A novel tracking method called AKF-CF, integrating improved Kalman filtering with temporal scale-adaptive KCF, has been developed to reconstruct the motion processes of safety and arming (S&A) mechanisms used within fuze technology. This advancement addresses the challenges posed by conventional methods, particularly due to the complex environmental forces these fuzes undergo during ballistic flight.
The significance of this research lies not only in enhancing tracking accuracy but also ensuring the operational reliability of S&A mechanisms, which are integral components of projectiles. Existing methods for analyzing these mechanisms often depend on costly physical tests, as they are typically destroyed upon impact. The newly proposed AKF-CF method leverages high-speed video capture and image processing techniques to effectively monitor and reconstruct motion trajectories, making it feasible to conduct safety evaluations without physical destruction.
Key innovations within the AKF-CF method include the extraction of grayscale images and Histogram of Oriented Gradients (HOG) features combined with the Adaptive Wave PCA-Autoencoder (AWPA) for efficient feature fusion. It also implements dynamic adjustments to the tracking box size through bilinear interpolation and hybrid filtering. An occlusion-aware mechanism using average peak correlation energy (APCE) has been integrated to maintain tracking accuracy even when the target is temporarily obscured. These features address two major challenges: non-linear motion patterns typical of mechanical systems and the difficulty of tracking during occlusion and motion blur.
Experimental results indicate the AKF-CF method shows substantial improvements, achieving 92.5% accuracy and 68.1% success rate on the OTB100 dataset, surpassing the performance of existing algorithms like ACSRCF. This performance enhancement is attributed to the method’s ability to adaptively adjust tracking parameters and its innovative feature extraction methodology. The AKF-CF algorithm effectively reproduces mechanical trajectories, aligning closely with actual motion paths observed during experimentation.
The study also highlights the miniature operational range of the S&A mechanism, which is approximately 65 μm, emphasizing the precision required for effective tracking and analysis. Through extensive testing across five distinct datasets, including the self-built S&A dataset and popular benchmarks like OTB50 and OTB100, the proposed methodology eliminates traditional testing limitations by offering reliable reconstruction of motion trajectories.
Notably, the mechanism demonstrated robustness against occlusion, maintaining accurate tracking through dynamic environments with minimal error. One of the most notable findings presents the motion curve obtained through AKF-CF closely mirroring the real curves with just a 6.67% error margin. The encompassing abilities of this technology provide insights not just for mechanical motion analysis but pave the way for future applications across precision tracking domains.
Conclusively, the AKF-CF tracking algorithm is positioned as a pivotal innovation, marrying mechanical analysis and computer vision to facilitate improved analysis and safety evaluations of S&A mechanisms. With the promising accuracy rates and robustness observed, future research could explore broader applications of this approach, potentially enhancing safety mechanisms across various ballistic systems. Such advancements may significantly influence the field, reducing reliance on destructive testing methods and allowing for innovation within the area of fuze technology.