增强现实系统中虚实结合采用的虚实遮挡处理技术,目前难以处理该技术中真实物体之间存在半遮挡现象的真实场景,为了判断真实物体之间的半遮挡现象,提出了一种基于双目图像物体边缘的半遮挡现象检测方法,该方法通过检测双目图像获得真实物体的边缘信息,提取每一个边缘点对应的若干边缘像素,将两幅图像的边缘像素进行灰度匹配,通过不匹配的边缘像素来判断是否存在真实物体之间的半遮挡现象。实验结果验证了该方法的有效性,相对遮挡处理过程,该方法具有较小的时间开销,并且适用于规则场景和规则对象的半遮档检测。
In order to determine the half occlusion phenomenon between real objects,a binocular image object edge detection method is studies and proposed for half occlusion phenomenon. The real object edge information is obtained by detecting binocular image, extracting a number of edge pixels corresponding to each edge point. The edge pixels in two images are dealt by gray matching and the half occlusion between real objects can be judged by mismatched edge pixel. The experimental results verify the validity of the proposed method. Compared with relative occlusion handling process, the method has small times cost, which is applicable for half occlusion detection of rule scenes and rule object.