摘要： 图像分割是进行图像处理的关键步骤。目前很多图像分割的技术都需要人工干预，而且是针对单纯的图像背景才能达到目标。为了解决有复杂背景的图像的分割问题，采用自适应阈值（阈值插值）的方法，并对其加以改进，用迭代阈值判断子图像的直方图是否是双峰分布，同时确定直方图有双峰的子图像的阈值；用双线性插值的方法确定直方图非双峰分布的子图像的阈值，使这种算法在实际中可行。实验证明这种方法适用于复杂背景的图像分割，且通用性比较好。

关键词: 图像分割, 自适应阈值, 迭代阈值, 灰度插值

Abstract: Image segmentation is a key step for image processing. At the present, many image segmentation technologies need intervention from man and they only work well under simple background. In this paper, an improved adaptive threshold (threshold interpolation) method is introduced for resolving segmentation of image in complicated background. The algorithm of iterative threshold is used to determine whether sub image has a bimodal gray-level histogram and at the same time decide the threshold of sub images. The algorithm of bilinear interpolation is used to determine the threshold of sub image which has not a bimodal gray-level histogram. The test shows that this method can perform well to image segmentation in complicated background and has good currency.

中图分类号: 

• TP391.41