Color image fusion for concealed weapon detection

Abstract

Recent advances in passive and active imaging sensor technology offer the potential to detect weapons that are concealed underneath a person's clothing or carried along in bags. Although the concealed weapons can sometimes easily be detected, it can be difficult to perceive their context, due to the non-literal nature of these images. Especially for dynamic crowd surveillance purposes it may be impossible to rapidly asses with certainty which individual in the crowd is the one carrying the observed weapon. Sensor fusion is an enabling technology that may be used to solve this problem. Through fusion the signal of the sensor that depicts the weapon can be displayed in the context provided by a sensor of a different modality. We propose an image fusion scheme in which non-literal imagery can be fused with standard color images such that the result clearly displays the observed weapons in the context of the original color image. The procedure is such that the relevant contrast details from the non-literal image are transferred to the color image without altering the original color distribution of this image. The result is a natural looking color image that fluently combines all details from both input sources. When an observer who performs a dynamic crowd surveillance task, detects a weapon in the scene, he will also be able to quickly determine which person in the crowd is actually carrying the observed weapon (e.g. "the man with the red T-shirt and blue jeans"). The method is illustrated by the fusion of thermal 8-12 µm imagery with standard RGB color images.

Fusion of colour and monochromatic images with edge emphasis

“…Cilj je da se sjedinjavanjem ove dve slike prenesu objekti vidljivi u IC slici, a da se zadrži visoka rezolucija i prirodnost TV slike. Metode za sjedinjavanje kolor i monohromatskih slika zasnovane su na transformaciji kolor slike iz RGB kolor sistema u neki drugi kolor sistem (HSV, LAB, lαβ, ...) i zatim sjedinjavanju monohromatske slike sa jednim od kanala kolor slike (Toet, 2003c), (Zhiyun, Rick, 2003), (Huang, et al, 2008). Metoda za sjedinjavanje jednog od kolor kanala sa monohromatskom slikom jedna je od standardnih metoda za sjedinjavanje monohromatskih slika (Burt, Adelson, 1983), (Petrović, Xydeas, 1999), (Petrović, Xydeas, 2004), (Toet, 1989).…”

“…Uspešnost sjedinjavanja zavisi od izbora kolor sistema za sjedinjavanje i metode monohromatskog sjedinjavanja. Veći broj metoda koristi sjedinjavanje monohromatske slike sa kanalom koji predstavlja intenzitet piksela na kolor slici (Toet, 2003c), (Huang, et al, 2008). Mana ovih metoda može da bude gubitak kontrasta na sjedinjenoj slici i narušavanje prirodnog izgleda kolor slike.…”

“…Slika 4 -Rezultat sjedinjavanja kolor i monohromatske slike prikazan na slici 1, HSV kolor transformacijom i Laplasovom piramidom Figure 4 -Colour fusion of the inputs in Figure 1 using the HSV space and the Laplacian pyramid fusion Jedan deo metoda za sjedinjavanje kolor i monohromatskih slika našao je primenu u detekciji skrivenog oružja (DSO) (Toet, 2003c), (Zhiyun, Rick, 2003). Metoda koja je opisana u (Toet, 2003c) zasnovana je na lαβ kolor transformaciji. Kolor slika se transformiše u lαβ kolor sistem i zatim se vrši sjedinjavanje l-komponente sa monohromatskom slikom.…”

Peer Review #2 of "Iterative guided image fusion (v0.1)"

“…Some potential benefits of image fusion are: wider spatial and temporal coverage, decreased uncertainty, improved reliability, and increased system robustness. Image fusion has important applications in defense and security for situational awareness (Toet et al 1997), surveillance (Shah et al 2013;Zhu & Huang 2007), target tracking (Motamed et al 2005;Zou & Bhanu 2005), intelligence gathering (O'Brien & Irvine 2004), concealed weapon detection (Bhatnagar & Wu 2011;Liu et al 2006;Toet 2003;Xue & Blum 2003;Xue et al 2002;Yajie & Mowu 2009), detection of abandoned packages (Beyan et al 2011) and buried explosives (Lepley & Averill 2011), and face recognition (Kong et al 2007;Singh et al 2008). Other important image fusion applications are found in industry (Tian et al 2009), art analysis (Zitová et al 2011), agriculture (Bulanona et al 2009), remote sensing (Ghassemian 2001;Jacobson & Gupta 2005;Jacobson et al 2007;Jiang et al 2011) and medicine (Agarwal & Bedi 2015;Biswas et al 2015;Daneshvar & Ghassemian 2010;Singh & Khare 2014;Wang et al 2014;Yang & Liu 2013) (for a survey of different applications of image fusion techniques see ).…”

Multiple reference images based on lookup-table color image fusion algorithm

“…In 2001, professor Erik Reinhard from University of Bristol first proposed color transfer in lαβ space, then Welsh [1], Vieira [2], Gary R. Greenfield [3], Yu-Wing Tai [4] from HK, Yufeng Zheng and Edward A. Essock [5] made contributions in image fusion improving. In 2003, Alexander Toet [6][7][8][9] from TNO laboratory in Holland innovatively took color transfer into color night vision research field. This years, Lingxue Wang and Weiqi Jin from Beijing Institute of Technology(BIT) proposed color transfer based image fusion algorithm in YUV space [10,11], and realized low light/IR double-band natural feeling color night vision real-time hardware processing and applying in equipment.…”