摘要
以卫星侦察红外图像背景为研究对象,在分析其特点的基础上,讨论了可用于红外图像背景分类的特征量。为了描述卫星侦察红外图像背景的纹理特征,基于灰度共生矩阵和直方图,计算了能量、惯性矩、熵、相关性、灰度均值和方差等统计特征指标。通过大量样本的识别实验,结果也验证了文中所提取的纹理特征量可以有效识别卫星侦察红外图像的背景。
The background of satellite infrared image is taken as an object of the study. Based on the trait analysis of satellite infrared image, the features, which can be used to classify the background of infrared image, are proposed. For describing the texture features of satellite infrared image, statistical measures included energy, inertia, entropy, correlation, gray mean and variance are computed on the base of gray co-occurrence matrix and one-dimensional histogram. A lot of tests validate that the texture features are efficient in classifying the background of satellite infrared image.
出处
《红外技术》
CSCD
北大核心
2005年第2期143-146,共4页
Infrared Technology
基金
国防预研项目(编号41101010506
41322020102)
关键词
分类
纹理特征
红外图像
卫星
classification
texture features
infrared image
satellite
