一种基于KPCA和形状先验知识的图像分割模型

Image Segmentation Based on KPCA and Shape Priori

对含有噪声、遮挡和信息缺失的图像进行分割,如果仅使用图像自身信息难以得到满意的结果。因此,本研究提出了一种新的融合图像信息和形状先验知识的可变形模型。在Chen等人的工作基础上,提出用核主元分析(KPCA)代替主元分析(PCA)来捕获形状信息。KPCA能更好地表示形状先验知识,允许待分割的目标形状与先验形状存在较大差异或非线性变形,而PCA需两者足够接近。同时,所用的分割模型包含了图像信息项和形状先验项,充分考虑了在分割过程中平衡全局图像信息和形状先验知识的相互作用。将本研究的模型和基于PCA的分割模型应用于合成图像和医学CT图像,结果表明KPCA更能准确地识别出与先验形状差异较大或背景污染严重的目标物体。

Segmentation means to separate an object from the background in a given image.Relying on image information alone can not yield satisfying results due to noise,occlusion,or missing parts existence.To effectively solve this problem,it is necessary to introduce shape priori into the segmentation model.This paper proposes a new deformable model based on the shape priori.Inspired by the works of Chen et al,we use KPCA（kernel PCA） instead of PCA（principal component analysis） to capture the shape information.KPCA can express better shape priori knowledge and allows nonlinear transformation or a quite difference between the object and the priori shape.However,PCA requires that the two shape need to be similar enough.Moreover,our segmentation model includes the image term and the shape term to balance the influence of the global image information and the shape priori knowledge in proceed of segmentation.Our model and the segmentation model based on PCA are applied to synthetic images and CT medical images.The comparative results show that KPCA can more accurately identify the object with large deformation or with serious background noise.