摘要
提出一种交互式结构光三维重建系统,在传统的结构光视觉的基础上,增加交互能力。通过人机交互的形式,对三维重建后的物体进行评价,对置信度较低的重建结果,返回原始的二维图像,利用人的知识对获取的结构光图像进行人工干预,然后再进行重建。对于部分物体还利用神经网络对系统进行自学习。实践结果表明,能够扩展结构光视觉系统的应用范围,对于因反射原因而用传统结构光方法难以获取的物体,也可以获得较好的结果。
This paper presents an interactive constucted lighting vision system. The interactive ability is added basing on the traditional constructed vision. By the human-computer interactive method, it can evaluate the 3D reconstructed result of the scanned object. To the unsatisfied result, it returnes the constructed lighting image and revise it with the help of original image. Then this revised constructed lighting image can be reconstructed repeatedly. To some objects, it uses the neural network to do the self-learning for the system. The experiments show that this method can extend the application of the classical constructed lighting vision. To some objects, which are very difficult to get the surface data because of reflecting reason, our method can obtain good result.
出处
《计算机与数字工程》
2007年第9期52-54,共3页
Computer & Digital Engineering
基金
国家自然科学基金(50175033)资助
关键词
结构光
三维重建
人机交互
constructed lighting, 3d-reconstruction, human-computer interactive method
