Gradient Based Dense Stereo Matching
Gradient Based Dense Stereo Matching
复制标题
基于梯度的密集立体匹配
DOI：
--
复制
发表时间：
2004
期刊：
影响因子：
--
通讯作者：
J. Borgosz
中科院分区：
文献类型：
--
作者：
T. Twardowski;B. Cyganek;J. Borgosz
本文研究了基于图像梯度的图像匹配技术。强度信号的梯度在人类视觉系统和计算机视觉中起着非常重要的作用。自从计算机图像处理开始以来，一直使用梯度形式的特征进行匹配。然而，匹配过程的第二部分是匹配标准的选择，称为比较度量。本文的主要贡献是对梯度比较度量进行了综述，最后提出了我们认为更适合于梯度立体匹配的两种新度量。文中给出了许多实验的结果，以验证我们的方法。
This paper contains a study on image matching techniques based on image gradients. Gradients of the intensity signal play the very important role in the human visual system and also in computer vision. Since the onset of computer image processing, features, in a form of gradients, have been used for matching. However, the second part of a matching process is a choice of the matching criterion, called a comparison measure. The main contribution of this paper is a survey of the gradient comparison measures, ended with a proposition of the two new measures that we believe are more appropriate for gradient stereo matching. The paper presents results of many experiments performed to validate our methods.