Single-Pixel Pattern Recognition with Coherent Nonlinear Optics

TL;DR

This paper introduces a nonlinear-optics method for pattern recognition using single-pixel imaging and deep neural networks, achieving high accuracy in digit classification even with noise, demonstrating potential for fast, efficient image analysis.

Contribution

It presents a novel nonlinear-optics approach employing mode selective image up-conversion for pattern recognition, combining optical processing with neural networks for high accuracy.

Findings

Achieves 99.49% accuracy on MNIST with 40 projection modes.

Maintains 95.32% accuracy under strong noise conditions.

Demonstrates potential for real-time large-scale image classification.

Abstract

We propose and experimentally demonstrate a nonlinear-optics approach to pattern recognition with single-pixel imaging and deep neural network. It employs mode selective image up-conversion to project a raw image onto a set of coherent spatial modes, whereby its signature features are extracted nonlinear-optically. With 40 projection modes, the classification accuracy reaches a high value of 99.49% for the MNIST handwritten digit images, and up to 95.32% even when they are mixed with strong noise. Our experiment harnesses rich coherent processes in nonlinear optics for efficient machine learning, with potential applications in online classification of large size images, fast lidar data analyses, complex pattern recognition, and so on.