磁共振截断频谱图像的层析法边缘检测技术

本文提出一种适合于含有截断伪影磁共振图像的边缘检测新算法。按奇异函数分析理论 ,把任何有截断伪影的信号表示为以奇异点为参量的截断奇异函数的加权和 ,奇异点和加权系数由该信号决定。层析边缘检测的关键在于一层层地剥去奇异点上的加权截断奇异函数 ,从而一层层地检测出奇异点。这些奇异点可以构成图像的边缘。本方法可以有效地剔除由截断伪影而引入的虚假边缘。实际和仿真结果都表明这种方法效果高于现有方法

THE NEW EDGE DETECTION TECHNOLOGY OF MR IMAGES WITH TRUNCATION ARTIFACT USING LAYER ANALYSIS METHOD

We propose a new edge detection algorithm suited to MR images with truncation artifact. According to the singularity functions analysis (SFA) theory, any digit signal with truncation artifact can be expressed as the weighted sum of truncated singularity functions (TSF) related to the signal singularity locations, the singularity locations and weight coefficients are determined from this signal. Layer analysis method of edge detection is that weight truncated singularity function on singularity point is reduced from the signal with truncated artifacts layer by layer, and then the singularity points are detected layer by layer. Edges of MR images with truncation artifacts can be constructed by these singularity locations. False edges due to truncation artifacts can be removed using this method. Experiment and simulation results showed that this method gave much better performance than currenthy available method.