Design And Implementation Of Fog Visibility Detection System Based On Image Analysis

Foggy weather has a great influence on traffic safety, to this situation, this paper conducted a study around visibility detection principle and fog level detection principle, and designed a fog visibility detection system, which includes a front device based on DSP and a backend server computer. The front DSP device detects the farthest visible point from the real-time traffic video images, then calculates the real visibility distance using the camera calibration parameters which have been calculated in the server computer. For several scenes, precise visibility distance could not be quantitative calculation, to this situation, the front DSP device estimates the fog level using the real-time traffic video images. Here are the main works of this paper:(1)After analysing the advantages and disavantages of the mainstream digital image processing systems, this paper designed a fog visibility dection hardware system, which uses DSP as main processer chip. In order to make this DSP system be able to process complicated algorithm independently, this paper designed external program memory and data memory modules, and also video signal decoding and encoding modules for the DSP processer. In addition, this paper designed a network transceiver module for the DSP system, so that the DSP system can connect with the remote server computer through the internet. In the end, the FPGA circuit module is designed for the system, to take charge of logic control of the system, and reserve for possible function expansion in future.(2) This paper studied a fog visibility detection method based on camera calibration, which establishes the conversion model between actual distance and image distance using the mark lines in highway road surface, then calculate the real visibility distance using this distance conversion model and the coordinates of the farthest visible point in image. As the experimental results shows, this method has good accuracy and practicability.(3)This paper proposes a fog level detection method based on image segmentation, which detects the vanishing point position of the highway lane line, then find out the boundary line of fog and road surface using the method of image segmentation, the fog level is estimated according to the level of road surface covered by fog.(4)This paper designed the software of fog visibility detection system, which includes DSP program and server computer program. The DSP program detects fog visibility and fog level, in addition to communicates with server computer through internet. The server computer program calculates camera calibration parameters and communicates with front DSP system. Both of DSP program and server computer program involve thread synchronization and communication, this paper designed appropriate thread scheduling rules, to realize the proper scheduling between threads and enhance the reliability of program. The test result shows that the program achieves the intended function, and has good stability.