Related Experiment Video
Updated: Jul 17, 2026
09:59
Fabrication of Flexible Image Sensor Based on Lateral NIPIN Phototransistors
Published on: June 23, 2018
A new silicon retina model and its advantages
Kuntal Ghosh1, Sandip Sarkar, Kamales Bhaumik
1Microelectronics Division, Saha Institute of Nuclear Physics, 1/AF Bidhannagar, Kolkata-700064, India phone: 091-033-23375346; fax: 091-033-23374637;
Summary
A novel silicon retina model integrates retinal layers for advanced visual processing. This neuromorphic system aids in computing zero-crossing maps, mimicking the human visual system for image retrieval.
Area of Science:
- Neuromorphic Engineering
- Computational Neuroscience
- Computer Vision
Background:
- Existing neuromorphic models often simplify retinal processing.
- Understanding the human visual system (HVS) is key for advanced AI.
- Retinal ganglion cell receptive fields are fundamental to visual information processing.
Purpose of the Study:
- To propose a new silicon retina model.
- To incorporate contributions from both inner and outer plexiform layers of the retina.
- To explore neuromorphic implications for visual processing and image retrieval.
Main Methods:
- Developed a silicon retina model based on retinal ganglion cell receptive fields.
- Implemented a linear combination of inner and outer plexiform layer contributions.
- Investigated the model's capability in computing zero-crossing maps using higher-order derivatives.
Main Results:
- The proposed model directly integrates retinal layer contributions.
- The system demonstrates capability in computing zero-crossing maps.
- The model shows potential for implementing derivative analyzers mimicking the HVS.
Conclusions:
- The new silicon retina model offers a more biologically plausible approach.
- This model has significant neuromorphic implications for visual processing.
- The system provides advantages for image retrieval applications.
