Demonstration of an improved nonlinear fringe-adjusted joint transform correlator for effective object detection and tracking in high-noise conditions

Abstract Optical correlators often struggle to identify objects in noisy environments, limiting their efficacy in real-world applications. Optical pattern recognition can be seriously hampered by severely reduced visibility and contrast deterioration, which compromises its reliability and often leads to failure in identifying objects in the query scene. To the best of our knowledge, adaptive optical recognition systems have yet to achieve consistent reliability in highly noisy real-world environments. In this study, we propose an advanced non-linear fringe-adjusted joint transform correlator designed to achieve robust resilience against various noise perturbations, including scene noise, hazy inputs, target occlusion, and atmospheric turbulence. The central order self-intensity terms in the conventional joint transform correlator (JTC) operation are significantly stronger than the conjugate products and consume most of the energy in the correlation outputs. Here, we present a simple yet effective zero-order elimination technique combined with wavelet processing, which significantly enhances correlation performance, improves peak detection, and increases stability against noise perturbations. The target-tracking capability of the proposed scheme in the presence of strong background noise has been demonstrated through numerical simulations. Additionally, we demonstrate the hybrid experimental realization of the modified non-linear fringe-adjusted JTC incorporated with chirp-encoding, which we believe provides a new perspective to the field. The proposed approach bridges the gap between numerical simulations and optical implementations, offering a robust and efficient solution for object recognition in challenging environments. Comparative analyses with the previously reported methods demonstrate the efficacy of the proposed algorithm, establishing its potential for real-time applications in optical pattern recognition and image processing.