Research On Infrared Polarization Image Processing Algorithm Based On FPGA

Infrared polarization imaging technology is a cutting-edge technology which further expands the advantages of polarization imaging to the field of infrared imaging.The new technology solves the problem,which cannot effectively recognize targets in complex background and camouflage radiation in traditional method.Therefore,the research on infrared polarization imaging has become a frontier and a hotspot in the field of infrared imaging technology.This thesis focuses on the research work based on the new domestic infrared polarization detector,the infrared polarization processing algorithm is studied based on the FPGA hardware platform.The main content of this thesis and the research results obtained are summarized as follows:1.Introduction to the basic theory of infrared polarization imaging.Firstly,the generation mechanism of polarized light is analyzed,which provides theoretical guidance for the research of this thesis and is the premise of this research,and the Stokes vector representation method of infrared polarized light used in this paper is introduced.At the same time,the architecture of the infrared polarization hardware imaging system built in this thesis is analyzed,and the key components used in the system are introduced.2.The infrared polarization interpolation algorithm is studied.The polarization array structure of infrared polarization detector is composed of multiple groups of micro grid super pixels,which will lead to the problem of instantaneous field of view error,and the infrared polarization image obtained by super pixel operation will reduce the image resolution.Therefore,an interpolation algorithm for infrared polarization mosaic elimination based on gradient weighted optimization is proposed in this thesis,and the algorithm can solve the problem of instantaneous field of view error,restore image resolution and obtain high-quality infrared polarization image.The results of the algorithm verification show that compared with the bicubic interpolation method which can have the best interpolation effect in the traditional interpolation algorithm,in the experimental scene,the interpolation algorithm proposed in this thesis improves the PSNR and SSIM indexes of polarization images by 19.25% and 19.34%.3.Research on infrared polarization noise reduction algorithm.The infrared polarization image is not directly output by the infrared polarization detector,which will cause image quality degradation in the process of obtaining the polarization image.Therefore,based on the traditional infrared filtering and noise reduction algorithm and by analyzing the characteristics of polarization imaging,this thesis proposes a secondary guidance correlation noise reduction method based on polarization phase correction.The experimental results of the algorithm show that the polarization image preprocessed by the algorithm has good evaluation results in both subjective and objective PSNR and SSIM indexes,and the noise reduction algorithm proposed in this thesis can be widely used in various scenarios.4.The infrared polarization enhancement algorithm is studied.Without any processing,the infrared polarization image has low contrast and poor visibility.Therefore,combined with the complementary advantages between infrared light intensity image and infrared polarization image,this thesis proposes a feature-level-based infrared light intensity map and polarization image detail enhancement algorithm,and studies the technical problems of how to realize each module in the algorithm on FPGA hardware platform,and the FPGA logic resources used by the hardware implementation of the algorithm only occupy 29% of the total logic units of the EP4CE55U14I9 chip.