Auto-detected areas in the Capsule Endoscope images
Wei Jin-qi · China Medical Engineering · 2011
【Objective】 Capsule endoscope produces large quantity of images,among which valueable images may be missed by human-eye selection.This article explores autodetection of ulcers,edema etc.among these images.【Methods】 Noise reduction,multi-resolution wavelet transform with db handle and HSB color space conversion are orderly processed.Adaptive value of the radius will be to make open and cloese operation and obtain the basic shape of the ulcers.Finally,the shape will be compared with,the ones experts outline manually.【Results and Conclusion】 Wavelet transform remains details of the target area.Similarity can be observed between autodetected areas and the ones experts outline,and less time is spent when the outline of ulcers preliminary obtained.