基于序列内窥镜视频图像的膀胱三维场景重建

内窥镜被广泛用于器官和体腔内部的诊断成像,尽管采集到的内窥镜视频中包含着丰富的信息,但是由于内窥镜的成像范围受到探头太小的限制,图像的放大倍数与视野范围不能兼得,且视频不能将个体二维图像与三维器官形态关联起来,大部分具有巨大数据潜力的内窥镜视频仅用于医师实时指导,并在收集后丢弃。提出基于序列内窥镜视频图像的膀胱三维场景重建的方法。先提取前后两幅序列图像中相对应的多个特征点,基于特征点采用运动恢复结构算法计算得到膀胱模型三维点云。然后,采用泊松曲面重建算法对各三维点进行网格化,并完成场景的三维重建。实验结果表明所提出的算法是可行的。

Three-dimensional reconstruction of bladder based on serial endoscopic video images

Endoscopes are widely used for diagnostic imaging inside organs and body cavities. Although the acquired endoscopic video contains rich of information, the scope of the endoscope is limited by the small size of the probe, and the magnification of the image is The range of vision cannot be both, and the video cannot associate individual two-dimensional images with three dimensional organ morphology. Most endoscopic video with huge data potential is only used by doctors for real-time guidance and discarded after collection. This paper proposes a three dimensional reconstruction method of bladder based on serial endoscopic video images. Firstly, the corresponding multiple feature points in two sequence images were extracted before and after the three dimensional point cloud of the bladder model was calculated based on the feature points using the motion recovery structure algorithm. Then, the three-dimensional points are meshed with the Poisson surface reconstruction algorithm, and the three-dimensional reconstruction of the scene was completed. Experimental results show that the proposed algorithm is feasible. These results validate the possibility of the proposed method.