Study On Recognition And Tracking Algorithm Based On Fast Object

Fast moving target recognition and tracking is one of the main research contents of computer vision. It has important applications in the military, intelligent visual surveillance system, industrial inspection and video image analysis, and other fields. Especially in the military field, being able to achieve real-time recognition and tracking of important military targets, such as aircraft, is of much importance for dominating the battle and improving the performance of the operational command system,In this paper, fast target recognition and tracking methods are discussed based on aircraft as a research object. First of all, the methods solving the problems in target recognition and tracking are introduced in details. Through comparison the real-time of common object detection algorithm, frame difference method has been selected to complete target detection, and morphological filtering method used to deal with the noise and empty problem. Then, with deep understanding and analysis of the mathematical basis of Gabor filter, we extract feature points in different scales and the direction from the response of Gabor filter with block-great value method. Comparing the energy value of these feature points neighborhood with the energy value calculated in the offline of the sample group, we complete the recognition of the target. Finally, the Mean-Shift Algorithm is introduced and improved with Kalman filter to complete the real-time tracking of targets. We also offered the scale and color template-updated strategy and methods to tackle the situation that the target was obscured. Experiments show that the algorithm can effectively detect, recognize and track target, is of high accuracy rate and the low mistaken recognition rate in the recognition process, is also of high speed and robustness in the process of tracking, and can overcome the problems such as scale changes and obscuration of targets, satisfy real-time processing requirements.However, we have lots of things to do to realize a target recognition and tracking system of robustness, accuracy and rapidity. In the end, improvement for the algorithm is described.