自动选取阈值方法比较研究

A Comparative Study of Automatic Selection Threshold Method

在数字图象处理中 ,阈值处理是非常有用的图象分割技术。它已被广泛地应用于数字图象处理的许多领域 ,近年来已有许多种阈值化方法被提出。阈值选取是图象处理与分析问题的基础 ,如何才能正确地找到适当的阈值 ,是一个非常棘手的问题。针对几种常用的图象二值化自动选取阈值方法 ,通过计算机仿真对实验结果进行了比较研究。并在此基础上 ,提出了一种新的图象二值化算法。该算法着重于在图象二值化时保留图象的边缘特征。其基本思想是 :首先 ,用微分算子检测图象的边缘 ;然后 ,在这些边缘象素点上进行二值化阈值的自动选择 ;最后 ,对于其它非边缘象素点则采取常规方法进行二值化处理。实验结果表明 ,这个基于边缘特征检测算子的算法能很好地保留原图的边缘特征 ,并能处理低质量的图象。

In digital image processing, threshold is a well-known technique for image segmentation. Because of its wide application to other areas of the digital image processing, quite a number of threshold methods have been proposed over the years. Threshold selection is very important for image segmentation. How to get fine threshold is a difficult problem in image processing. Through experiment with several general threshold selection methods, the results were analyzed. Based on operators for edge feature detection, a new method of automatic threshold selection was proposed. The main idea of the algorithm is as follows: first, using differential operator detects the edge of image; then, the threshold is selected automatically on the pixel points of the edge; finally, using the general method copes with the pixel points of the non-edge. The experiment shows that the new algorithm keeps the original edge features well and is efficient for processing low-quality image.