Arbitration algorithm of FIR filter and optical flow based on ANFIS for visual object tracking

• A new ANFIS-based FIR filter and optical flow arbitration (AFOA) is proposed.
• The AFOA integrates Kalman-like UFIR filter and optical flow with arbitration rules.
• The AFOA is a real-time visual tracking algorithm with robustness.
• The AFOA exhibits better performance than a single FIR filter and a single optical flow.
• The AFOA is superior to the Kalman filter and optical flow arbitration (KOA).

This paper proposes a new visual object tracking algorithm based on an adaptive neuro fuzzy inference system (ANFIS) for arbitration algorithm between a finite impulse response (FIR) filter and optical flow (OF). The proposed algorithm is called ANFIS-based FIR filter and OF arbitration (AFOA). The AFOA operates as an FIR filter for normal situations, keeping the computational cost low, and, when abrupt turns occur, converts to an OF to compensate for the inaccuracy of the FIR filter. An ANFIS-based arbitration algorithm constructs a mapping system from given inputs to an output using fuzzy logic and determines tracking mode of the tracking process between the FIR filter and the OF. The effectiveness of the AFOA algorithm is demonstrated by experiments employed on real-time video clips along with a comparative analysis with the ANFIS-based Kalman filter and OF arbitration (AKOA).