自动对焦
内窥镜
计算机视觉
计算机科学
缩放
人工智能
视野
镜头(地质)
折反射系统
光学
物理
放射科
光学(聚焦)
医学
作者
Wei Zhang,Yi-Tao Jin,Xin Guo,Jinhui Su,Suping You
标识
DOI:10.1364/josaa.33.001970
摘要
A traditional capsule endoscope can only take 2D images, and most of the images are not clear enough to be used for diagnosing. A 3D capsule endoscope can help doctors make a quicker and more accurate diagnosis. However, blurred images negatively affect reconstruction accuracy. A compact, autofocus capsule endoscope system is designed in this study. Using a liquid lens, the system can be electronically controlled to autofocus, and without any moving elements. The depth of field of the system is in the 3-100 mm range and its field of view is about 110°. The images captured by this optical system are much clearer than those taken by a traditional capsule endoscope. A 3D reconstruction algorithm is presented to adapt to the zooming function of our proposed system. Simulations and experiments have shown that more feature points can be correctly matched and a higher reconstruction accuracy can be achieved by this strategy.
科研通智能强力驱动
Strongly Powered by AbleSci AI