摘要：Wireless capsule endoscopy is the golden standard in the investigation of the small bowel, but is time consuming and burdening for the physicians, as the imaging conditions are challenging and the variability of the acquired frames is great. This paper carries out a study on the automatic detection of bubbles and debris, based on an algorithm combining colour slicing techniques, textures and position within the WCE frame and the intestinal tract. Their identification represents an important phase in the investigation of a video file furnished by the endoscopic capsule, as it allows the reduction of the important frames submitted to analysis for potential lesion detection. Experimental results prove that the algorithm is effective in reducing the number of WCE images with relevant content, reducing thus the time spent for the analysis of the images acquired by WCE.Wireless capsule endoscopy is the golden standard in the investigation of the small bowel, but is time consuming and burdening for the physicians, as the imaging conditions are challenging and the variability of the acquired frames is great. This paper carries out a study on the automatic detection of bubbles and debris, based on an algorithm combining colour slicing techniques, textures and position within the WCE frame and the intestinal tract. Their identification represents an important phase in the investigation of a video file furnished by the endoscopic capsule, as it allows the reduction of the important frames submitted to analysis for potential lesion detection. Experimental results prove that the algorithm is effective in reducing the number of WCE images with relevant content, reducing thus the time spent for the analysis of the images acquired by WCE
