基于区域生长的图像分割研究进展

摘要

Image segmentation is one of the fundamental problems in image processing and image analysis.The inherent ill-posedness of image segmentation is the greatest challenge of this research field.Region growing methods are widely used in image segmentation because of their simplicity and efficiency.According to certain major region growing-based segmentation models at present,we summarize the approach of region growing,including seeds selection,homogeneity criterion,applicable condition and improvement direction.Finally,the trend of region growing methods in future is prospected.%图像分割是图像处理与图像分析的基础步骤,主要难点是图像分割的不适定性.区域生长法算法简单,计算速度较快,是一种运用广泛的图像分割技术.本文根据当前主要的基于区域生长的分割模型,讨论了区域生长法中的种子点自动选取、相似性判断准则、适用情况及其改进方向,并对区域生长技术的发展进行了展望.