Research On Flicker Noise Test Technology And Noise Reduction Method Based On Low Illumination CMO

With the improvement of image intensifier image quality,scintillation noise becomes an important factor that limits effective observation distance and affects image clarity.The traditional SNR test results are not completely consistent with the perception characteristics of noise by human eyes.Under the condition of the same SNR ratio,the scintillation noise also has a great difference,which requires quantitative analysis of the scintillation noise of image intensifier.This paper focuses on the scintillation noise test technology and noise reduction method.In this paper,the source and mechanism of scintillation noise are studied according to the structure of image intensifier,and the main forms of scintillation noise are classified.On this basis,a comprehensive scintillation noise measurement system based on low illumination CMOS is developed.The system is composed of plane uniform light source system,imaging optical system,CMOS imaging system,illuminance meter,digital power supply and testing upper computer.Among them,CMOS imaging system is the core of the design,and the design process includes system circuit design,FPGA logic design and upper computer design.The system can clearly collect images on the fluorescent screen under the condition of resolution256×256 and frame rate 300fps,which can meet the requirements of flicker noise testing.Aiming at the limitation of SNR ratio test and the characteristics of scintillation noise,two detection algorithms are designed in this paper.The dispersion coefficient of the noise caused by thermal emission in a large area of the image is tested to obtain the overall situation of the flicker noise on the fluorescent screen of the image intensifier,the uniformity of the noise distribution and the serious flicker defects.The ion feedback noise bright spot counting test is used for the highlighted and irregularly distributed noise in the image,and the number of ion feedback noise bright spot per unit time can be accurately calculated by multi-frame local extremum comparison algorithm.In this paper,a multi-frame noise reduction algorithm based on motion detection is proposed based on ICMOS camera.The moving object is detected by comparing the dynamically updated background image with the current image and different filtering strategies are adopted for the moving area and the background area.Experimental results show that at the lowest illumination of 2.03×10^-4lx,the proposed algorithm can effectively suppress a large number of flicker noises in the image and improve the image contrast.