Multi-scale feature fusion based on swarm intelligence collaborative learning for full-stage anti-interference object tracking

Abstract

Highly maneuverable object locating and tracking under anti-stealth and anti-interference environment is a challenging research subject. Since the locality-constrained linear coding and the cooperative representation are intrinsically linear model, this will cause the discriminant information is insufficient in object tracking. Therefore, a multi-scale feature fusion based on swarm intelligence collaborative learning for full-stage anti-interference object tracking is proposed in paper. This paper combines multiple features to describe the object so as to improve the representation performance of the single feature, and uses local constrained linear coding to obtain better classification performance; then use the swarm intelligence kernel function to extend the cooperative learning of local constraints to the kernel space, and derive the kernel sparse representation. Simulation results show that the improved algorithm has obvious advantages in real-time, stability and quantitative indexes, and is suitable for high-performance, low-cost video surveillance.