Hierarchical segmentation based on a multilevel thresholding

Segmentation plays an important role in image processing, machine vision, pattern recognition and so on. Numerous segmentation methods based on bi-level thresholding have been proposed, assuming there are only two main populations in an image. However, for a complex image consisting of more populations, a multilevel thresholding is required. This paper presents a new hierarchical segmentation method based on multilevel thresholding. The hierarchical segmentation utilizes an iterative threshold selection method as a basis to partition an image into two regions. Similarly, every region is segmented into two parts. This process continues until a better segmentation is obtained. Several objective measures are considered to evaluate the quality of segmentation. The experimental results indicate the proposed method can obtain an effective segmentation for an image with more populations.