基于纹理分析的去噪声方法研究  

Study of Removing Stripe Noise Based on the Measure Analysis

作　　者：白香花[1] 刘素红[2] 唐世浩[1] 朱启疆[1] 帅艳民[1] 

机构地区：[1]北京师范大学遥感与地理信息系统研究中心,北京100875 [2]中国科学院遥感应用研究所,北京100101

出　　处：《遥感技术与应用》2003年第1期36-40,共5页Remote Sensing Technology and Application

基　　金：973国家重点基础研究规划项目(G20000779)。

摘　　要：中巴资源一号卫星高分辨率CCD相机接收的图像数据,灰度值偏低,对比度较差,且存在明显的条带噪声;当图像被增强时,这种条带更加突出,严重影响了图像的质量以及清晰度,不利于数据的应用。从与以往不同角度出发,提出了基于纹理分析的去条带方法。即将图像上具有特定特征的噪声信息看作是某种对应的纹理特征,在基于纹理特征滤波法去噪声基础上,再用灰度共生矩阵法进一步进行直方图平滑,以消除局部噪声。同时将此方法与基于统计的均值调节法进行对比分析。The lower value of gray level, relatively poor contrast and obvious stripe noises exist in the image received by the high\|resolution power CCD camera carried on China\|Brazil Resources satellite No.1. The stripe affects the quality of data and clarity; furthermore, it restricts the use of the data. With enhance of the image, the stripe noise is more obvious. In this paper, a method of removing the stripe based on the texture analysis is brought up from a different aspect, that is, we consider the certain noise information in the image as the some counterpart texture feature. Furthermore, gray level co\|occurrences matrix is applied to smooth the histogram based on the texture filter so as to remove the local stripe noise. In addition, the mean value adjustment method is used to compare with the texture analysis one.

关 键 词：纹理分析 去噪声方法 条带噪声 纹理特征 灰度共生矩阵 特征参数 遥感图像 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]