Research And Implementation Of Online Image Dehazing Algorithm Based On Zyn

For daily image processing applications,image defogging techniques have received a lot of attention as a key node of pre-processing.In the process of real-world application of image defogging algorithm,due to the generation of bad weather such as haze,there are a large number of particles in the air and atmospheric light scattering effect,which in turn makes the outdoor scene low visibility,and when the acquisition equipment for image and video capture,it will lead to blurred effect and low contrast of the acquisition source.Due to the complexity of the application environment of the capture device,the complexity of the defogging algorithm is high and it is difficult to meet the real-time requirements.To address this issue,this paper implements a system acceleration design using a combination of hardware and software on the logic and processor sides of the fully programmable system-on-a-chip to meet the realtime requirements.The system designed in this paper is improved in three aspects: algorithm,hardware and software parallel optimization.In view of the high complexity of the defogging algorithm,this paper proposes an improved defogging algorithm through the theoretical study of the dark channel a priori defogging algorithm based on the physical model and the image enhancement algorithm based on the nonphysical model.The improved algorithm uses image local adaptive threshold segmentation to get a more realistic atmospheric light value,color space conversion for image enhancement of luminance values to get the dark channel map,and combines the calculated transmittance with guided filtering for refinement to get a reduced fog-free image.Compared with the commonly used defogging algorithm,this improved algorithm has advantages in both computational complexity and defogging efficiency.At the system software level,this paper validates the functional simulation of each module and improves the software operation efficiency by optimizing the design of data structure and planning the parallel logic of algorithms.At the hardware level,this paper utilizes Zynq’s unique architecture to combine hardware programmability at the PL side and software programmability at the PS side,and uses AXI4 for high-speed communication,clock tree and reset tree optimization design,etc.to implement hardware design and optimization of modules such as hardware acceleration of the defogging algorithm,video capture,and HDMI output of processed data at the PS side and PL side,respectively.By performing physical platform verification,the improved algorithm in this paper can meet the real-time requirements and maintain the effectiveness of the defogging algorithm in the embedded hardware platform.