【24h】

Fast Stereo Matching Using Block Similarity

机译:使用块相似度的快速立体声匹配

获取原文
获取原文并翻译 | 示例

摘要

A lot of engineering applications related to computer vision use stereo vision algorithms and they usually require fast processing capability near to real time. In this paper, we present a new technique of area-based stereo matching algorithm which provides faster processing capability by using block-based matching. Our algorithm employs block-based matching followed by pixel-based matching. By applying block-based matching hierarchically, representative disparity values are assigned to each block. With these rough results, dense disparity map is acquired under very short search range. This procedure can reduce processing time greatly. We test our matching algorithms for various types of images, and shall show good performance of our stereo matching algorithm in both running time and correctness' aspect.
机译:许多与计算机视觉相关的工程应用程序都使用立体视觉算法,它们通常需要接近实时的快速处理能力。在本文中,我们提出了一种基于区域的立体声匹配算法的新技术,该技术通过使用基于块的匹配提供了更快的处理能力。我们的算法采用基于块的匹配,然后再进行基于像素的匹配。通过分层地应用基于块的匹配,将代表性视差值分配给每个块。通过这些粗略的结果,可以在非常短的搜索范围内获得密集的视差图。该程序可以大大减少处理时间。我们针对各种类型的图像测试了我们的匹配算法,并且应该在运行时间和正确性方面都表现出立体匹配算法的良好性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号