一种基于灰度共生矩阵的云图检索方法

Satellite Cloud Image Retrieval Based on Gray Level Co-Occurrence Matrix

卫星云图检索可帮助气象预报人员快速定位历史相似天气,根据云图纹理特征区分度较大的特点提出一种采用纹理特征对卫星云图进行相似性检索的方法。针对找到一个普遍适用的纹理特征非常困难的问题,提出一种根据特征值的方差分布情况从大量备选特征中快速找出适合某类图像检索所需的纹理特征值的方法,并以灰度共生矩阵的特征值提取为例,对卫星云图进行相似性检索。检索流程为:首先对云图进行云地分离的预处理,然后从云图的灰度共生矩阵中提取有效的检索特征生成特征值,并与历史云图库对应的特征库进行相似距离计算,最后根据距离的排序顺序输出最终的检索结果。实验表明,该方法能有效地从历史云图库中检索出具有相似视觉特征的云图,说明该方法可以用于卫星云图的相似性检索。

Satellite cloud image retrieval can help forecasters positioning similar weather in the past quickly. As the texture characteristics of satellite cloud image has a greater degree of differentiation, puts forward a method of satellite cloud image retrieval based on texture characteristics. It is very difficult to find a universally applicable texture feature, proposes a method based on the distribution of variance to find out features needed for certain types of image retrieval. Takes the extraction of gray level co-occurrence matrix as example to carry out similarity retrieval of cloud images. Carries out the pretreatment of separation on the cloud image. Extracts effective re- trieval features from the gray level co-occurrence of the cloud image to generate feature vector and calculate the similarity distance of this feature vector to the history feature vector library. Outputs the results according to the sorted order. Experiments show that this method can retrieve satellite cloud images with similar visual characteristics from the history cloud image atlas. It can conclude that this method can be used in similarity retrieval for satellite cloud image.