Retinomorphic Optical Spiking Neuron

---
name: retinomorphic-optical-spiking-neuron
description: "Hodgkin-Huxley-based optical spiking neuron (OSHN) methodology for energy-efficient retinomorphic vision processing and camouflaged object detection. Uses 2D anti-ambipolar phototransistor in subthreshold regime, emulates retinal center-surround receptive fields, achieves sub-picojoule spike energy."
---

## Retinomorphic Optical Spiking Neuron

Methodology from arXiv:2606.00818 (Submitted 30 May 2026). Authors: Srilagna Sahoo, Adwaaiit Pande, Kartikey Thakar, Shubham Sahay, Saurabh Lodha.

## Core Innovation

**Optical Spiking Hodgkin-Huxley Neuron (OSHN)** — a retinomorphic hardware device that implements biological Hodgkin-Huxley neuron dynamics using a 2D anti-ambipolar phototransistor operated in the subthreshold regime. The device emulates multiple retinal preprocessing functionalities in a single hardware element, enabling energy-efficient event-driven vision systems.

## Key Specifications

| Metric | Value |
|--------|-------|
| Energy per spike (dark) | 0.9 pJ |
| Energy per spike (480 nm) | 2 pJ |
| Energy per spike (800 nm) | 24.5 pJ |
| Spiking rate range | 0 - 2 kHz |
| Response time | 4.2 µs - 1.25 ms |
| Human retina response | 30 ms - 60 ms |

## Mathematical Framework

### Hodgkin-Huxley Optical Implementation

The OSHN maps the classic Hodgkin-Huxley equations to optical-electrical dynamics:

```
C_m * dV/dt = -I_Na(V) - I_K(V) - I_L(V) + I_photo(λ, I_light)
```

Where:
- `I_photo(λ, I_light)` is the wavelength- and intensity-dependent photocurrent from the anti-ambipolar phototransistor
- The anti-ambipolar transfer characteristic provides the **non-monotonic I-V curve** required for spike generation (analogous to sodium channel activation-inactivation)
- Subthreshold operation minimizes power consumption

### Center-Surround Receptive Field (CSRF) Emulation

OSHN implements retinal antagonistic center-surround receptive fields:

```
Response_CSRF = w_center * I_center(λ, t) - w_surround * I_surround(λ, t)
```

- Operates at a single wavelength (480 nm OR 800 nm) with varying intensities
- Enables spatial edge detection and contrast enhancement in spiking domain

### L-M Cone Opponency

Emulates midget ganglion cell color opponency:

```
Response_opponent = k_L * I_800nm(t) - k_M * I_480nm(t)
```

- Enables spectral discrimination in spike timing
- Forms basis for color-aware spike encoding

## Usage Patterns

### Pattern 1: Retinomorphic Spike Encoding

Use OSHN principles to convert optical scenes into spike trains:

1. Map pixel intensities to phototransistor input current
2. Apply Hodgkin-Huxley dynamics to generate spike timing
3. Energy-efficient encoding: 0.9-24.5 pJ per spike (vs. µJ-mJ in conventional sensors)

### Pattern 2: CSRF-Augmented SNN for Object Detection

Build spiking neural networks with retinal preprocessing:

1. Apply center-surround receptive field emulation to input
2. Feed preprocessed spike trains to downstream SNN layers
3. Achieves significant accuracy improvements:
   - +4.4% on FMNIST (over conventional SNN)
   - +10.4% on COD10K (camouflaged object detection)
   - +28.4% on synthetic camouflaged datasets

### Pattern 3: Visual Adaptation for Dynamic Range

Implement biological visual adaptation:

1. Adjust spike threshold based on ambient illumination (at 480 nm)
2. Prevent system saturation in high-intensity conditions
3. Maintain sensitivity across orders of magnitude

## Implementation Guidelines

### Hardware Requirements
- 2D anti-ambipolar phototransistor (e.g., MoS2/WSe2 heterostructure)
- Subthreshold bias operation for minimum power
- Wavelength-selective illumination (480 nm and 800 nm bands)

### SNN Architecture Design
- CSRF preprocessing layer → standard SNN classifier
- Event-driven processing (only active pixels generate spikes)
- Compatible with existing SNN frameworks (SpikingJelly, Norse)

### Energy Budgeting
- Dark conditions: 0.9 pJ/spike (ultra-low baseline)
- Mid-wavelength (480 nm): 2 pJ/spike (typical operation)
- Long-wavelength (800 nm): 24.5 pJ/spike (higher sensitivity mode)

## Activation Keywords
- retinomorphic optical spiking neuron
- OSHN methodology
- Hodgkin-Huxley optical neuron
- anti-ambipolar phototransistor spiking
- center-surround receptive field SNN
- camouflaged object detection SNN
- event-driven vision system
- optical spiking neuron
- 视网膜类光脉冲神经元
- 光学霍奇金-赫胥黎神经元
- 反双极性光电晶体管
- 中心-周围感受野
- 伪装目标检测

## Related Skills
- `neuron-photonic-spiking-laser` — photonic spiking neurons using VCSELs
- `aquatic-neuromorphic-optical-flow` — SNN for optical flow estimation
- `spiking-neural-network-analysis` — SNN paper analysis patterns
- `spikingjelly-framework` — SpikingJelly SNN framework guide

## References
- **arXiv:2606.00818**: "A Retinomorphic Optical Spiking Neuron for Camouflaged Object Detection" (physics.app-ph, quant-ph)
- **PDF**: https://arxiv.org/pdf/2606.00818
- **DOI**: https://doi.org/10.48550/arXiv.2606.00818