A Neuroinspired Image Sensor System With A Hierarchical Multiscale Gabor Filtering Circuit

• doi:
• title: A neuro-inspired image sensor system with a hierarchical multi-scale Gabor filtering circuit
• publisher: IEEE
• isbn: 979-8-3315-4446-1
• issn:
• rank: 1183
• access_type: LOCKED
• content_type: Conferences
• abstract: A set of Gabor filters with selectivity for various spatial frequencies, phases, and orientations is one of the most important spatial features, not only for developing image-based applications such as image recognition, but also for simulating the visual nervous system. In this study, we designed an efficient neuro-inspired multi-scale Gabor filtering circuit that provides Gabor filtered images with four orientations and two orthogonal phases for each scale. To reduce the computational cost, we employed the following techniques: (i) neuro-inspired hierarchical filtering that starts with a simple low-pass filter and progressively achieves Gabor-like characteristics, (ii) separating one two-dimensional filter into two one-dimensional filters by using the separability of Gaussian and Gabor functions, (iii) multi-scale filter kernels that differ in width but not in the weight values. We implemented the filter circuit into a field-programmable gate array (FPGA), and developed an image sensor system that consists of the FPGA and a CMOS image sensor. The system output a set of Gabor filtered images with $160\times 120$ pixels at approximately 110 frames per second. The set of filtered images was used to visualize responses of a model of visual neurons that respond selectively to particular spatial frequencies and orientations, regardless of spatial phase, and the model's response was reproduced appropriately.
• article_number: 10805153
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10805153
• html_url: https://ieeexplore.ieee.org/document/10805153/
• abstract_url: https://ieeexplore.ieee.org/document/10805153/
• publication_title: 2024 SICE Festival with Annual Conference (SICE FES)
• conference_location: Kochi City, Japan
• conference_dates: 27-30 Aug. 2024
• publication_number: 10804877
• is_number: 10804885
• publication_year: 2024
• publication_date: 27-30 Aug. 2024
• start_page: 1194
• end_page: 1201
• citing_paper_count: 0
• citing_patent_count: 0
• download_count: 22
• insert_date: 20241224
• index_terms:
  • ieee_terms:
    • Semiconductor device modeling
    • Visualization
    • Filtering
    • Low-pass filters
    • CMOS image sensors
    • Universal Serial Bus
    • Gabor filters
    • Integrated circuit modeling
    • Field programmable gate arrays
    • Testing
  • author_terms:
    • multi-scale
    • Gabor
    • FPGA
    • neuromorphic
    • image sensor
  • dynamic_index_terms:
    • Image Sensor
    • Camera Sensor
    • Gabor Filters
    • Filter Circuit
    • Gaussian Kernel
    • Gaussian Function
    • Spatial Orientation
    • Spatial Direction
    • Filter Kernel
    • Spatial Phase
    • Multiscale Filter
    • Convolution
    • Visual Cortex
    • Visual Cortices
    • Gaussian Filter
    • Preferred Orientation
    • Preferential Orientation
    • Odd Number
    • Uneven Number
    • Even Number
    • Even Integer
    • Number Of Weights
    • Energy Model
    • Spatial Filter
    • Filtering Algorithm
    • Number Of Scales
    • Conventional Circuit
    • Preferred Frequency
    • Frequency Preference
    • Bit-width
    • Sophisticated Functions
    • Sophisticated Mapping
    • Digital Circuits
    • Digital Circuitry
    • Simple Cells
    • USB Interface
    • Filter Scale
    • Shift Register
    • Kernel Width
• isbn_formats:
  • format: Print on Demand(PoD) ISBN, value: 979-8-3315-4446-1, isbnType: New-2005
  • format: Electronic ISBN, value: 978-4-9077-6483-8, isbnType: New-2005
• authors:
  • Author Name: Shutaro Kubo
    Affiliation: Graduate School of Information Science and Technology, Osaka Institute of Technology, Osaka, Japan
    Author URL: https://ieeexplore.ieee.org/author/228070395611188
    ID: 228070395611188
    Order: 1
    Author Affiliations:
    • Graduate School of Information Science and Technology, Osaka Institute of Technology, Osaka, Japan
  • Author Name: Yuki Yamaji
    Affiliation: Graduate School of Information Science and Technology, Osaka Institute of Technology, Osaka, Japan
    Author URL: https://ieeexplore.ieee.org/author/295248964903356
    ID: 295248964903356
    Order: 2
    Author Affiliations:
    • Graduate School of Information Science and Technology, Osaka Institute of Technology, Osaka, Japan
  • Author Name: Hirotsugu Okuno
    Affiliation: Faculty of Information Science and Technology, Osaka Institute of Technology, Osaka, Japan
    Author URL: https://ieeexplore.ieee.org/author/37288777100
    ID: 37288777100
    Order: 3
    Author Affiliations:
    • Faculty of Information Science and Technology, Osaka Institute of Technology, Osaka, Japan

Image Sensor

• sensor_type: CMOS
• resolution: 160x120
• dynamic_range: Not specified
• pixel_size: Not specified
• dark_current: Not specified

Optical Data

• focal_length: Not specified
• aperture: Not specified
• field_of_view: Not specified
• distortion: Not specified

Performance Metrics

• frame_rate: 110 fps for a single scale, 83 fps for two scales

Applications & Benefits

• cell_imaging: The system produces eight filtered images for each scale, i.e., four-orientation and two-phase Gabor filtered images.
• benefits: Real-time sensor testing for various visual nervous system models.

Supporting Organizations

• supported_by: JSPS KAKENHI Grant Number JP24H02338

Manuscript Details

• publication_date: 27-30 Aug. 2024

Relevancy Score

• score: 8
• missing_fields:
  • dynamic_range
  • pixel_size
  • dark_current
  • focal_length
  • aperture
  • field_of_view
  • distortion
  • signal_to_noise_ratio
  • sensitivity
  • shutter_speed
  • power_consumption
  • noise

Processed JSON Filename

request_12eb6215-42ae-4c60-a686-05bb5b493f39-a_neuroinspired_image_sensor_system_with_a_hierarchical_multiscale_gabor_filtering_circuit.json