Doppler Weather Radar Echo Images’ Clutter Suppression And Storm Clouds’ Segmentation And Tracking

Since always, weather is in close relationship with people’s production and life, disaster weather is particularly affecting production of human sociaty and living. Thus, people’s life and the loss of property can effectively be avoided by accurate forecasting of the disaster weather. So disaster weather’s accurate identification and tracking become the most important thing of weather forecasting. A series of methods proposed by this paper play an important role in image denoising, image segmentation and target tracking under Doppler weather radar’s circumstance.For radar image denoising, the purpose is to remove general ground clutter and anomalous propagation clutter(AP clutter). Doppler weather radar’s clutter suppression based on the texture feature of Gray-level Co-occurrence Matrix(GLCM) is proposed, and a series of the texture characteristics, including contrast, energy, entropy, characteristics of inertia, are tested on a large mount of experimental datas. Tests show the conclusion that inertia can effectively remove the radar image noise clutter. Image denoising is the basis of storm cell segmentation.For the storm cell segmentation, level set method and relevant image segmentation is applied to the weather field first time in history. This method is apt to extract nonrigid targets from images whose background are unstable, just because it can effectively deal with the condition of merging, division and the appearing and disappearing phenomenon. The method of stom cells segmentation makes the target edge smooth, highly consistent with actual situation. It provides a strong basis for the extraction of cloud characteristics, and for the storm cell tracking.For the storm cell tracking, storm cell targets detected in the current time frame are labeled first, and then are matched with targets in the previous frame. Through matching we can detect changes of targets in targets’ evolution and realize tracking algorithm. The algorithm save the targets evolution status tabel through which we can find history information of storms’ evolution. Finally the tracking algorithm is assessed by LSBEM algorithm, results prove that storm cell tracking algorithm in this paper is effective.The use of doppler weather radar reflectivity factor data, radial velocity data and vertical integrated liquid data, the combination with some characters of strong convective weather and background, and the using of image processing, statistical pattern recognition and numerical method of partial differential equation, the above all together effectively solved the strong convective weather disaster image denoising, image segmentation and target tracking problem.