基于国产FPGA的红外图像处理算法实现

Realization of infrared image processing based on domestic FPGA

随着国内政策与资源的倾斜,国产化芯片取得长足进步,但核心器件与高端芯片仍需在探索与应用中发展。现场可编程门阵列(FPGA)作为信号处理与数据传输的通用芯片需要大量应用,另一方面,因国产红外探测器存在盲元、闪元等无效像元,使盲闪元剔除成为红外机芯必要的图像处理算法。因此采用国产化的FPGA进行基于时空滤波的红外盲闪元实时检测算法的移植,对FPGA进行逻辑优化和时序约束,并通过片上RAM和片外DDR3分别进行数据缓存。经算法移植,红外图像缺陷像元被有效抑制,盲闪元率从5‰降为0.73‰,验证了在国产FPGA上实现算法移植的可行性,对促进我国自主可控的核心芯片的工程应用与推广具有深远意义。

With the inclination of domestic policies and resources,localized chips have made great progress,but core devices and high-end chips still need to be developed in exploration and application.On the one hand,domestic infrared detectors inevitably have defect such as blind pixel,overheating pixel,and flick pixel.Therefore,the blind pixel elimination algorithm is an indispensable image processing algorithm in infrared detectors.On the other hand,Field programmable gate arrays used as a chip for signal processing and data transmission,require a large number of applications.The domestic FPGA is used to transplant the infrared blind flash element real-time detection algorithm based on space-time multi-dimensional filtering,logic optimization and timing constraints are carried out on the FPGA,and the on-chip RAM and the on-chip domestic DDR3 are used for data storage.Through experiments,the blind flash element of the infrared image after the algorithm transplantation is effectively suppressed,and the blind flash element rate is reduced from 5‰to 0.73‰.The feasibility of implementing algorithm transplantation on domestic FPGA is verified,which is of great significance to the engineering application of China′s independent controllable core chip.