| Wireless capsule endoscopy is a revolutionary technology that allows physicians to examine the digestive tract of a human body in the minimum invasive way. Physicians can detect diseases such as polyps, ulcers, Crohn’s disease and so on. Although this technology is really a marvel of our modern times, but it exists several drawbacks, such as lower sampling frame rate, passive forward, random shooting, hard to focus on observation of suspicious lesion site, the large number of images, low treatment efficiency, and no 3-D representation of the objects is captured from the camera of the capsule.In this thesis, we provided solutions to deal with each of the above issues improving the current technology without forcing hardware upgrades. These methods worked together to create a smooth three-dimensional image in a good visual, while retaining the observed object structure to ensure the realism of the image. Firstly, the characteristics of the gastrointestinal tract image were analyzed. Through observation we found that some remnants of the gastrointestinal tract or its mucus bubbles were remained, resulting in the formation of serious reflection, the highlighted area would appear in the output image. Based on the principle of "neighboring pixels’ characteristics are similar", we used the digital morphology means iterative algorithm to finish removing the highlights. Three methods of interpolating were tried to be compared. Because of the lower sampling frame rate, we adopt monocular visual image reconstruction, a reasonable choice of optimization algorithms and suitable surface constraints of SFS methods to obtain the depth. In order to ensure that the results were close to the real gastrointestinal surfaces, we introduced local cubic curve interpolation algorithm of the NURBS to the three-dimensional surface smoothing treatment, aimed at improving the deficiencies of the existing three-dimensional reconstruction. Therefore, the purpose of our method is to provide convenient and effective treatment of the gastrointestinal tract lesions.Our results were compared with other methods, based on large number of sources of different endoscopic images, we can conclude that our algorithm for image visualization can appear a stronger and clearer cavity outline. Finally, illustrating the results are given in the end of the article. |
