Video Image Dehazing Algorithm Research And Hardware Implementation

Foggy weather conditions have a greater impact on visibility,greatly reducing the quality of images collected by imaging equipment and affecting computer vision algorithms.For example,driving recorders and monitors require high-quality images or videos.Therefore,defogging and clearing the collected images or videos is of great research significance.This thesis studies the image defogging algorithm based on non-physical model and the image defogging algorithm based on physical model,focusing on the dark channel prior defogging algorithm.Aiming at the problems of white halo on the edge of the image depth and distortion of the sky area generated by the dark channel prior,this thesis proposes an image defogging algorithm based on the dark channel prior and optimized automatic level.This algorithm introduces an improved dark channel prior to the transmission The transmission rate is oriented to filter,refine the transmittance propagation map,improve the situation that there is no defogging at the edge,set the lower threshold of the transmittance to avoid the color gradation in the highlight,and use the weighted least squares smoothing filter to retain the image edge in post-processing.Information,reduce the ringing effect to optimize the automatic color gradation to enhance the image,make the image more natural,prevent artifacts and color dim.The method of combining qualitative and quantitative analysis is used to evaluate the algorithm of this thesis and several defogging algorithms in recent years.It shows that the algorithm of this thesis can keep the image color quality while keeping the details clear,and the defogging effect is better.The hardware implementation of this thesis takes EP4CE10F17C8 N of FPGA Cyclone IV as the main control chip.First build the overall video image capture buffer display hardware framework.Finally,the video image defogging module is connected,which is completed by the rgb?dark module,the Transmittance?A module,and the defogging module in a pipeline manner.The hardware optimization includes:improving the hardware minimum filtering algorithm,proposing a fast minimum filtering method suitable for FPGA to obtain dark channel images;improving the coarse transmittance estimation method to streamline FPGA operations.The final verification results show that this thesis has completed the primary defogging hardware system suitable for FPGA,and achieved the video image defogging effect.