图像二维阈值分割的数据场方法

Data Field-Based Method for Image Segmentation with Two-Dimension Threshold

针对图像分割中最优阈值选择这一难题,提出了一种新的图像二维阈值分割方法.该方法引入数据场的思想,将图像从灰度值空间映射到数据场的势空间;将二维灰度直方图的频率作为数据场对象的质量,计算二维直方图元素之间的相互作用和影响,生成二维直方图的三维数据场;再通过势心削除、势心合并等环节获得最优分割阈值,在不明显增加时间复杂度的前提下得到较好的分割结果.对标准图像数据集以及部分加噪声图像的分割实验表明,该方法是合理、有效的,能够适应大多数图像的分割,具有一定的抗噪性能,是经典一维最大类间方差法的有效补充.

In order to correctly select the optimal threshold for image segmentation,a novel method of image segmentation based on data field is proposed.The method maps the image from grayscale space to the appropriate potential space in data field,and measures the interactions of the elements in the two-dimension histogram by taking the frequency of two-dimension gray histogram as the mass of data field,thus generating a three-dimension data field.Then,by employing the potential center elimination and combination,the optimal threshold is determined and good segmentation result is obtained without significantly increasing the time complexity.It is indicated by the experiments for standard image datasets and some noisy images that,as an alternative to OTSU,the proposed me-thod is reasonable and effective with certain noise resistance.