一种基于直方图的图像二值化算法

摘要

图像二值化在计算机图像处理技术中应用十分广泛，其关键问题是阈值的选取。对于具有明显双峰直方图的图像，可选取双峰的谷底作为阈值，这是基于直方图的图像二值化算法的基本思想。根据其基本思想，提出了一种新的阈值选取算法——最佳阈值迭代法。算法先根据图像的最大灰度值和最小灰度值设定一个初始阈值，然后通过迭代循环计算在该阈值下的目标和背景的中心值，当目标和背景中心值的平均值和假定的阈值相同时，则循环中止，并以此值为阈值。实验结果表明，该算法可以较好地保留图像的细节信息，且具有一定的抗干扰性和稳定性。%Image binaryzation is widely used in image processing technology. The key problem is to select an appropriate threshold value. For images with obvious bimodal histogram, the bottom of bimodal histogram was selected as threshold value, which was based on the basic idea of image binaryzation algorithm with histogram. Thus a new optimal threshold value iteration method was selected. This algorithm set an original threshold value according to the maximum and mini?mum gray values of images, then calculated the center value of target area and background by iterative method with threshold value. When the average value of center value of target area and background was as same as the assumed thresh?old value, the circle stopped and the value was set as the final threshold value. The experiment shows that the algorithm can retain the details of images effectively, and it has a certain anti-interference and stability.