Mean Shift算法的基本原理，着重研究了Mean Shift的迭代过程和收敛性以及特征空间的结构。通过Mean Shift结构分析，本章提出了一种快速区域合并算法并将其应用于改进的Mean Shift图像分割算法。-The basic principle of the Mean Shift algorithm, focusing on the Mean Shift iteration process and convergence and the structure of the feat

Automatic Image Segmentation by Dynamic Region Merging

图像分割是一种重要的图像分析技术，对图像分割的研究一直是图像技术研究中的热点和焦点。分水岭算法是基于数学形态学理论的图像分割算法，但是对噪声敏感且存在过分割的现象。为提高图像分割效果，本文通过对影像进行滤波处理，用改进的快速区域合并算法优化分水岭算法进行影像分割。-In this work a technique for cloud detection and classification from MSG–SEVIRI (Meteosat Second Generation–Spinning

In the proposed algorithm, these two issues are solved by a novel predicate, which is defined by the sequential probability ratio test (SPRT) and the minimal cost criterion. Starting from an over-segmented image, neighboring regions are progressivel