基于局部统计特征约束的Snake模型图像分割方法

Image Segmentation Algorithm Based on Snake Model of Local Statistical Characteristic Restriction

针对血管内超声(IVUS)图像边界模糊不清的问题,提出一种基于局部统计特征约束的Snake模型图像分割方法.首先利用IVUS图像的先验知识建立图像特征的统计信息,并以此信息构造变化的膨胀力,同时构造了轮廓曲线的内部平衡力,以此构造新的Snake模型,然后利用该模型进行图像分割.实验结果表明,该Snake模型对初始轮廓依赖性小,对噪声不敏感,能准确分割出具有模糊边界的目标图像.

In order to clarify the fuzzy boundary of intravascular ultrasound（IVUS） image,an image segmentation algorithm based on the Snake model of local statistical characteristic restriction is presented.In this algorithm,the prior knowledge of IVUS image is used to establish the statistical information of image characteristics,with which the expanding force is set up and the internal balance force of the contour is proposed to construct a new Snake model.The newly constructed model is then adopted to segment the IVUS image.Experimental results show that the proposed Snake model is little dependent on the initial contour and is not sensitive to noise.Thus,the target image with fuzzy boundary can be accurately segmented.