A modified region-scalable fitting model was put forward against the defects such as being less divided and the slow convergence of outline during the segmentation of certain medical images by the RSF model.K-means was employed to process the medical image globally, and then a new kernel function replaced the Gaussian function.On the basis of the new kernel function, a new energy function was re-established, and the internal energy was introduced into the level set model as a penalty function.Compared with traditional RSF model, the results showed that the accuracy of the improved model increased by nearly 40%, and the rate increased by about 30%.%针对可伸缩区域拟合能量(RSF)模型在分割某些医学图像时会存在欠分割以及轮廓收敛速度慢等问题,提出一种改进的RSF模型.利用K均值对医学图像进行全局处理,用一个新的核函数代替高斯函数.在新的核函数基础上重新建立能量泛函,并将一个内部能量项作为罚函数项引入到水平集模型中.结果表明,与传统的RSF模型相比,改进模型的分割精度提高了近40%,分割速度提高了近30%.
展开▼
