Research On Moving Small Target Detection In Clutter Ground Environment

At present, optical imaging guidance technology has become an important way in precision strike. Detection of small target in ground environment is one of the key technologies of optical imaging guidance. Because of the complexity of the scene, targets can be easily overwhelmed by noises and background. At the same time, the imaging platform's irregular motion will make the target locations very unstable. All of these will increase the difficulty of the target's detection. In this article, we do some research of how to solve these problems. The main is as follows:1. Two feature matching methods to eliminate the influence of the imaging platform's irregular motion are proposed. One is using EMDHT algorithm for motion detection after picking-up the images'edges, and the other is using corner points'matching method. In the experiments we compare these two methods, and the results show that corner points'matching method can estimate the motion parameters of the imaging platform more effectively. Then we can get the differential images after motion-eliminated.2. To weaken the interference of the false targets and noises, a self-adaptive threshold partition method is used to convert the differential gray images into black-white ones. After that, we use the two-dimensional Gabor filter for further targets'choosing. And the experiments show that our plan is effective and the potential targets can pop out while eliminating nearly all of the noises.3. A neighborhood-connected target's detection algorithm using Kalman filter is discussed. Here targets moving characteristics are fully used. We estimate the targets'state in the next frame, compare the observation and the estimation, and get the true state values of the targets through error analyses. And the experiments tell that the noises'interference can be easily eliminated. At the same time, the accuracy and robustness of detection is improved highly.