Computer Vision - Ballistic Characterization - Fig I

Ballistic Characterization for Artillery Weapons

Another useful application of the developed technique has been demonstrated in the field of ballistic characterization for artillery weapons. High-speed launch video analysis has previously been used by test engineers to verify important launch events such as rocket ignition, hood extraction, and proper fin deployment. Using the automated analysis method, it is also possible to estimate projectile’s speed, pitch and yaw angles (Fig.Ia), and even a spin rate (if a projectile carries fiducial markings or stripes painted on its nose). Figure Ib displays a segmented 155mm projectile and the keypoints used to represent its perimeter in an active-shape model.

a) Figure J-a  b) Figure J-b               

Figure I. Projectile attitude (a), and example of computer vision analysis of an artillery projectile (b).

Understanding aerial platform geometry in a camera frame allows employing fitting techniques that incorporate the epicyclic equations for projectile motion and estimating the history of a projectile pose accurate to 1/20th of a degree. The automated post-processing method is able to re-create the projectile trajectory immediately following a cannon launch (Fig.J). In addition, the algorithm can estimate several ballistic coefficients used in trajectory modeling and quantify the observed pitching motion in terms of the precession and nutation frequency, phase and amplitude. Having two cameras (Fig.K) enables a full inertial pose estimation.

Figure J. Analysis of artillery projectile attitude in a camera frame.

Figure J. Analysis of artillery projectile attitude in a camera frame.

a) Figure K-a   b)

Figure K. Analysis of three-dimensional dynamics of artillery projectile.

An extension of this project involved having more than just two cameras installed along a test article path. Figure L shows the images taken by four pairs of high-speed high-resolution cameras, and Fig.M demonstrates the result of applying the developed code to recover bullet’s six-degree-of-freedom motion. In this particular test the goal was to determine at which angle a bullet hits the target and reconstructed motion allowed to answer this question with a high probability.

Figure L. The bullet images as captured by four pairs of cameras with the known positions.

 

Figure L. The bullet images as captured by four pairs of cameras with the known positions.

 

Figure M. Determining bullet motion and estimating angle of target impact.

 

Figure M. Determining bullet motion and estimating angle of target impact.