Algorithm For Bleeding Determination Based On Object Recognition And Local Color Features In Capsule Endoscopy

Yong-Gyu Lee, Jin Hee Park, Youngdae Seo, Gilwon Yoon
2013 Zenodo  
Automatic determination of blood in less bright or noisy capsule endoscopic images is difficult due to low S/N ratio. Especially it may not be accurate to analyze these images due to the influence of external disturbance. Therefore, we proposed detection methods that are not dependent only on color bands. In locating bleeding regions, the identification of object outlines in the frame and features of their local colors were taken into consideration. The results showed that the capability of detecting bleeding was much improved.
doi:10.5281/zenodo.1070181 fatcat:cnzok6ye2fduhg4zpzor24xhfi