根据科学级CMOS(sCMOS)图像传感器开发的工程实际需求, 设计了基于FPGA的二阶多项式拟合的非均匀性校正算法。该非均匀性校正算法首先通过定标得出每个像素的响应曲线, 然后采用最小二乘法将sCMOS像素响应曲线参考平均响应曲线进行拟合, 得出校正系数并通过FPAG对像素值进行实时校正。通过处理后的图像非均匀性从17.1%降低到7.6%。该算法有效提高了sCMOS器件对高动态、非线性响应特性的适应能力, 并具备实时计算校正优点, 通过模块化设计可以方便移植和集成。

According to the actual demands of scientific CMOS (sCMOS) image sensor development project, a two-order polynomial non-uniformity correction fitting algorithm based on FPGA was designed. Firstly, calibration will produce a group of response curve of each column pixels. The least square method was used to fit the pixel response curves to the average response. The corrected coefficient will be saved in the on-board flash and correct pixel value real-time. The non-uniformity of the processed images decreases from 17.1% to 7.6%. This algorithm can effectively improve the devices’ adaptability to the characteristics of high dynamic and nonlinearity, and presents advantages of real-time calculation and correction, thus it can be easily transplanted and integrated through modular design.