Research On Bleeding Detection Algorithm And Storage System For Wireless Capsule Endoscopy Image

Wireless capsule endoscopy(WCE) is an emerging endoscopy to achieve full gastrointestinal examination. With the development of technology, it shows a great popularization and application prospects in the digestive tract health, by function improvement and cost reduction. At the same time, as the WCE image data increasing, the technical problems is becoming obvious. In order to develop the WCE real-time telemedicine system, especially in the background of telemedicine, it needs efficient lesion image recognition algorithm and high-performance image data storage technology.This paper aim to solve the problem of low efficiency of lesion image recognition in big WCE image data. The feature that adjacent WCE images have strong correlation is analyzed by proposing correlation. A novel method is proposed for bleeding detection based on hierarchy strategy and rule modification. Considering the WCE image data storage requirements in the background of telemedicine, we design a WCE image data storage system based on MongoDB and implement the management of image data and lesion recognition function for future real-time telemedicine system. The main work of this paper is shown as follows:In order to solve the problem of detecting bleeding images from massive WCE image data, we analyzes the image pattern recognition and the characteristics of the WCE image, and put forward an algorithm by fusing and improving some existing bleeding detection algorithms in this paper. The algorithm is estimated in three levels including a rough level, a sliding window for modification level and a precise level, which not only exerts the characteristics and advantages of different algorithms, but also balances the cost time and the performance of detection. The experimental results suggest the effectiveness of the proposed method.In order to meet the needs of high-performance storage of WCE image data for telemedicine, we research the framework and deployment of MongoDB, and analyses the feasibility and the key issues of the WCE image storage system based on MongoDB, such as parallel processing, database schema, etc. Finally, a storage system is designed in the experimental environment.According to the design mentioned above, an application software is programmed by C++ in the development environment, including the collaborative architecture between MongoDB and SQLSERVER, a database cluster, C++ access driver, etc. Finally, performance comparison and function verify are carried out in the application software. The experimental result shows that the system performs well in WCE image storage and bleeding detection, which lays the foundation for future system expansion and practical products.