Research On Uncooled Infrared And ICMOS Image Fusion Technology

In the field of low-light-level night vision,low-illumination CMOS has light weight and can work 24 hours,but its signal-to-noise ratio is low under moonlight level illumination.ICMOS is more suitable for night work because of its high gain characteristics provided by image intensifier.Infrared images are totally independent of time.Although they lack details,they could provide target information.In order to improve the performances of night vision and make the hardware smaller,an image fusion system is developed using domestic third generation image intensifier,NSC1105 CMOS and UL04472 uncooled infrared focal plane.Furthermore,the real-time image fusion algorithm of uncooled infrared and ICMOS based on FPGA is studied and implemented in this paper.Firstly,based on the existing ICMOS hardware solutions of the research group,this paper designed uncooled infrared and ICMOS image fusion system hardware platform using FPGA,then drove these two image sensors at the same time and debugged the hardware module.Secondly,the preprocessing algorithm and dual source image registration of uncooled infrared and ICMOS images are studied and implemented.The ICMOS image preprocessing mainly eliminates Gaussian noise under night illumination.The infrared image preprocessing includes non-uniform correction and median filtering,both images are histogram equalized to enhance image contrast;the registration process is mainly to compress and translate infrared images.Thirdly,the image fusion algorithm suitable for FPGA hardware platform is studied,and the wavelet transform pyramid image fusion algorithm is implemented in the system.Finally,the imaging experiment is carried out and the imaging effect of the system is evaluated.The system imagines at night,taking the low-light level image as the background,and the infrared image to enhance the target.Compared with average fusion algorithm,the experimental results show that the algorithm proposed in this paper can combine infrared and ICMOS images in real time and the effect is better.