摘要
针对双目视觉重建方法在三维重建过程中步骤繁琐与重建效果不理想的问题,文中提出了一种基于双视角下可见外壳的三维重建方法。该方法主要借助平面镜成像原理,在不同视角拍摄包含物体与其成像的两幅图像,并根据成像原理确定物体之间的位置关系。通过相机与周围物体的运动关系求得相机参数,实现相机自标定。然后根据阈值分割及区域生长算法处理边缘信息得到目标轮廓,通过可见外壳方法计算侧影轮廓线,连接轮廓线形成重建模型。该方法通过自适应可见外壳种子体素生长在不了解物体先验知识的情况下完成重建。文中对比文献"基于图像轮廓的三维重建方法"中的单视角重建物体方法,提出了双视角重建物体模型方法。实验结果表明,双视角重建方法简单实用,生成的三维模型准确真实。
For the problem that binocular vision reconstruction method is complicated and its reconstruction effect is not ideal in the 3D reconstruction process,put forward a 3D reconstruction method for visible hull based on double perspectives. The method mainly uses mirror imaging principle,taking tw o pictures in different perspectives containing tw o images of object and its image,then determine the location relationship by imaging principle. It uses the mobile relationship betw een cameras and objects to get the cameras parameters and achieve self calibration method. Then use the threshold segmentation and region grow ing algorithm for edge information processing to get the target contour,computing silhouette contour line with visible shell method and connecting contour lines to form a reconstruction model. The 3D object reconstruction method uses seed body of grain growth to complete the adaptive visible hull. Compared with the literature"3D reconstruction method based on image silhouette",put forward a 3D reconstruction method based on double perspectives. The experimental results show that this reconstruction method is simple and practical,and the model generated is accurate and real.
出处
《计算机技术与发展》
2015年第3期50-53,共4页
Computer Technology and Development
基金
辽宁省自然科学基金(20102163)
沈阳市科技基金项目(F13-316-1-38)
关键词
相机自标定
图像预处理
侧影轮廓线
可视外壳
三维重建
camera self calibration
image preprocessing
silhouette contour line
visible hull
3D reconstruction