基于GPGPU的快速医学图像FCM分割算法

由于医学图像生成容易受到空间时间影响,噪声较大,具有不确定性,传统的硬分割方法很难取得理想的分割结果.模糊分类技术能很好地处理医学图像中的不确定性,却由于计算量大不能保证实时性.灰度统计方法和通用计算图形处理器技术的引入,保证了初始聚类中心的准确性.又由于模糊C均值聚类算法是可并行的,将其改进并在图形处理器上完成计算,降低了算法迭代次数和计算时间,保证了实时性.实验结果表明,使用该方法对医学图像分割得到了良好的结果.

A Fast FCM Algorithm for Medical Image Segmentation Based on GPGPU

The generation of medical image is easily influenced by external environment,so the medical image contains much noise and uncertainty.The traditional hard image segmentation method is difficult to achieve the ideal result.The fuzzy classification technique can process the uncertainty in medical image.However,it contains large computational quantity and can not guar-antee the real time.The method of histogram and the technology of general computing on graphic processing units are utilized.The better initial cluster centers are given to reduce the iteration. The parallelization of the image segmentation algorithm is realized on GPU platform to reduce the computational time and ensure the real time.The experiment results show that the method can produce better results.