阿尔法滤波自适应红外去盲元算法及硬件实现

Algorithm and hardware implementation of adaptive infrared blind pixelremoval with alpha filtering

为了解决红外焦平面阵列盲元问题对红外图像质量的影响,提出一种针对动态场景下的阿尔法滤波加时域累积去盲元算法,并对其进行硬件实现。通过阿尔法滤波获取图像估计,在单帧盲元分布判断上加入窗口均匀判断机制来防止图像边缘误检,采用时域累积的方法确定最终盲元分布并进行盲元补偿。在FPGA硬件实现上采用流水线结构提高处理实时性,采用多次时域累积迭代的方法提高盲元检测效果。通过软硬件实现验证该算法的有效性。相比其他时域累积方法能有效防止边缘过检。因此该硬件实现方法可以应用到实际红外成像系统进行实时处理,在长时间的时域累积中会有更好的盲元检测效果。

In order to eliminate the influence of blind pixels in infrared focal plane array on infrared image quality,an algorithm of alpha filtering and time⁃domain cumulative blind pixel removal in dynamic scene is proposed and its hardware implementation is conducted.The image estimation is obtained by alpha filtering,and the window uniform judgment mechanism is added to the single frame blind pixel distribution judgment to prevent image edge false detection.The final blind pixels distribution is determined by means of the time⁃domain accumulation method and the blind pixel compensation is carried out.A pipeline structure is used for FPGA hardware implementation to improve the real⁃time processing.The multiple time⁃domain cumulative iteration method is used to improve the effect of blind pixel detection.The effectiveness of the algorithm is verified by the implementation of software and hardware.In comparison with other time⁃domain accumulation methods,it can effectively prevent edge over⁃detection.Therefore,the hardware implementation method can be applied to the actual infrared imaging system to conduct real⁃time processing,and it will have a better effect of blind pixel detection after the long⁃term time⁃domain accumulation.