胶囊内窥图像出血识别的BP神经网络算法

An Algorithm of Bleeding Detection in WCE Images Based on BP Neural Network

在大量的胶囊内窥图像中寻找出血区域或相关病理特征是一件非常费时费力的工作,使用计算机进行胶囊内窥图像出血区域智能检测是必然趋势。本文设计了一种BP人工神经网络应用于内窥图像出血模式的识别,并通过软件编程实现了基于BP神经网络的内窥图像出血区域智能检测的新方法。实验表明该方法能正确检测出内窥图像中的出血区域,从而将内窥图像分类为出血模式与非出血模式,达到了理想的效果。

It＇ s is very laborious and time-consuming to find the bleeding regions and other related abnormal characters from mass WCE （wireless capsule endoscope） video. Utilizing the technique of intelligent digital image recognition with computer to deal with the WCE video is an ideal substitute. In this paper, a BP artificial neural network is built to detect the bleeding regions in WCE images. The bleeding detection software application based on the BP artificial neural network is programmed, and the experiments demonstrate that the bleeding regions in WCE images can be correctly detected. Therefore the WCE images are classified into bleeding frames and non-bleeding frames.