Single-observer automatic identification method of multi-target tracks

Abstract

The invention provides a single-observer automatic identification method of multi-target tracks, comprising the steps of (1) image addition, (2) targets connection, (3) directional filtering and (4) quantity filtering. The method of the invention is advantageous over the prior art in that: interpretation experiences of operators are summarized and used in a single-observer automatic identification of multi-target tracks, which means adding captured images and identifying authenticity of the tracks in a manner of combining the directional filtering and the quantity filtering, thereby avoiding the use of prediction models employed in other methods and providing a better adaptability in image interpretation, so that the ingle-observer automatic identification of multi-target tracks is carried out without a manual intervention in interpretation procedure.

Description

technical field [0001] The invention relates to a multi-target trajectory single-station automatic identification method, which belongs to the technical field of data processing. When it is applied to the continuous shooting test of multi-barreled weapons in the shooting range, the multi-target trajectory recognition situation. Background technique [0002] Measuring the trajectory of the flying target when the multi-barreled weapon is fired continuously is an important content of the range weapon system test. A prominent problem in photometric image processing is that it is difficult to distinguish multiple target trajectories appearing on the image. For example, in experiments such as the salvo of multi-barreled weapons and the throwing of submunitions, the image of each station may contain dozens or even hundreds of targets. Small targets, and the targets are similar in shape and size, and there may be occlusion between the targets. Since the targets to be measured in t...

AI Technical Summary

Problems solved by technology

[0006] The above method uses the prediction method to determine the target filter window, but the prediction model is difficult to adapt to the complexity of the measured image, so the adaptability of this method to the single-station automatic identification of multi-target trajectories is not ideal, and usually requires manual intervention by the operator, and cannot realize automatic identify

Images