Reconfigurable hardware implementation of a phase-correlation stereoalgorithm

This paper describes the implementation of a stereo-vision system using Field Programmable Gate Arrays (FPGAs). Reconfigurable hardware, including FPGAs, is an attractive platform for implementing vision algorithms due to its ability to exploit parallelism often found in these algorithms, and due to the speed with which applications can be developed as compared to hardware. The system outputs 8-bit, subpixel disparity estimates for 256× 360 pixel images at 30,fps. A local-weighted phase correlation algorithm for stereo disparity [Fleet, D. J.: {Int. Conf. Syst. Man Cybernetics 1:48–54 (1994)] is implemented. Despite the complexity of performing correlations on multiscale, multiorientation phase data, the system runs as much as 300 times faster in hardware than its software implementation. This paper describes the hardware platform used, the algorithm, and the issues encountered during its hardware implementation. Of particular interest is the implementation of multiscale, steerable filters, which are widely used in computer vision algorithms. Several trade-offs (reducing the number of filter orientations from three to two, using fixed-point computation, changing the location of one localized low-pass filter, and using L1 instead of L2 norms) were required to both fit the design into the available hardware and to achieve video-rate processing. Finally, results from the system are given both for synthetic data sets as well as several standard stereo-pair test images.     

Material state remapping in computational solid mechanics

A computational procedure for remapping material state information from one finite element mesh to another is described. The procedure is useful in connection with evolving meshes for inelastic problems, as for example occur in the context of fracture simulation and adaptive mesh refinement. The proposed method is based on weak enforcement of equality between corresponding fields on the two meshes, where piecewise‐constant fields on both meshes are generalized from the quadrature‐point values. The essential algorithmic problem is that of calculating the volume partition of an arbitrary convex region with respect to a covering set of disjoint convex regions. Instead of geometrically resolving the associated intersections, the problem is herein approximated by a constrained optimization problem, which may be readily and efficiently solved computationally. This formulation is a main contribution of the paper. Computational examples are given that illustrate the effectiveness of the proposed procedure. Copyright © 2002 John Wiley & Sons, Ltd.     

The visual control of reaching and grasping: Binocular disparity and motion parallax.

The primary visual sources of depth and size information are binocular cues and motion parallax. Here, the authors determine the efficacy of these cues to control prehension by presenting them in isolation from other visual cues. When only binocular cues were available, reaches showed normal scaling of the transport and grasp components with object distance and size. However, when only motion parallax was available, only the transpont component scaled reliably. No additional increase in scaling was found when both cues were available simultaneously. Therefore, although equivalent information is available from binocular and motion parallax information, the latter may be of relatively limited use for the control of the grasp. Binocular disparity appears selectively important for the control of the grasp. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

一种基于MPEG-2的立体视频编码中的视差匹配快速算法

高效，快速的视差匹配是立体视频处理中的一项关键技术。本文在分析立体图像序列的视差矢量与运动矢量之间的相关性的基础上，提出一种基于MPEG-2的立体视频编码中的视差匹配快速算法。实验结果表明，与全搜索法相比，在保证重建图像质量的前提下，快速算法能显著降低视差估计的计算复杂度。     

基于梯度值的自适应窗口立体图像匹配算法

针对基于区域的立体图像匹配算法支持窗口难以选择,容易出现窗口过大或过小的问题,提出一种新的自适应窗口立体图像匹配算法。该算法利用Sobel梯度算子计算像素梯度值,并根据其梯度值动态地获取具有自适应的支持窗口,然后分别选择相似性测度函数SAD或NCC搜索最佳匹配点,获得视差图。此外,算法在窗口选择过程中进行优化,减少了计算量。实验结果表明,改进后的算法提高了匹配正确率且计算时间缩短了近5%。     

基于多尺度注意力网络的立体匹配方法

为解决网络深度与训练图像块大小耦合问题及进一步提高弱纹理区域及边缘处的匹配精度,提出了一种基于多尺度注意力网络的立体匹配方法.该方法将立体匹配过程分为2个阶段：第1阶段提出了一种成本网络用于计算匹配成本,该网络由基础网络层和缩放层组成.第2阶段提出了一种基于多尺度注意力的视差求精网络,该视差求精网络综合了多种视差线索,并加入多尺度注意力机制进一步提高立体匹配精度.该方法在KITTI 2012、KITTI 2015和SceneFlow数据集上的3像素坏点百分比分别为1.13%,1.87%和2.29%.实验结果表明,与国内外同类方法相比,采用多尺度注意力网络的立体匹配方法在匹配精度上获得了较大的提升,尤其是在弱纹理区域及物体边缘处表现较好.     

基于素质差距的装备保障训练需求分析应用研究*

针对我军新型装备缺乏系统规范的保障训练需求分析的问题,提出了“基于人员素质差距确定训练需求”的思路,并结合案例建立了我军装备保障训练需求分析的仿真模型,经验证,仿真结果科学有效。     

Aging and availability of binocular disparity and pictorial depth cues in 3D graphics contents

Some stereoblind observers do not perceive depth of 3D stimuli that depends on binocular disparity. These individuals, who have no disabilities, comprise over 5% of the general population. In addition, 17–30% of nonstereoblind young and young–middle people do not use disparity information in certain 3D environments, a phenomenon known as pseudo‐stereoblindness. This study aimed to investigate the relationship between aging and the proportion of pseudo‐stereoblindness in the general population. In an experiment, 134 nonstereoblind participants, ranging in age from 17 to 83 years, judged subjective depth of 3D stimuli containing binocular disparity and pictorial depth cues. Results showed that the proportion of pseudo‐stereoblindness among young (17–24 years old) and young–middle aged observers (25–39 years old) was 29%, in both cases. However, the proportion of pseudo‐stereoblind observers increased in older populations: 65% and 82% in the middle (40–54 years old) and senior (55–83 years old) age groups, respectively. These results suggest that a number of people, especially in elderly populations, have trouble perceiving depth from binocular disparity in 3D graphic contents despite their essential ability to use disparity information.     

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。     

立体图像视差线性调整算法

为使立体图像能在特定显示设备上舒适显示,本文提出一种调整立体图像的方法.首先利用立体图像融像区知识,计算不同显示设备舒适显示的水平视差范围;利用匹配的SIFT(scale-invariant features)特征点来估计立体图像的视差范围,建立起立体图像原有视差到舒适显示视差的线性映射;最后利用奇异值分解估计该映射的变换矩阵,通过变换矩阵计算出调整后的图像,消除立体图像垂直视差,并将立体图像的水平视差调整到舒适观看的范围.实验结果表明,本文提出的方法可以解决不同显示设备上立体图像视差调整的问题,得到更好的显示效果.