|
|
Multidimensional illumination and image processing techniques in the W-band for recognition of concealed objects Authors: J. Stiens, L. Zhang, A. Elhawil, I. Jäger, G. Koers, H. Sahli and P. Schelkens Publication Year: 2007
Abstract: In recent years the research field of mm waves, the spectral band between 30 and 300 GHz has expanded in security and biomedical applications, because mm-waves yield a convenient compromise between acceptable resolution and sufficient penetration depth through a variety of materials, non-transparent in the visible. E.g. mm waves penetrate clothing to image concealed weapons, but also allow skin cancer detection and skin wound healing assessment. In indoor applications (the biggest market in any sense) the temperature contrast between the body, objects under investigation and the background of the scene (walls of a room or hall) is too low for making high contrast imaging in a reasonable exposure time. Active imaging techniques can outperform passive systems in contrast and resolution when the typical image artifacts inherent to active systems are understood and can be eliminated.
In this paper we focus on security applications in the W-band (75뀶 GHz) Illuminating the object in the mm-band is not obvious due to the lack of the equivalent of our well known flash lamp. On the contrary one needs to cope with coherent illumination sources such as Gunn or Impatt diodes or BWOs, resulting in a series of typical coherent specific image artifacts such as speckle, ringing, glint and interferometric noise, ... In this paper we will show how the disadvantages of coherent illumination can be converted in more favorable conditions for image improvement and material or pattern recognition. The parameters of the multidimensional illumination set that will be discussed are: multi-angle, multi-frequency, multi-polarization and multi-phase patterns. Afterwards the various multi-dimensional arrays of mm wave images will be analyzed and correlated with each other to distinguish between different materials (metals, dielectrics, depolarizing materials). We will apply and adapt various common image processing techniques such as PCA, intensity histogram analysis, despeckling algorithms, etc. Different examples specific to security applications will be discussed in detail.
|
|
