Design Of Simultaneous Polarization Imaging Defogging System Based On DSP

In recent years,polarization defogging technology has become a research hotspot in the field of video image processing.The use of polarization information for image dehazing can effectively improve image quality.It has been widely used in forest detection,forest landslide,fire risk warning and other forest detection and maintenance fields.The DSP platform has the advantages of small hardware,high computing power,and the ability to process data in parallel.The fusion of the DSP platform and the polarization dehazing technology can effectively improve the image processing speed and algorithm accuracy.Aiming at the problem that the amount of polarized image data is large and the defogging processing speed is insufficient in the current research,the paper studies the design of simultaneous polarization image defogging system based on DSP.The main research contents of the thesis are as follows:Firstly,the polarization imaging technology is studied,and the polarization camera device is designed to complete the four-way image information collection.Three of them are polarized image information of different angles,and one is the intensity image information,which provides data support for the subsequent defogging process.Secondly,the hardware platform of high-speed image processing system is designed and implemented.The platform integrates FPGA and dual DSP chips.It uses FPGA to process data in parallel and drives the network port to complete four-way image information synchronous transmission.It uses floating-point DSP to complete data processing and realize complex operations.The use of dual DSP "Pang Pong" alternate transmission mechanism to improve the overall processing efficiency of the system.Finally,the polarization fusion algorithm and the dehazing algorithm are optimized.Aiming at the high complexity of floating-point operations in polarization fusion algorithm,the addition and right shift are used instead of the multiplication and division operations that DSP is not easy to implement.The 16-bit look-up table method and right shift are used to realize floating-point numbers.Square root operation.Under the premise of ensuring the accuracy of the algorithm,The polarization defogging algorithm based on atmospheric attenuation model is implemented on the hardware platform.The algorithm uses the polarization degree information to combine the intensity image to achieve the dehazing effect,and optimizes the algorithm flow to effectively reduce the algorithm complexity.Experiments show that the quality of the restored image is improved,the definition is high,and the defogging effect is obvious.