Research On Dynamic Detection Algorithm Of Precipitation Clouds

Severe convective weather on a small to medium scale is often accompanied by heavy precipitation and flooding,threatening the personal health and property safety of the people.Monitoring precipitation weather and conducting effective forecasts can gain time for flood fighting and disaster relief.In the current research,most of them use Doppler weather radar to monitor precipitation clouds.The Doppler weather radar is expensive,bulky,and the location of the infrastructure is complex,making it difficult to achieve widespread installation.Although the X-band ship navigation radar is not as good as the Doppler weather radar,it can monitor the dynamics of precipitation clouds within a range of 100 kilometers.Compared with the Doppler weather radar,the ship navigation radar is cheap and small,and can be used to monitor the dynamics of precipitation clouds in a small area.Due to these characteristics of ship navigation radar,the use of ship navigation radar instead of Doppler weather radar to track precipitation cloud has a good research value.This paper mainly studies how to use the ship navigation radar and the theoretical methods in computer vision to realize the detection and tracking of precipitation clouds.The research in this paper is mainly divided into the following sections.(1)Radar image processing.The image morphology processing method was studied for the small isolated point processing after precipitation cloud detection.The radar image near-range static strong echo is removed and the geographic information is superimposed on the echo image.(2)Identification of precipitation clouds.The classic target detection algorithm(optical flow method,interframe difference method,background subtraction method)is studied.The advantages and disadvantages of the three detection algorithms in the detection of precipitation clouds are compared and analyzed.Based on the background subtraction method,the hybrid Gaussian background modeling algorithm is studied.The effect of the hybrid Gaussian background modeling algorithm is analyzed experimentally.(3)Tracking of precipitation clouds.Common target tracking algorithms(model-based tracking algorithm,contour-based tracking algorithm,region-based tracking algorithm,feature-based tracking algorithm)are studied.The applicability of four tracking algorithms in precipitation cloud tracking is analyzed.Based on the feature matching based tracking algorithm,combined with the characteristics of the precipitation cloud target,the area and moment invariants are used as the tracking feature quantity.The experiment analyzed the tracking effect of precipitation cloud.(4)Based on the identification and tracking algorithm of this paper,the ship navigation radar precipitation cloud detection and tracking software is completed according to the experimental requirements.The software designed radar video reading module,precipitation cloud detection module and precipitation cloud tracking module to realize video path selection and display function,radar original video display function,precipitation cloud detection display function,and precipitation cloud dynamic area range display.Function,precipitation cloud tracking track display function,video playback pause and end function.The software interface is designed according to the module function.