A Hardware Architecture For Cellbased Featureextraction And Classification Using Dualfeature Space

• doi: 10.1109/TCSVT.2017.2726564
• title: A Hardware Architecture for Cell-Based Feature-Extraction and Classification Using Dual-Feature Space
• publisher: IEEE
• isbn:
• issn: 1558-2205
• rank: 736
• access_type: LOCKED
• content_type: Journals
• abstract: Many computer-vision and machine-learning applications in robotics, mobile, wearable devices, and automotive domains are constrained by their real-time performance requirements. This paper reports a dual-feature-based object recognition coprocessor that exploits both histogram of oriented gradient (HOG) and Haar-like descriptors with a cell-based parallel sliding-window recognition mechanism. The feature extraction circuitry for HOG and Haar-like descriptors is implemented by a pixel-based pipelined architecture, which synchronizes to the pixel frequency from the image sensor. After extracting each cell feature vector, a cell-based sliding window scheme enables parallelized recognition for all windows, which contain this cell. The nearest neighbor search classifier is, respectively, applied to the HOG and Haar-like feature space. The complementary aspects of the two feature domains enable a hardware-friendly implementation of the binary classification for pedestrian detection with improved accuracy. A proof-of-concept prototype chip fabricated in a 65-nm SOI CMOS, having thin gate oxide and buried oxide layers (SOTB CMOS), with 3.22-mm2 core area achieves an energy efficiency of 1.52 nJ/pixel and a processing speed of 30 fps for 1024 × 1616-pixel image frames at 200-MHz recognition working frequency and 1-V supply voltage. Furthermore, multiple chips can implement image scaling, since the designed chip has image-size flexibility attributable to the pixel-based architecture.
• article_number: 7979565
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7979565
• html_url: https://ieeexplore.ieee.org/document/7979565/
• abstract_url: https://ieeexplore.ieee.org/document/7979565/
• publication_title: IEEE Transactions on Circuits and Systems for Video Technology
• conference_location:
• conference_dates:
• publication_number: 76
• is_number: 8506638
• publication_year: 2018
• publication_date: Oct. 2018
• start_page: 3086
• end_page: 3098
• citing_paper_count: 9
• citing_patent_count: 0
• download_count: 522
• insert_date: 20170713
• index_terms:
  • ieee_terms:
    • Feature extraction
    • Computer architecture
    • Histograms
    • Training
    • Microprocessors
    • Prototypes
  • author_terms:
    • Dual feature
    • HOG
    • Haar-like
    • pixel-based pipeline
    • cell-based recognition
    • complementary classifier
  • dynamic_index_terms:
    • Hardware Architecture
    • Energy Efficiency
    • Feature Space
    • Parallelization
    • Parallel Computing
    • Image Sensor
    • Camera Sensor
    • Image Frames
    • Nearest Neighbor Search
    • Nearest Neighbor Method
    • Nearest Neighbour Method
    • Working Frequency
    • Histogram Of Oriented Gradients
    • Pedestrian Detection
    • Coprocessor
    • Multiple Chips
    • Support Vector Machine
    • Positive Samples
    • Horizontal Plane
    • Horizontal Direction
    • Horizontal Distance
    • Window Size
    • Input Image
    • Negative Samples
    • Image Object
    • Intermediate Results
    • Intermediate Outcomes
    • Dual Character
    • Dual Feature
    • Integral Image
    • Positive Training
    • Hardware Implementation
    • Scale-invariant Feature Transform
    • Scale Invariant Feature Transform
    • Clock Cycles
    • Negative Images
    • Positive Image
    • Human Figure
    • Regeneration Buffer
    • Framebuffer
    • Frame Buffer
    • Display Memory
• authors:
  • Author Name: Fengwei An
    Affiliation: Institute of Engineering, Hiroshima University, Higashi-Hiroshima, Japan
    Author URL: https://ieeexplore.ieee.org/author/38235502500
    ID: 38235502500
    Order: 1
    Author Affiliations:
    • Institute of Engineering, Hiroshima University, Higashi-Hiroshima, Japan
  • Author Name: Xiangyu Zhang
    Affiliation: Graduate School of Engineering, Hiroshima University, Higashi-Hiroshima, Japan
    Author URL: https://ieeexplore.ieee.org/author/37085836690
    ID: 37085836690
    Order: 2
    Author Affiliations:
    • Graduate School of Engineering, Hiroshima University, Higashi-Hiroshima, Japan
  • Author Name: Aiwen Luo
    Affiliation: Graduate School of Advanced Sciences of Matter, Hiroshima University, Higashi-Hiroshima, Japan
    Author URL: https://ieeexplore.ieee.org/author/37086102073
    ID: 37086102073
    Order: 3
    Author Affiliations:
    • Graduate School of Advanced Sciences of Matter, Hiroshima University, Higashi-Hiroshima, Japan
  • Author Name: Lei Chen
    Affiliation: HiSIM Research Center, Hiroshima University, Higashi-Hiroshima, Japan
    Author URL: https://ieeexplore.ieee.org/author/37085589660
    ID: 37085589660
    Order: 4
    Author Affiliations:
    • HiSIM Research Center, Hiroshima University, Higashi-Hiroshima, Japan
  • Author Name: Hans Jürgen Mattausch
    Affiliation: Research Institute for Nanodevice and Bio Systems, Hiroshima University, Higashi-Hiroshima, Japan
    Author URL: https://ieeexplore.ieee.org/author/37269154600
    ID: 37269154600
    Order: 5
    Author Affiliations:
    • Research Institute for Nanodevice and Bio Systems, Hiroshima University, Higashi-Hiroshima, Japan

Image Sensor

• sensor_type: CMOS
• resolution: 1024x1616
• 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: 30 fps
• power_consumption: 75.48 mW

Applications & Benefits

• cell_imaging: Used for pedestrian detection and object recognition in robotics, automotive systems, and security systems.
• benefits: High energy efficiency (1.52 nJ/pixel), real-time processing capabilities, flexibility in image resolutions.

Supporting Organizations

• supported_by: Hiroshima University TAOYAKA Program, Ministry of Education, Culture, Sports, Science and Technology (Japan)

Manuscript Details

• publication_date: Oct. 2018

Relevancy Score

• score: 9
• missing_fields:
  • fill factor
  • quantum efficiency
  • readout speed
  • temporal noise
  • fixed-pattern noise
  • analog-to-digital conversion techniques
  • use of microlenses
  • on-chip colour filters
  • global or rolling shutters
  • backside illumination

Processed JSON Filename

request_0b7ea73e-3e7d-48da-a175-93e1774eaf62-a_hardware_architecture_for_cellbased_featureextraction_and_classification_using_dualfeature_space.json