Skip to Main content Skip to Navigation
Conference papers

Bio-inspired quantization

Abstract : This paper introduces a novel retina-inspired sparse representation which is applied to temporally constant 2D inputs. This architecture consists of the recently released retina-inspired filtering which mimics the transformation of the visual stimulus into current as it takes place in the retina. This transform is very redundant. As a result, we propose the Perfect Leaky Integrate and Fire (Perfect-LIF) as a model which sparsifies the over-complete retina-inspired decomposition mimicking the spike generation mechanisms of the neurons. The Perfect-LIF is a thresholding function based on a time-depended deadzone. Numerical results show the efficiency of our architecture which provides almost equivalent reconstruction results between the over-complete and the sparse representation of the input image.
Document type :
Conference papers
Complete list of metadata

Cited literature [10 references]  Display  Hide  Download
Contributor : Lionel Fillatre <>
Submitted on : Tuesday, June 4, 2019 - 4:12:34 PM
Last modification on : Tuesday, March 30, 2021 - 9:24:31 AM


Files produced by the author(s)


  • HAL Id : hal-01562644, version 1



Effrosyni Doutsi, Lionel Fillatre, Marc Antonini, Julien Gaulmin. Bio-inspired quantization. XXVIeme Colloque GRETSI, Sep 2017, Juan-les-Pins, France. ⟨hal-01562644⟩



Record views


Files downloads