【摘要】 针对灰度不均的弱边界现象通常发生在真实世界的图像中,可能会增加图像分割的难度,传统的局部二值拟合(LBF)模型分割方法需要建立复杂的数学模型,局部化属性也导致了该模型更容易陷入局部极值,且需要较多的算法迭代次数。为了克服上述缺陷,提出了一种新的图像分割处理方式,首先用形态学运算去除血管图像的背景,以达到增强图像的目标。然后对处理后的血管图像采用双边滤波器滤波,使用最大类间方差法(Otsu)进行分割,将最优阈值代入Canny算法中,进行再分割。实验结果表明,该方法不需建立复杂的数学模型和数值分析算法,可以用较少的迭代次数,对于灰度不均且弱边界的医学图像实现较好的分割。更多还原

【Abstract】 Gray uneven weak boundary phenomenon often happens in the real world images which may bring more difficulty for image segmentation.Traditional Local Binary Fitting( LBF) model segmentation method needs complicated mathematical model to be set up.Localization properties make the model easier to fall into local extremum,and require a longer algorithm iterations.In order to solve these defects,a new way of image segmentation was proposed.Firstly,the background of the blood vessel image was removed by the morphological operation,in order to enhance the image.Then after processing of the blood vessel image filter by using bilateral filter,using the most between-cluster variance method( Otsu)segmentation,the optimal threshold was introduced into the Canny algorithm for segmenting again.The experimental results show that this method does not need complex mathematical model and numerical analysis algorithm,and less iteration for medical images with uneven gray level and weak boundary to achieve better segmentation.更多还原