基于图像分割的云量估计

Cloud Cover Estimation Based on Image Segmentation

摘要: 云是天空中人类可以直接通过肉眼感知到的主要气象要素之一，云的探测能够帮助人们识别阴晴风雨，预知天气变化。云量估计是云探测组成部分。本文主要采用数字图像处理技术中阈值分割方法估计云量，相比传统人工目测方法和一般器测方法具有便利性和实用性。本文利用地基的可见光范围的云图，在饱和度空间运用分割算法估计云量。对比分析了最大类间方差法、最大熵法、基于数学期望分割法的分割效果，并提出了改进算法。该方法在云底高度较高的云类获得的分割效果更符合云量估计标准。

Abstract: Cloud is one of the important meterological elements that could be perceived by eyes. Cloud ob-servation could help us make weather forecast. Cloud cover estimation is part of cloud observation. The paper uses the method of threshold segmentation of digital image processing to estimate cloud amount. Compared with traditional artificial and ordinary instrument method, the measurement is more convenient and practical. The paper analyzes the result of three methods of image segmentation which are Otsu, max entropic thresholding and Mathematics expectation, and puts forward a modified method. The modified method works better on cloud image which shows higher cloud.