基于三次多项式拟合的快速非均匀校正算法

Fast Nonuniformity Correction Algorithm Based on Cubic Polynomial Fitting

由于两点法校正的动态范围窄,而多点法系数存储量大,所以研究出一种温度范围宽、能够适应弹载计算机有限的资源和计算能力,实时、高精度非均匀校正算法是一项迫切的研究任务。提出了一种基于三次多项式拟合的快速算法,使校正的动态范围扩展到正常成像的全部温度范围,而校正系数的存储空间比多点法存储系数要少得多。对于320×256像素的16位图像,计算速度达到0.6 ms/帧,完全满足实时性要求。试验证明了算法的优秀性能。

Two-point-correction has a narrow temperature dynamic range and multi-point-correction has huge storage burden. It becomes urgent to develop an algorithm which has the virtue of wide temperature range, low computation burden, and high correction precision. A correction approach based on cubic polynomial fitting is proposed. This algorithm extends the temperature dynamic range to the whole temperature of detector imaging, but the storage memory cost of cubic polynomial fitting algorithm is greatly shorter than that of multi-point-correction algorithm. For the image of 320 ×256 16bit pixels, the algorithm performance is 0.6 ms/frame, which satisfies the real time application well. Experiments illustrate the virtue of the proposed approach.