Research Of Ocean Wind Direction Retrieval Algorithm Base On Marine Radar Images

The all-weather,all-round three-dimensional ocean dynamic environment monitoring has a vital role of human marine activities,marine exploration and research.Owing to convenience,small investment,and high resolution,navigation radar has become a hot topic in the study of scholars in overseas and domestic research status.At this stage,using the corresponding algorithm,we can derive the ocean dynamic parameters from the sea-clutter images collected by marine radar.This approach has an outstanding performance than other remote marine monitoring sensors.This thesis researches the technology of ocean wind-direction retrieval according to project of marine lab,and dedicates to giving comprehensive and precise ocean-wind information.This thesis reviewed the project significance and background,gave the summaries of radar ocean previous research results at home and abroad.Meanwhile,we gave an deep investigation on the technology of ocean wind-direction retrieval based on marine radar,designed two overall ocean wind-direction retrieval schemes using small scale wind streak characteristics.Finally,we test and verify the two different retrieval technology availability using real-time radar sequences.In order to remove the radar image radial attenuation,we design a least squares adaptive piecewise fitting algorithm to correct radar image according to radial attenuation characters.The algorithm application of least square piecewise fitting radar radial echo intensity,achieves adaptive judgment using the absolute error of fitting function value and real value.This algorithm uses multivariate function extremum to solve least-squares regression coefficient of Lagrange function,hence guarantees the continuity of piecewise fitting.The least squares using the second order as the highest order,means this approach can not only simplify the fitting function model but also improve the algorithm running speed.Results show that the fitting function value and radar radial echo intensity deviation is 0.83°,standard error is 22.19°,and correlation coefficient is 0.9986,matches the requirements of the radar correction.It verifies the effectiveness of the proposed relative radiation correction algorithm.Research of wind streak imaging mechanism is the foundation to retrieve ocean wind direction,but all the research achievements are based on the wind streaks from SAR at percent stage.Navigation radar and SAR have a bigger difference in terms of working principle,measuring method,which will cause the algorithm incorrect.This thesis begin with the influence of sea surface wind field modulation for radar imaging,theoretically analysis the key reasons of marine radar imaging characteristics generated from sea surface wind field.According to the sea surface wind field static characteristics,we extract small scale wind streaks from navigation radar image sequences,and ultimately identify the scale,cycle and axial direction of small scale wind streaks using the measured radar data.To overcome the problem that conventional local gradient method cannot adapt the small scale wind streak changing,we proposed an adaptive reduction operator method(ARM)to extract wind direction from marine radar image.Firstly,this algorithm use adaptive reduction operator to processing image in space,then a stable coefficient can be derived from probability diagram using reduction gradient direction information.Finally,we can extract the wind direction using optimal Sobel operator from the reduction image,and eliminate the inversion 180°ambiguous according to the quadrant of single point value.The correlation coefficient between real-time inversion results and reference ocean wind direction is 0.9956,with a standard deviation 7.62°,a bias-1.04°,owning a 58.3% improvement compared to conventional approach.The space domain image processing technique can corrupt small scale wind streaks,as well as lose a part of image,leading to lower inversion accuracy.To address this problem,an energy spectrum method(ESM)ocean wind direction extraction is proposed.The image will be convert into frequency domain energy spectrum diagram using 2-D Fourier transform,then small scale wind streaks can be derived after a band filter.Finally,ocean wind direction information is reflected by small scale wind streaks energy spectrum.The correlation coefficient between real-time retrieval results and reference sea wind direction is 0.99,with a standard deviation 8.29°,a bias-0.64°,verify the algorithm effective.To test the effective of the prior proposed sea wind inversion algorithm,we design a real ocean experiment based on marine radar at Pingtan island of Fujian province.The results shows that the EMS has a 45.7% improvement compared ARM,and the EMS correlation coefficient achieve 0.98,with a standard deviation 12.13°,a bias 1.68°.All indexes are better than ARM ocean wind inversion algorithm.Further,7s the average operation time shorten 7s for EMS than ARM.This experiment meanwhile analyzed measurement area,ocean surface wind speed,significant wave height and wave current influence on the ARM and EMS algorithm.The results show that EMS has a better engineering significance compared to ARM,although ARM has a little advantage in several experiment cases.