An Automated Method for Computer Vision Analysis of Cannon-Launched Projectile Video

摘要

This paper describes a cost effective automated methodology to analyze launch video of artillery projectiles. Image processing and computer vision techniques are used to segment and classify the projectile shape in each video frame. Within minutes of being fired, the initial position, velocity, and orientation history of the projectile in three-dimensional space is determined at the gun site. An overview of several standard methods used by the Army to characterize pitching and yawing motion of projectiles is included, as well as a discussion of the limitations of these methods. Results from real artillery testing using the automated video analysis method are validated through comparisons to results measured using conventional techniques.