Image edge detection with a photonic spiking VCSEL-neuron

Area of Science:

• Neuromorphic computing
• Optical signal processing
• Image analysis

Background:

• Traditional edge detection methods can be computationally intensive.
• Neuromorphic computing offers potential for faster, more efficient information processing.
• Vertical-cavity surface-emitting lasers (VCSELs) are key components in optical communication.

Purpose of the Study:

• To demonstrate edge detection in digital images using an artificial optical spiking neuron.
• To leverage VCSELs for high-speed, neuromorphic image feature extraction.
• To explore the compatibility of this technique with existing optical technologies.

Main Methods:

• Utilized a VCSEL-based artificial optical spiking neuron.
• Employed convolution techniques for optical input pre-processing.
• Applied individual kernel operators and gradient magnitude for edge detection.

Main Results:

• Successfully detected target edges with varying directionalities in digital images.
• Achieved complete image edge detection via gradient magnitude.
• Demonstrated sub-nanosecond response times, significantly faster than biological neurons.

Conclusions:

• The VCSEL-based spiking neuron offers a powerful and rapid method for neuromorphic edge detection.
• The technique operates at the 1300 nm telecom wavelength, ensuring compatibility with optical communication and data centers.
• This approach represents a significant advancement in high-speed optical image processing.