Face acquisition camera design using the NV-IPM image generation tool

Howell, Christopher L.; Choi, Hee-Sue; Reynolds, Joseph P.

doi:10.1117/12.2179619

Cited by 4 publications

(3 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…LeMaster and Eismann [40] demonstrated how the face detection problem could be used to impact sensor design by transforming imagery with pyBSM and evaluating the performance of a Haar feature-based cascade classifier. Similarly, Howell et al [43] used the NV-IPM image generation tool for designing a camera to optimize face recognition with the PITT-PATT algorithm.…”

Section: E Cv-based Modelingmentioning

confidence: 99%

Remote Sensor Design for Visual Recognition With Convolutional Neural Networks

Jaffe

Zelinski

Sakla

2019

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

While deep learning technologies for computer vision have developed rapidly since 2012, modeling of remote sensing systems has remained focused around human vision. In particular, remote sensing systems are usually constructed to optimize sensing cost-quality tradeoffs with respect to human image interpretability. While some recent studies have explored remote sensing system design as a function of simple computer vision algorithm performance, there has been little work relating this design to the state of the art in computer vision: deep learning with convolutional neural networks. We develop experimental systems to conduct this analysis, showing results with modern deep learning algorithms and recent overhead image data. Our results are compared to standard image quality measurements based on human visual perception, and we conclude not only that machine and human interpretability differ significantly but also that computer vision performance is largely self-consistent across a range of disparate conditions. This paper is presented as a cornerstone for a new generation of sensor design systems that focus on computer algorithm performance instead of human visual perception.

show abstract

Section: E Cv-based Modelingmentioning

confidence: 99%

Remote Sensor Design for Visual Recognition With Convolutional Neural Networks

Jaffe

Zelinski

Sakla

2019

IEEE Trans. Geosci. Remote Sensing

View full text Add to dashboard Cite

show abstract

“…In 2015, NVESD researchers used NV-IPM's image generation feature to test the performance of a facial recognition algorithm [6] (PITTPATT) against various ranges, levels of blur, and levels of illumination. No effort was made to compare the results to human performance, and the authors caution against attempting to use their results for that purpose.…”

Section: Non-tod Approachesmentioning

confidence: 99%

Automated image-based range performance measurement using TOD

Short¹

2023

Infrared Imaging Systems: Design, Analysis, Modeling, and Testing XXXIV

View full text Add to dashboard Cite

Image processing (including histogram equalization, local area processing, and edge sharpening) is a key component of practical electro-optical imaging systems. Despite this, the range performance impact of such processing remains difficult to quantify, short of running a full human perception experiment. The primary difficulty is that current analytic range performance models—best exemplified by the Targeting Task Performance (TTP) model—can only account for linear and shift-invariant (LSI) image effects. We present our efforts towards developing a quantitative, image-based range performance metric that does not require LSI assumptions. Our proposed metric is based on a Triangle Orientation Discrimination (TOD) target set and observer task, with automatic scoring accomplished through a simple template correlator. The approach is compatible with both synthetic and real imagery.

show abstract

“…In [40], LeMaster et al demonstrate how the face detection problem could be used to impact sensor design by transforming imagery with pyBSM and evaluating performance of a Haar feature-based cascade classifier. Similarly, Howell et al use the NV-IPM image generation tool for designing a camera to optimize for face recognition with the PITT-PATT algorithm [43].…”

Section: E Cv-based Modelingmentioning

confidence: 99%

Remote Sensor Design for Visual Recognition with Convolutional Neural Networks

Jaffe,

Zelinski,

Sakla

2019

Preprint

View full text Add to dashboard Cite

While deep learning technologies for computer vision have developed rapidly since 2012, modeling of remote sensing systems has remained focused around human vision. In particular, remote sensing systems are usually constructed to optimize sensing cost-quality trade-offs with respect to human image interpretability. While some recent studies have explored remote sensing system design as a function of simple computer vision algorithm performance, there has been little work relating this design to the state-of-the-art in computer vision: deep learning with convolutional neural networks. We develop experimental systems to conduct this analysis, showing results with modern deep learning algorithms and recent overhead image data. Our results are compared to standard image quality measurements based on human visual perception, and we conclude not only that machine and human interpretability differ significantly, but that computer vision performance is largely self-consistent across a range of disparate conditions. This research is presented as a cornerstone for a new generation of sensor design systems which focus on computer algorithm performance instead of human visual perception.

show abstract

Face acquisition camera design using the NV-IPM image generation tool

Cited by 4 publications

References 3 publications

Remote Sensor Design for Visual Recognition With Convolutional Neural Networks

Remote Sensor Design for Visual Recognition With Convolutional Neural Networks

Automated image-based range performance measurement using TOD

Remote Sensor Design for Visual Recognition with Convolutional Neural Networks

Contact Info

Product

Resources

About