2019 IEEE International Instrumentation and Measurement Technology Conference (I2MTC)

1 / 2018-12-07 11:53:27

Sparse Representation Target Tracking Based on Visual Attention Mechanism

Attention mechanism; Target tracking; Sparse representation; Salient model

Aiming at the problem of low real-time and accuracy of traditional target tracking algorithm in complex scenes, this paper proposes an improvement. Inspired by human visual system, a sparse representation target tracking algorithm based on visual attention mechanism is proposed. The sparse representation coefficients of different modal features of each object are calculated by sparse representation of different modal features, and the sparse reconstruction error is calculated. Then the visual attention mechanism is integrated into the multi-modal sparse algorithm, and the target features are updated by using the working principle of human cerebral cortex structure and information attention mechanism for reference. Motivation drives oneself to learn new environment or new things constantly, adjusts the learning rate, reduces the matching time of sample characteristics, thus guarantees the real-time performance of optimization methods.