Neuromorphic Circuits on FDSOI Technology for Computer Vision Applications
| Autor: | Yogesh Singh Chauhan, Amit Ranjan Trivedi, Dinesh Rajasekharan |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Pixel
business.industry Computer science Boolean circuit Transistor ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Hardware_PERFORMANCEANDRELIABILITY law.invention symbols.namesake CMOS law Logic gate Hardware_INTEGRATEDCIRCUITS symbols Inverter Computer vision Artificial intelligence business Hardware_LOGICDESIGN Von Neumann architecture Electronic circuit |
| Zdroj: | VLSI Design |
| DOI: | 10.1109/vlsid.2019.00108 |
| Popis: | Potential of neuromorphic circuits on FDSOI technology for computer vision applications is demonstrated in this paper. Computer vision systems based on conventional Von Neumann architecture consume large area and energy. The FDSOI inverter-based circuits proposed in this work, require only 11 transistors per pixel for colour detection, and only 59 transistors per pixel for erosion and dilation operations, whereas the CMOS-based Boolean circuit requires more than 300 transistors per pixel, and 2700 transistors per pixel, respectively, for these operations. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|