Imaging Ladar Target Tracking Under Correlated Estimate And Amelioration On Tracking Performance Based On Centroid Estimate

Having high angle and range resolution, the imaging laser radar is capable to produce range and intensity image of the target at the same time, which can form high resolved angle-angle-range-intensity images. And it has bright foreground in applying in many areas. Image tracking is the hardcore of the laser radar. The whole ladar system is depending on it. Right now, centroid estimator is widely used in laser radar image tracking system. This estimator is quite easy to build in hardware and it has high run speed. Correlated estimator is also one of the most important estimators in image tracking. It's relatively more complicated and slower compared with centroid estimator, rarely applied in imaging laser radar tracking system. But along with the development of the hardware, the high- accurate correlated estimator is quite sure to become one of the most important tactics in laser radar target tracking.Researches and Experiments which this dissertation presented can mainly be divided into two parts: the part of correlated estimator and the part of centroid part.In the first part, traditional correlated estimators are ameliorated, so that they can be put into the process of laser radar tracking operation. The author uses a new criterion to judge whether the target is overflow and whether the template has to expand. Later, emulational tracking experiments are carried out under kinds of scenes, including buildings and vehicles. Tracking results and tracking errors are given at last. The result has shown that the ameliorated algorithm proposed has the high accuracy and speed, and is able to expand self-adaptively as the target become larger. Moreover, the author tried to use some image processing technics before the images were passed to the single frame estimator. The purpose of image pre-processing is to reduce noises and enhance images. The results have proved that tracking errors of the multi-frame images after image pre-processing are lower than those do not.In the second part, centroid estimator is analyzed statistically. Important conclusion then is drawn from the analyse, that is the estimate of centroid estimator is a biased estimate, and contains signal depended noise. Then author expatiated on the shortage of the pointing optics system. This shortage, which is named lag phenomenon, posed a significant negative effect on the radar tracking performance. To resolve this problem, referring kalman signal processing, the author designed an adaptive filter. The filter can be divided into prediction and update. In first part, former target's centroids are brought to a fusion, an prediction of the target's centroid in the next frame is made. The latter part updates the prediction and some parameters, by using forward predictive error and single frame estimate. From the experiments given later, the author concludes that lag phenomenon has been suppressed remarkably by this filter. By fixing the target in the center of the field of view, this filter keeps the tracking covariance in a low level, thus reduced the tracking error. So this filter is quite helpful to current laser radar tracking system to improve its precision and tracking ability.