Research Of Image Dehazing Algorithm And Realization Based On FPGA

In the foggy environment,due to the scattering of the atmosphere in the air,the captured image will be degraded and become blurred,resulting in reduced contrast of the image,and miss a lot of detailed information,thus affecting the work of various systems such as monitoring systems which rely on optical imaging systems.In order to remove the influence of fog and restore the detailed information of the image,it is of great research significance to use image dehazing technology to enhance or repair the image.With the continuous advancement of technology and the improvement of image clarity,the use of software to achieve image dehazing in PCs can no longer meet the requirements,and the development of FPGA has injected new vitality into it.However,when implementing image dehazing on FPGA,there are still problems in the bottleneck and complex implementation.In this paper,the dehazing algorithm and FPGA design are deeply studied,and the dehazing algorithm is improved.Finally,the image dehazing system based on FPGA is designed and implemented.Firstly,this paper introduces the image dehazing method based on image enhancement and the image dehazing method based on physical model.The advantages and bottlenecks of the two are analyzed.Finally,the dark channel prior image dehazing algorithm based on the physical model is selected to recover to get clear image.Because the soft mapping algorithm used in it is too complicated,and some of the step s are not suitable for the problems implemented on the FPGA,it will limit the processing speed and waste valuable hardware resources.Therefore,based on the full study of the original algorithm,this paper conducts further improvement.Secondly,based on the original algorithm,this paper reduces the complexity of the algorithm under the premise of ensuring the defogging effect.In order to make the transmittance calculation and atmospheric light estimation faster and more convenient in hardware,and save hardware resources,the calculation process is optimized and simplified.In view of the problem that the halo phenomenon and the soft sputum algorithm for removing halos are too complicated,other methods for refining the transmittance are emphasized.Two different schemes of guided filtering and combining the edge detection to process were implemented and simulated on the software platform.The results of subjective experience and objective evaluation were compared.Finally,the method of guided filtering is used to refine the transmittance to achieve fast dehazing.Through the above process,the basic flow of the image dehazing algorithm is determined.Finally,this paper implements an improved image dehazing algorithm on FPGA.With the FPGA chip as the core processor,under the premise of maintaining the defogging effect,the advantages of the FPGA are fully utilized,and the FPGA-based implementation optimization method is used to implement the system.The hardware logic of multiple modules such as dark channel acquisition,transmittance calculation,atmospheric light estimation and final image restoration is realized,which reduces the computational complexity and saves resources.The focus is on the minimum filtering module of the dark channel acquisition and the implementation of the guided filtering module to refine transmittance,and the boundary part of the image is designed.Finally,the function of each module was verified,and the actual foggy image was defogged.