对立体视觉中对称性假设的检验

指出了基于立体视觉研究中的隐性对称性假设,即认为匹配的选取与哪只眼观察哪一图像无关,而仅取决于两眼映像特征的相似性,因此交换左眼和右眼图像这样一种实验操作,被认为提供了一种非常简便有效的技术,用于鉴定双眼性视差和其他不同于双眼性视差的因素对立体深度的贡献.采用2:2直线型常规立体视构型,对比了左右眼图像交换前后的匹配结果,发现匹配融合产生中间第3根直线的两根直线在图像交换前后会发生改变,并导致中间直线的深度在图像交换前后不发生颠倒.实验结果证明:立体视觉研究中的隐性对称性假设不成立;交换左眼和右眼图像后深度是否颠倒,并不是判断深度是来源于匹配视差还是其他因素的依据.

Testing of the Symmetry Hypothesis in Stereovision

Based on the implicit symmetry hypothesis in the current theories of stereovision , that is, the matches chosen should not depend on the eye that receives a given image but only on the similarity of the images in the two eyes , the experimental manipulation of interchanging the images in the left and right eyes was considered to be a very simple but effective technique to identify the contributions of variables other than binocular disparity to stereoscopic depth .By adopting the two-to-two linetype conventional stereoscopic configuration , the matching results before and after the interchange of the images in the two eyes were compared , and it was found that the two lines matched to produce the middle third line would change and the depth of the middle line would not reverse when the images in two eyes were interchanged .Therefore, the implicit symmetry hypothesis assumed by current theories of stereovision was not supported , and interchanging the images in two eyes cannot be relied on to identify the contributions of variables other than binocular disparity to stereoscopic depth .