两种改进的局部阈值分割算法

Two Improved Local Threshold Algorithms

局部阈值分割法存在两个明显的缺点:当一个子块几乎完全属于背景或目标时,直方图为单峰,很难进行分割;当目标被分在不同子块时,分割结果存在块状效应。为此,结合全局阈值中的迭代法和局部阈值法,提出最近邻插值阈值分割算法,并在这种算法的基础上,利用线性插值,提出等间距插值阈值分割的算法,并通过仿真试验说明这两种算法在分割图像时的有效性和可行性。

The local threshold have two obvious shortcomings：when one sub- block nearly belongs to the background or the goal, the histogram is unimodal and very difficult to divide; When the goal is divided in the different sub - block, segmentation result existence massive effect. For this,the article unifies global threshold iterative algorithm and local threshold segmentation, and proposes the nearest neighbor interpolation threshold. Based on this method of threshold segmentation, using the linear interpolation,the equal interval interpolation threshold is proposed, and then through simulation experiments showing two methods＇ validity and feasibility.