Abstract:

Using TVcommandguided missile video images to determine the landing place of the missiles, which can be used to carry out accurate battle damage assessments, is a new kind of assessment tools. Image feature matching is the key step in determining the missile landing place according to video images. Considering the characteristics and the operational applications of missile video images, we improve the speeded up robust features algorithm in two aspects at the feature matching stage, which is based on the principles of accuracy and realtime performance. The first improvement is to restrict the extraction area of feature points and to define the restricted area factors. The second one is to limit the number of feature points via the algorithm threshold and random sample consensus algorithm. Experimental results show that when dealing with typical video clips, the improved algorithm can enhance the matching accuracy by 13.4%, and the matching time is reduced by 12.6% in comparison with the original speeded up robust features algorithm. Compared with the scale invariant feature transform algorithm, the matching time is enhanced by 74.9%. At the same time, the false matching points are eliminated effectively. Through the test simulation on three sections of videos, matching time of the improved algorithm is enhanced by 14.9% on the whole compared with the original algorithm, and it is generally suitable for feature points matching of video images.

Key words: video image feature extraction;image matching;speeded up robust features;TV-commandguided missile