Research On The Characteristics Correlation Tracking Algorithm For Imaging Laser Radar

Target tracking algorithm is one of the key technology in the radar system. Steak tube imaging radar is a novel radar which can not only get the distance image but also intensity image. Aiming at the feature of steak tube images, we research on the application of target tracking algorithm in the radar tracking imaging system, trying to find a suitable algorithm for this system.Compared to the normal tracking algorithm, correlation tracking algorithm has some advantages such as a strong anti-jamming performance, robustness, and the stability even in a complicated situation. However characteristic coherent tracking algorithm only focuses on the characteristic information, so that it has a few amount of the calculation, a fast calculating speed and more stable tracking tha n the normal one. There are three sections in this thesis: theoretical analysis of characteristic coherent tracking algorithm, numerical simulation of characteristic coherent tracking algorithm, and the experiment of it.In the first part, we analyze the traditional coherent tracking algorithm, and introduce the common characteristic point in the image tracking as well as three key parts of coherent tracking algorithm- matching measure, template upgrading strategy and the searching strategy.In the second part, we obtain successive frame images of radar by numerical simulation. We try to simulate the radar images acquired in the air. Then we do the experiment to the images we obtained. We utilize two kinds of coherent tracking algorithms, SUSAN corner and edge feature respectively, and choose one of them through comparison. Then we improve the algorithm, select an optimal size of template and complete the characteristic coherent tracking of radar images obtained by simulation. Experiment results show that the improved algorithm reduces the error.In the last part, we obtain the real radar images by steak laser radar. We can get consecutive frame images by interpolation amplification and equal proportion cutting. Then we track the images by the improved algorithm in the second part. Especially, we utilize the distance information coming from distance image of the laser radar combined with the intensity image to complete the image tracking by characteristic coherent tracking. Experimental results show that the edge characteristic coherent tracking with a high accuracy is more suitable for tracking the laser radar image.