一种具有抗噪性的图像分割方法

Image Segmentation Method with Noise Immunity

基于图论的图像分割方法对有噪声污染的图像必须先进行预处理,算法自身不能抑制噪声。针对该问题,提出一种具有抗噪性的图像分割方法。该方法将图谱划分测度作为划分目标与背景的阈值分割准则,采用基于灰度值的权值矩阵代替基于图像像素个数的权值矩阵,描述像素之间的关联,并在图权计算中增加像素点与其邻域的空间相关信息,以提高算法的抗噪性。实验结果表明,使用该方法进行图像分割具有较好的分割效果,抑制噪声能力较强。

Aiming at the problem that the image segmentations based on graph theory have to use the image pre-processing to resist noise,this paper presents an image segmentation method with noise immunity.The proposed algorithm uses the normalized graph spectral cut measure as the thresholding principle to distinguish an object from the background.Instead of commonly using the image pixels,the algorithm uses the weight matrices based on the gray levels of an image to describe the relationship between image pixels.In order to improve the ability of resisting noise,the correlation between the pixel and its neighborhood is added in the graph weight computation.Experimental results show that the method of image segmentation has a very good segmentation results,and a strong ability to resist noise.