摘要
针对不含深度信息的单幅RGB图像,提出一种新颖的主平面检测算法。首先利用图像分割技术将图像主体部分提取出来,并检测出该部分区域的轮廓;然后将轮廓拟合成多边形,从多边形的顶点中找到三平面的交点,从而找出平面间的分界线,得到一个粗糙的平面边界;最后利用图像中的边缘信息优化平面边界。
A novel dominant plane detection algorithm is proposed on a single RGB image without depth information.Firstly we use image segmenta tion technology to extract the main part of the image and detect the outline of the area.Secondly fit the outline into a polygon,find the inter section of three planes from the vertices of the polygon,and find the boundary between the planes,then a Rough planar boundary is ob tained.Finally we use the edge information in the image to optimize the planar boundary.
作者
李志明
LI Zhi-ming(College of Computer Science,Sichuan University,Chengdu 610065)
出处
《现代计算机》
2020年第7期79-83,共5页
Modern Computer
