FIGURE 4.
The point cloud results when using the red channel images with the IC dye. The red pyramids represent the estimated endoscope camera poses and the gray points represent the 3D points. For Subject A to D, we show the comparison between the initial point cloud, , and the final result of our outlier removal algorithm,
, while only the final results are shown for Subject E to G due to the space limitation. From these results, we can confirm that our outlier removal algorithm can produce a clean point cloud. In some subjects, some parts have a noticeable hole. It is because some parts of the stomach are not captured in the endoscope video.