Abstract: The paper proposes to use the Total Variation(TV) algorithm to eliminate the labeled objects in the satellite cloud images. The shapes of labeled objects are often long and narrow. The main effort focuses on improving the process of discreteness for the total variation algorithm through introducing different weights. The improved algorithm fulfills that it only fills the eliminating areas according to the surrounding information but keeping the surrounding information unchanged. The experiments show that the proposed method is effective for eliminating the labeled objects in the satellite cloud images.

Key words: Total Variation(TV), satellite cloud image, object eliminating

摘要： 将整体变分算法模型用于卫星云图中标注对象的剔除应用，根据云图标注对象的形状多为线形区域的特点，通过引入权值，对整体变分算法的离散化过程作了改进，使算法在利用邻域信息对待剔除区域进行填充时不改变待剔除区邻域的像素值。实验表明该算法对窄长条状或线性区域的剔除是有效的。

关键词: 整体变分, 卫星云图, 对象剔除

CLC Number: 

• TP391.4