基于明暗信息的胶囊内镜图像三维形状恢复

3-d shape recovery from vce Image Based on Shading Information

利用明暗恢复形状技术,将视频胶囊内镜二维图像的明暗信息重构小肠内腔表面的三维场景信息。针对传统的明暗恢复形状技术的约束条件,提出了一种针对胶囊内镜图像的预处理方法,利用预处理后的图像恢复出更为平滑的视觉场景,同时保留被检查对象的结构信息。基于临床数据的实验结果表明,三维形状恢复技术可以增强胶囊内镜视频的可读性,提高小肠疾病的诊断准确率。

Shape from shading （SFS） technique was used to generate 3-D structures from 2-D capsule endoscopic im- age. For the assumptions of the SFS technique, a preprocessing method for endoscopic image was proposed. Smooth and visually scene could be created based on the preprocessed endoscopic image, while preserving the structure of the observed objects. Experiments with real capsule endoscopic data demonstrate the 3-D shape enhanced the video and therefore leading to a more accurate diagnosis.