain good results with a small fixed number of message passing iterations, independent of the size of the input images. Taken together these techniques speed up the standard algorithm by several orders of magnitude. In practice we obtain results that are as accurate as those of other global methods (e.g., using the Middlebury stereo benchmark) while being nearly as fast as purely local methods.     

Fast algorithms for frequent itemset mining using FP-trees

Efficient algorithms for mining frequent itemsets are crucial for mining association rules as well as for many other data mining tasks. Methods for mining frequent itemsets have been implemented using a prefix-tree structure, known as an FP-tree, for storing compressed information about frequent itemsets. Numerous experimental results have demonstrated that these algorithms perform extremely well. In this paper, we present a novel FP-array technique that greatly reduces the need to traverse FP-trees, thus obtaining significantly improved performance for FP-tree-based algorithms. Our technique works especially well for sparse data sets. Furthermore, we present new algorithms for mining all, maximal, and closed frequent itemsets. Our algorithms use the FP-tree data structure in combination with the FP-array technique efficiently and incorporate various optimization techniques. We also present experimental results comparing our methods with existing algorithms. The results show that our methods are the fastest for many cases. Even though the algorithms consume much memory when the data sets are sparse, they are still the fastest ones when the minimum support is low. Moreover, they are always among the fastest algorithms and consume less memory than other methods when the data sets are dense.     

Real-Time Correlation-Based Stereo Vision with Reduced Border Errors

This paper describes a real-time stereo vision system that is required to support high-level object based tasks in a tele-operated environment. Stereo vision is computationally expensive, due to having to find corresponding pixels. Correlation is a fast, standard way to solve the correspondence problem. This paper analyses the behaviour of correlation based stereo to find ways to improve its quality while maintaining its real-time suitability. Three methods are suggested. Two of them aim to improve the disparity image especially at depth discontinuities, while one targets the identification of possible errors in general. Results are given on real stereo images with ground truth. A comparison with five standard correlation methods is provided. All proposed algorithms are described in detail and performance issues and optimisation are discussed. Finally, performance results of individual parts of the stereo algorithm are shown, including rectification, filtering andcorrelation using all proposed methods. The implemented system shows that errors of simple stereo correlation, especially in object border regions, can be reduced in real-time using non-specialised computer hardware.     

Joint view synthesis and disparity refinement for stereo matching

Typical stereo algorithms treat disparity estimation and view synthesis as two sequential procedures. In this paper, we consider stereo matching and view synthesis as two complementary components, and present a novel iterative refinement model for joint view synthesis and disparity refinement. To achieve the mutual promotion between view synthesis and disparity refinement, we apply two key strategies, disparity maps fusion and disparity-assisted plane sweep-based rendering (DAPSR). On the one hand, the disparity maps fusion strategy is applied to generate disparity map from synthesized view and input views. This strategy is able to detect and counteract disparity errors caused by potential artifacts from synthesized view. On the other hand, the DAPSR is used for view synthesis and updating, and is able to weaken the interpolation errors caused by outliers in the disparity maps. Experiments onMiddlebury benchmarks demonstrate that by introducing the synthesized view, disparity errors due to large occluded region and large baseline are eliminated effectively and the synthesis quality is greatly improved.     

基于立体视觉的目标检测与轨迹预测研究综述

目前基于立体视觉信息的运动目标识别定位、跟踪及轨迹预测是机器视觉领域的研究热点.通过归纳整理相关文献,从双目立体视觉技术、运动目标检测技术、运动目标轨迹预测技术三个方面对基于立体视觉的运动目标检测及轨迹预测进行了概述,分别阐述了相机标定的常见方法、图像特征提取及立体匹配不同算法的适用场景、各运动目标检测方法的优缺点、常...     

Real-time stereo on GPGPU using progressive multi-resolution adaptive windows

We introduce a new GPGPU-based real-time dense stereo matching algorithm. The algorithm is based on a progressive multi-resolution pipeline which includes background modeling and dense matching with adaptive windows. For applications in which only moving objects are of interest, this approach effectively reduces the overall computation cost quite significantly, and preserves the high definition details. Running on an off-the-shelf commodity graphics card, our implementation achieves a 36 fps stereo matching on 1024 × 768 stereo video with a fine 256 pixel disparity range. This is effectively same as 7200 M disparity evaluations per second. For scenes where the static background assumption holds, our approach outperforms all published alternative algorithms in terms of the speed performance, by a large margin. We envision a number of potential applications such as real-time motion capture, as well as tracking, recognition and identification of moving objects in multi-camera networks.