TDI-CCD图像固有条带噪声的消除方法及实现  

Methods and Realization for Removing Inherent Stripe Noises in TDI-CCD Images

作　　者：赵变红[1,2] 何斌[1] 杨利红[1,2] 王文华[1] 禄金波[1,2] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院研究生院,北京100039

出　　处：《液晶与显示》2010年第5期752-758,共7页Chinese Journal of Liquid Crystals and Displays

基　　金：国家"863"高技术研究发展计划资助项目(No.863-2-5-1-13B)

摘　　要：存在于TDI-CCD图像中的条带噪声会影响图像的质量,降低系统的测量精度。针对TDI-CCD图像固有条带噪声的灰度值在原始信息中变化比较缓慢的特点,利用傅里叶变换域内的频谱图映射确定条带噪声频率的方法,分别采用低通、带阻滤波器对条带噪声进行消除。另外,提出了一种改进阈值的小波变换法来消除条带噪声,该方法能在小波分解后各个尺度的垂直方向上自适应地确定阈值。实验结果表明在消除条带噪声方面,改进阈值的小波变换法优于传统的傅里叶变换法。在整个频率域内,改进阈值的小波变换法能够较彻底地消除条带噪声,同时较好地保持了原图像的特征。The existence of stripe noises influences the qualities of TDI-CCD images,and reduces the measurement accuracy of the system simultaneously.For the characteristic of slow change of the stripe noises＇grey level compared with the primary information,the stripe noises are eliminated via low-pass filter and band-elimination filter separately using the method of determining the stripe noise frequency through the frequency spectrum mapping in the Fourier transform domain.Moreover,an improved threshold of Wavelet Transform method of eliminating stripe noise is presented in this paper.The each vertical direction＇s threshold in various scales is determined by itself after the wavelet decomposition.It is proved that in removing the TDI-CCD stripe noises,the improved threshold of Wavelet Transform is better than the traditional Fourier Transform.The stripe noises can be removed drastically via the improved threshold of Wavelet Transform method,the goal which reserves the spectral characteristic of original image is also achieved.

关 键 词：TDI-CCD 条带噪声 傅里叶变换 小波变换 自适应阈值 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]