Image Reconstruction For Quanta Image Sensors Using Deep Neural Networks

• doi: 10.1109/ICASSP.2018.8461685
• title: Image Reconstruction for Quanta Image Sensors Using Deep Neural Networks
• publisher: IEEE
• isbn: 978-1-5386-4659-5
• issn: 2379-190X
• rank: 2694
• access_type: LOCKED
• content_type: Conferences
• abstract: Quanta Image Sensor (QIS) is a single-photon image sensor that oversamples the light field to generate binary measurements. Its single-photon sensitivity makes it an ideal candidate for the next generation image sensor after CMOS. However, image reconstruction of the sensor remains a challenging issue. Existing image reconstruction algorithms are largely based on optimization. In this paper, we present the first deep neural network approach for QIS image reconstruction. Our deep neural network takes the binary bit stream of QIS as input, learns the nonlinear transformation and denoising simultaneously. Experimental results show that the proposed network produces significantly better reconstruction results compared to existing methods.
• article_number: 8461685
• pdf_url: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8461685
• html_url: https://ieeexplore.ieee.org/document/8461685/
• abstract_url: https://ieeexplore.ieee.org/document/8461685/
• publication_title: 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
• conference_location: Calgary, AB, Canada
• conference_dates: 15-20 April 2018
• publication_number: 8450881
• is_number: 8461260
• publication_year: 2018
• publication_date: 15-20 April 2018
• start_page: 6543
• end_page: 6547
• citing_paper_count: 18
• citing_patent_count: 0
• download_count: 772
• insert_date: 20180913
• index_terms:
  • ieee_terms:
    • Image reconstruction
    • Neural networks
    • Transforms
    • Pipelines
    • Image sensors
    • Maximum likelihood estimation
    • Random variables
  • author_terms:
    • Quanta Image Sensor
    • single-photon imaging
    • image reconstruction
    • deep neural networks
  • dynamic_index_terms:
    • Neural Network
    • Deep Neural Network
    • Image Reconstruction
    • Iterative Reconstruction
    • Tomographic Reconstruction
    • Image Sensor
    • Camera Sensor
    • Quanta Image Sensor
    • Denoising
    • Noise Reduction
    • Bitstream
    • Binary Sequence
    • Binary Measure
    • Reconstruction Approach
    • Image Reconstruction Algorithm
    • Random Variables
    • Low-pass
    • Low-pass Filter
    • Gradient Descent
    • Steepest Descent
    • Convolutional Layers
    • Nonlinear Function
    • Nonlinear Mapping
    • Poisson Distribution
    • Poissonian
    • Reconstruction Quality
    • Maximum A Posteriori
    • Maximum-a-posteriori
    • Deconvolutional Layers
    • Model Mismatch
    • Variance Stabilizing Transformation
    • Variance-stabilizing Transformation
    • varianceStabilizingTransformation
• isbn_formats:
  • format: Print on Demand(PoD) ISBN, value: 978-1-5386-4659-5, isbnType: New-2005
  • format: USB ISBN, value: 978-1-5386-4657-1, isbnType: New-2005
  • format: Electronic ISBN, value: 978-1-5386-4658-8, isbnType: New-2005
• authors:
  • Author Name: Joon Hee Choi
    Affiliation: School of ECE and Dept of Statistics, Purdue University, West Lafayette, IN
    Author URL: https://ieeexplore.ieee.org/author/37086189005
    ID: 37086189005
    Order: 1
    Author Affiliations:
    • School of ECE and Dept of Statistics, Purdue University, West Lafayette, IN
  • Author Name: Omar A. Elgendy
    Affiliation: School of ECE and Dept of Statistics, Purdue University, West Lafayette, IN
    Author URL: https://ieeexplore.ieee.org/author/37086150685
    ID: 37086150685
    Order: 2
    Author Affiliations:
    • School of ECE and Dept of Statistics, Purdue University, West Lafayette, IN
  • Author Name: Stanley H. Chan
    Affiliation: School of ECE and Dept of Statistics, Purdue University, West Lafayette, IN
    Author URL: https://ieeexplore.ieee.org/author/37600910000
    ID: 37600910000
    Order: 3
    Author Affiliations:
    • School of ECE and Dept of Statistics, Purdue University, West Lafayette, IN

Image Sensor

• sensor_type: Quanta Image Sensor (QIS)
• resolution: Not specified
• dynamic_range: Not specified
• pixel_size: Not specified
• dark_current: 1-250 photoelectrons

Optical Data

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

Performance Metrics

• frame_rate: 10k fps (up to 156k fps)
• sensitivity: Single-photon sensitivity

Applications & Benefits

• cell_imaging: Not specified
• benefits: High spatial resolution and single-photon sensitivity suitable for various applications.

Supporting Organizations

• supported_by: U.S. National Science Foundation

Manuscript Details

• publication_date: 15-20 April 2018

Relevancy Score

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

Processed JSON Filename

request_3a8174b5-3015-47e5-9acf-f73de97957ed-image_reconstruction_for_quanta_image_sensors_using_deep_neural_networks.json