HDR luminance normalization via contextual facilitation in highly recurrent spiking neural networks

White, Alexander James; Hung, Chou P.; Lo, Chung-Chuan

doi:10.1117/12.2623789

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Grouping mechanisms that employ Gestalt laws of perceptual organization, such as, good continuation, proximity, and similarity, have been defined in other biologically plausible models (Grossberg and Mingolla 1985;Grossberg et al 1997;Li 1998) and have already inspired computer vision algorithms for contour enhancement (Parent and Zucker 1989;Guy and Medioni 1996;Hung et al 2020;White et al 2022). The architecture outlined here emphasizes that the proposed canonical principles of computation i.e., feedforward and feedback processing combined with competition and normalization, can achieve contextual amplification through grouping, as well as de-amplification or suppression in case of reduced information content of stimulus components.…”

Section: Perceptual Binding In Contour Grouping and Texture Suppressionmentioning

confidence: 99%

Canonical circuit computations for computer vision

2023

View full text Add to dashboard Cite

Advanced computer vision mechanisms have been inspired by neuroscientific findings. However, with the focus on improving benchmark achievements, technical solutions have been shaped by application and engineering constraints. This includes the training of neural networks which led to the development of feature detectors optimally suited to the application domain. However, the limitations of such approaches motivate the need to identify computational principles, or motifs, in biological vision that can enable further foundational advances in machine vision. We propose to utilize structural and functional principles of neural systems that have been largely overlooked. They potentially provide new inspirations for computer vision mechanisms and models. Recurrent feedforward, lateral, and feedback interactions characterize general principles underlying processing in mammals. We derive a formal specification of core computational motifs that utilize these principles. These are combined to define model mechanisms for visual shape and motion processing. We demonstrate how such a framework can be adopted to run on neuromorphic brain-inspired hardware platforms and can be extended to automatically adapt to environment statistics. We argue that the identified principles and their formalization inspires sophisticated computational mechanisms with improved explanatory scope. These and other elaborated, biologically inspired models can be employed to design computer vision solutions for different tasks and they can be used to advance neural network architectures of learning.

show abstract