As colonoscopy is needed for screening colorectal polyps, and the first step of classification these polyps is recognizing them, it is worth to research, whether the colonoscopy databases could be improved by image processing and contour detection. In the following considerations a reflection filtering and background subtraction with large-sized mean filter is used as image preprocessing tools, and hybrid method for segmentation of pathological forms, based on template matching and active contour model as contour fitting for segmentation of the image is made.