基于横噪声消除的医学粘连图像边缘分割算法

Based on the Horizontal Noise Elimination of Medical Adhesion Image Edge Segmentation Algorithm

针对CCD型医学图像中,在细胞粘连区域信号附近出现的较强的横条纹噪声干扰,影响图像信息的正确性,分割后会存在边界模糊和锯齿条文的问题,为提高医学图像分割效果,提出了一种基于横条纹噪声消除的医学粘连图像边缘分割算法。分析了医学图像中边沿横条纹噪声的原因,通过wold纹理模型与多尺度马尔可夫随机场模型,利用确定性随机场和不确定性随机场的谱属性不同的特征,将医学图像边沿的干扰特征分离开,有利于对粘连医学图像进行分割。实验证明,方法有效地去除了横条纹噪声并很好地保留了图像的边缘和细节信息,同时运算复杂度低。

According to the CCD type medical image, in cell adhesion area near signal in strong horizontal stripe noise interference, the correctness of the image information influence, the division will exist boundary fuzzy and sawtooth provisions. In order to improve the effect of medical image segmentation, this paper puts forward a based on the transverse stripe noise elimination of medical adhesion image edge segmentation algorithm. Analysis of the medical image edge transverse stripe noise reasons, by the use of combined with wold texture model and multi-scale markov random field model, using the certainty and uncertainty with the airport with the airport spectral properties of different features, will be the medical image edge interference characteristics apart. Is advantageous to adhesion medical image segmentation. Experimental results prove that the method can effectively removed the transverse stripe noise and it is quite good to keep the image edge details and information, and at the same time operation complexity is low.