| It is observed that the outdoor visual sensor devices in fields such as autonomous driving,safety monitoring,and military reconnaissance are severely affected by the haze weather,resulting in reduced image contrast,color distortion,and loss of detail information in the collected images.This greatly limits the extraction of effective information during subsequent image processing.A kind of improved dehazing algorithm based on dark channel prior(DCP)was studied in this paper.An FPGA logic circuit was designed to complete the mapping of the algorithm and the collection,storage,and display of video streams.An FPGA chip is ultimately used as the command control center and data processing center to realize a real-time dehazing system,effectively reducing the interference of haze.The main research content of this paper includes:(1)An improved fog removal method based on DCP is studied in this paper.The atmospheric light value is obtained by using the dynamic thresholding method based on the bright channel prior theory.The transmission propagation map is refined by using the Gaussian filter based on the gradient thresholding,and the compensation mechanism is applied to the transmission propagation map of the sky region through a segmentation approach.The experimental results show that the proposed method can produce images with moderate brightness,rich colors,and natural dehazing effect in the sky region,and it is also suitable for implementation on FPGA.(2)The FPGA mapping of the dehazing algorithm is completed.Standard setting and parallel analysis were performed on the dehazing algorithm studied in this article,and hardware computing techniques such as converting multiplication and division operations to shift operations and converting floating-point calculations to fixed-point calculations were used to optimize the algorithm.Verilog hardware description language is used to design the logical circuits of each module in the dehazing algorithm with a pipeline structure.Questa Sim software is used to simulate and verify the logical circuits of each module.(3)The real-time dehazing system is designed and the hardware platform of the real-time dehazing system is established to achieve video image acquisition,buffering,and real-time display.The video image is captured by configuring the CMOS register using the SCCB protocol.The AXI Interface is used to control the DDR3 SDRAM to achieve high-bandwidth transmission through burst read and write,completing the image data buffering.The TMDS transmission technology is used to achieve real-time display of image data.(4)Video dehazing is achieved by combining the logical circuit of dehazing algorithm and the hardware platform of dehazing system.Hardware single-frame image dehazing tests are completed through co-simulation of MATLAB and Questa Sim.Natural scenes with different haze concentrations are captured by a CMOS camera for hardware video dehazing tests.The test results show that the FPGA-based dehazing system could process video images at 60 fps in real time.The video image after defogging has clear details and high contrast. |
