Method for reconstructing safety and arming motion process by integrating Kalman filter and KCF

Area of Science:

• * Mechanical Engineering
• * Computer Vision
• * Signal Processing

Background:

• * Reconstructing the motion of safety and arming (S&A) mechanisms in fuzes is crucial for performance analysis and safety.
• * Traditional methods face challenges in accurately tracking dynamic targets, especially under occlusion.
• * Target detection and tracking offer a promising approach for motion reconstruction.

Purpose of the Study:

• * To develop an advanced target tracking method for reconstructing S&A mechanism motion.
• * To improve tracking accuracy and robustness in complex environments.
• * To validate the proposed method's effectiveness through experimental analysis.

Main Methods:

• * Fusion of an improved Kalman filter with a temporal scale-adaptive Kernelized Correlation Filter (AKF-CF).
• * Feature extraction using Adaptive Wave PCA-Autoencoder (AWPA) on grayscale images and Histogram of Oriented Gradients (HOG).
• * Integration of an occlusion-aware mechanism with Average Peak Correlation Energy (APCE) for Kalman-based prediction.

Main Results:

• * Demonstrated significant improvements in tracking accuracy and success rates on OTB50 and OTB100 datasets.
• * Achieved 92.50% accuracy and 68.10% success rate on OTB100, outperforming existing algorithms.
• * Reconstructed motion curves accurately replicated mechanical trajectories, proving robustness in occlusion.

Conclusions:

• * The proposed AKF-CF tracking method effectively reconstructs S&A mechanism motion.
• * The method exhibits superior performance in challenging tracking scenarios, including occlusion.
• * This approach offers a robust solution for analyzing dynamic mechanical systems.