论文部分内容阅读
提出了一种场景自适应的非均匀性校正固定图案噪声去除方法.基于多帧配准和相邻帧之间的全局位移估计,利用最小均方算法迭代计算均方误差函数,同时结合误差函数均方差以及互相关峰值,从而能够自适应场景变化.所提出的算法充分利用相邻多帧之间的相关性,对模拟和实际红外图像序列非均匀性的校正效果突出.实验结果表明本方法能够精确估计非均匀性参数,收敛速度较快,几乎不残留鬼影.
A scene adaptive nonuniformity correction fixed pattern noise removal method is proposed.Based on multi-frame registration and global displacement estimation between adjacent frames, the mean square error function is iteratively calculated by the least mean square algorithm, meanwhile the error function Mean square error and cross-correlation peak, so as to adaptively change the scene. The proposed algorithm makes full use of the correlation between adjacent frames to highlight the effect of non-uniformity of simulated and actual infrared image sequences. Experimental results show that this method It can accurately estimate the non-uniformity parameters and has a fast convergence speed with almost no residual ghosting.