Research On The Parallel Segmentation Algorithms Of Medical Image Based On Mapreduce

Medical image segmentation is a very important work in medical image processing, ich is the key step from processing to analysis. It is the foundation of target separation, feature extraction and the parameter measurement and so on. Medical image segmentation make it possible that we can have better image diagnosis and understanding.Compared to other images, medical image has different characteristics. Such as low contrast, more noise and fuzzy boundaries between different tissues and organs or lesions. Those characteristics make it to be a difficult research. So far, any kind of segmentation has its own problems. Therefore, research of medical image segmentation algorithm is currently a hot research topic. Reasonable image segmentation is one of the most important step in the medical image processing. This paper takes brain MRI image sequence as the research object, we do some research about the detection and segmentation of brain different organization. The main research work is as follows:This paper designed a medical image processing system, system consist of two parts. The first part is MRKDCA algorithm; The second part is the MRTSCS algorithm.Based on Chebyshev inequality of kernel density estimation algorithm (KDCA algorithm) is a new background modeling algorithm, the MRKDCA algorithm is a distributed background programming model, which is based on KDCA algorithm and MapReduce.Image segmentation algorithm based on pixel connected rate (TSCS) algorithm is a new kind of image segmentation algorithms, the MRTSCS algorithm is a distributed image segmentation algorithm based on TSCS and MapReduce.The instantaneity and complexity of image segmentation algorithm need us improve hardware technology and parallel processing algorithm. A single point of serial algorithm has too much problems in facing with large-scale data, and it is relatively low efficiency and weak, it is hard to meet the needs of the people to handle huge amounts of data, in this paper, on the basis of the Hadoop framework, we put forward two kinds of parallel algorithm based on Hadoop.They are efficient, cheap. We have apply it to the actual brain MRI images segmentation, and experiments show that those methods have a accuracy, and when it face to large-scale of data they can meet the needs of people to handle huge amounts of data.