【作者】 隋修宝； 柏连发； 陈钱； 顾国华；

【Author】 Sui Xiubao Bai Lianfa,Chen Qian Gu Guohua National Defense Key Laboratory of Optoelectronic Engineering,Nanjing university of science and technology, Nanjing,210094,China

【机构】 南京理工大学光电工程重点学科实验室；

【摘要】 条纹噪声是红外图像中的一种噪声类型,它的主要特点是同一行(列)中噪声的幅度和相位相同;而不同的行(列)中噪声幅度和相位不同,因此,为了消除该噪声,去除条纹噪声算法需要在一帧时间(标准PAL制,40毫秒)内完成。本论文分析了条纹噪声的产生机制,并提出了能够在一帧时间内完成的去除条纹噪声算法。该算法利用局部窗口遍历一帧图像的所有像元,获得每行像元的极值,并通过相邻行之间灰度值的继承性完成一帧图像的校正参数的计算;通过设定合适的阈值消除图像边缘和系统噪声对参数精度的影响。实验结果和理论分析表明,本论文提出的算法能够在一帧时间内有效的消除红外图像的条纹噪声。更多还原

【Abstract】 Strip noise is a kind of typical noise in infrared image whose feature is that,the amplitude and phase of the noise are the same at the same row(or column) and different at different row(or column).In order to eliminate the noise,the eliminating algorithm for the strip noise should be completed in one frame time(standard PAL,40ms,the same below).This paper studied the mechanism of strip noise and brought forward the eliminating algorithm for strip noise in one frame time.The algorithm obtains the extreme value of every row pixels by traversing all pixels of a frame in local window and computes noise gain and offset parameters through the succession of adjacent row pixels.By setting adaptive threshold value,eliminating the influences of image edges and system noise,we can achieve the precise gain and offset parameters and complete the eliminating algorithm for strip noise.The last experiment and theoretical analysis indicate that our algorithm can eliminate strip noise effectively in one frame time.更多还原