中文摘要：

为了解决传统的Otsu法均分像素点的缺点以及Kapur法易受干扰的不稳定性,分析了这两种算法理论上的联系并结合它们的各自优势,提出一种新的算法,实现两者在图像阈值选取上的一种均衡。实验结果表明,新算法对图像直方图不具有明显双峰、目标像素点占比很小、图片存在干扰信息源等情况均能得到理想的分割效果。该算法可用于更大范围的图像分割处理,并且展现了更强的稳定性和自适应性。

英文摘要：

In order to overcome the defects of the equally dividing trend of Otsu algorithm as well as the instability of Kapur＇ s algorithm, this paper analyzed the theoretical correlation between these two algorithms and combined their respective superiorities. It proposed a novel algorithm to realize the balance between these two algorithms in image thresholding. The testing results illustrate that the new algorithm can get perfect segmentations for images with non-ideal bimodal histogram, little targets and disturbing information. This algorithm is applicable in a more general scope of image thresholding, and performs better stability and adaptivity.