为了提高水平集图像分割的质量和减少水平集迭代次数, 本文提出了新的能量公式和水平集函数. 在粗糙集数据离散化基础上引入了针对图像数据的离散化方法, 根据图像离散区域的信息对新能量函数进行直接加权并且对核函数进行间接加权, 使用加权的核映射函数将原始离散图像数据映射到高维空间, 从而使得该模型可以处理多种类型的图像甚至是一定信噪比的噪声图像. 新的能量公式联合由它导出的区域参数能够更好地表达同质区域的灰度信息, 从而能够更精确地分割图像.与传统水平集图像分割不同, 在迭代过程中新的水平集函数中的水平集元素可以拥有不同的步长, 步长越大水平集元素的更新速度越快并且水平集函数能够快速达到收敛状态, 实现快速图像分割. 人工合成图像和真实图像的分割实验表明本文方法可以获得更好的分割效果.%In order to improve the quality of image segmentation and decrease the iterations of level set evolution for image segmentation, a new energy function and a level set function are proposed. Image data discretization is introduced on the basis of rough set theory, in which the local information of the discrete region from the image is used to weight the new energy function directly and the kernel function indirectly. It can handle many types of images, even in a certain noise-signal ratio, by using the weighted kernel function to map discrete image data into higher dimension. The new energy function and the region parameters which are deduced by the new energy function can better express the gray level of the homologous region. Therefore, the proposed method can properly and accurately segment the image. Compared with the traditional methods, the new level set function is constructed by using a new energy function and the information of the discrete regions from the image, each element from the level set may have a different step-size during iteration. The higher the weight value, the faster the element is updated and the less iteration the method has. The proposed method shows the promising performance in the experiments based on synthetic and real images.
展开▼
