A Visual Feature Detection Algorithm Inspired by Spatio-Temporal Properties of Visual Neurons

Enhancing or detecting visual features related both to shape and motion is an important step in motion recognition. In the present study, we developed a spatio-temporal visual feature detection algorithm that provides information for estimating both orientation and velocity based on a physiological...

Full description

Saved in:
Bibliographic Details
Published inNeural Information Processing pp. 634 - 643
Main Authors Horiguchi, Eisaku, Okuno, Hirotsugu
Format Book Chapter
LanguageEnglish
Published Cham Springer International Publishing 2021
SeriesLecture Notes in Computer Science
Subjects
Image processing
Motion
Neuromorphic
Robot vision
Online AccessGet full text
ISBN3030922723
9783030922726
ISSN0302-9743
1611-3349
DOI10.1007/978-3-030-92273-3_52

Cover

More Information
Summary:Enhancing or detecting visual features related both to shape and motion is an important step in motion recognition. In the present study, we developed a spatio-temporal visual feature detection algorithm that provides information for estimating both orientation and velocity based on a physiological model of motion sensitive neurons. The algorithm consists of a low number of simple processing steps, and therefore, is suitable for circuit implementation and for edge computing. We evaluated the algorithm by using computer simulated movies and real-world movies. The results showed that the algorithm has high selectivity for motion direction and little dependence on contrast, and provides enough information for speed estimation.
ISBN:3030922723
9783030922726
ISSN:0302-9743
1611-3349
DOI:10.1007/978-3-030-92273-3_52