摘要
光流法是研究视觉和客观物体之间关系的数学分析方法,在图像识别和追踪的相关领域已有广泛的成功应用。本文通过介绍经典的Horn-Schunck方法和Lucas-Kanade方法,引入光流方法的基本概念和原理,进而在此基础上进一步用统一的变分方法框架解释绝大部分光流场算法。最后介绍光流方法在气象领域的应用及其前景。
Optical flow method is a mathematical analysis of the relationship between the visual sense and its object.It has been widely applied successfully in related areas of image recognition and tracking.This paper introduces the basic concept and principle of optical flow method by examples such as the Horn-Schunck and the Lucas-Kanade method,and then on the basis of further using the unified variational method to explain the vast majority of optical flow field algorithm.Finally,it introduces the application of the optical flow method in the field of meteorology and its prospects.
出处
《气象科技进展》
2015年第4期16-21,共6页
Advances in Meteorological Science and Technology
基金
国家重大科学研究计划项目(2012CB957803)
关键词
光流
变分法
图像识别
气象
optical flow,variational method,image identification,meteorology
