Research On Haze Removal Algorithm Based On FPGA For Stereo Vision Camera

With the continuous application of stereo camera in the field of computer vision,the image plays an important role as the main way of machine vision observation and perception of the world.Images collected in haze and other weather can not reflect the real scene information,reducing the value of the image,which greatly affected the image-based computer vision in security surveillance,driving aids,drones and other fields application.Therefore,it is necessary to study foggy images taken by stereo camera and realize the restoration of image quality.In this paper,we take the stereo camera as the object of study,aiming at improving the imaging quality of the foggy image,focusing on the detection of fog,atmospheric light optimization,transmittance solving algorithm,and FPGA as the acceleration platform of the stereo camera de-fogging algorithm.The main work of this paper is as follows:1.Fog detection and identification.Taking the physical model of the atmosphere as the research starting point,the dark channel algorithm is used to process the image and make in-depth research and analysis to realize the automatic detection of the mist.2.Single-image image de-fog algorithm.Based on the general idea of optimizing the contrast algorithm,this paper makes an in-depth research and analysis of the single-image de-fogging algorithm,studies and optimizes the quad-tree algorithm based on the atmospheric light value,and studies the sky segmentation algorithm to optimize the transmittance,so as to achieve better Image to fog effect.3.Stereoscopic camera to fog algorithm key parameters consistent research.Explore the atmospheric light value fusion scheme of the left and right images,further study the transmission solution based on the SGM matching window and the transmission optimization scheme based on the ORB feature points,and analyze the refinement scheme based on the edge-preserving filter transmittance to realize the pixel level Restore effect.4.FPGA-based stereo camera to fog acceleration program.Based on the algorithm of image de-fogging on the PC side,aiming at the parallel flow characteristics of FPGAs,the algorithm reconstruction is designed and studied.The realization process of atmospheric light module,transmittance module and lead filter module in the FPGA is studied and designed,and using timing simulation and analysis to achieve the reliability of FPGA algorithms.