Ground-Based Hyperspectral Image Surveillance Systems for Explosive Detection: Part II-Radiance to Reflectance Conversions


Koz A.

IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, vol.12, no.12, pp.4754-4765, 2019 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 12 Issue: 12
  • Publication Date: 2019
  • Doi Number: 10.1109/jstars.2020.2964483
  • Journal Name: IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.4754-4765
  • Keywords: Reflectivity, Hyperspectral imaging, Surveillance, Object detection, Explosives, Explosive detection, hyperspectral image surveillance (HSI), incident light, radiance to reflectance conversion, Spectralon, teflon, white reference, CALIBRATION
  • Middle East Technical University Affiliated: Yes

Abstract

The usage of white references for reflectance conversion in a captured scene for potential ground-based hyperspectral image surveillance systems suffers from security, as it exposes the location of the system, and could be expensive for a broad application due to the high prices of standard Spectralons. In Part I of this two-part article, the state-of-the-art of ground-based hyperspectral imaging systems for explosive trace detection was discussed along with the future challenges for generic surveillance systems. The part II of the article proposes different approaches for reflection conversion to compensate the usage of white references at every capturing instant during surveillance, which is one of the main challenges of these systems. As an initial approach, recording the radiance of white reference only at the initialization of the system is tested by capturing an experimental setup in different days and times throughout a duration of two months. The correlations between the incident lights obtained from white references at different times, days, and orientations are compared. The effects of using a prerecorded white radiance for the reflectance conversion in the current scene are investigated. Second, a practical approach for reflectance calculation is tested by employing cheap-price reference objects having smooth characteristics in short-wave infrared spectrum, instead of expensive Spectralons. Finally, a novel reflectance conversion method is developed by using the reflectance of the static background materials in the scene rather than standard references by considering the stationary characteristics of surveillance systems. The proposed approaches are compared and verified on the baseline detection methods.