A Novel Fuzzy Controller for Visible-Light Camera Using RBF-ANN: Enhanced Positioning and Autofocusing

Area of Science:

• Optics and Photonics
• Control Systems Engineering
• Artificial Intelligence

Background:

• Visible-light camera autofocusing and focal length fitting require high precision and speed.
• Traditional control algorithms often suffer from issues like gear backlash, affecting performance.
• Existing methods for focal length fitting and autofocusing may lack the necessary accuracy and efficiency.

Purpose of the Study:

• To develop an improved control algorithm for precise positioning of zoom and focus lens groups in visible-light cameras.
• To enhance autofocusing speed and focal length fitting accuracy simultaneously.
• To eliminate backlash caused by gear gaps in camera mechanisms.

Main Methods:

• Implementation of an improved Radical Basis Function (RBF) adaptive neural network (ANN) fuzzy Proportional Integral Derivative (PID) control algorithm.
• Utilizing the Levenberg-Marquardt iterative algorithm for focal length fitting.
• Application of an improved area search algorithm for autofocusing and backlash elimination.

Main Results:

• The improved RBF ANN fuzzy PID algorithm demonstrated a performance index improvement of over two orders of magnitude compared to traditional fuzzy PID.
• Settling time was reduced by 6.4 seconds, and autofocusing time improved to under 2 seconds (more than double the traditional method).
• The Levenberg-Marquardt algorithm showed fast convergence and high fitting precision, outperforming quadratic and exponential fitting methods.

Conclusions:

• The proposed improved RBF ANN fuzzy PID control algorithm significantly enhances precision and speed in camera lens control.
• The Levenberg-Marquardt and improved area search algorithms effectively address focal length fitting and autofocusing challenges, including backlash.
• This integrated approach leads to faster, more accurate autofocusing and clearer image acquisition in visible-light cameras.