Several targets are set-up outside and imaged by a passive millimeter-wave sensor over a 24 hour period. The sensor is capable of measuring two linear polarization states simultaneously and the contrasts of the targets are compared for the different polarizations. The choice of polarization is shown to have an impact on the contrast of different targets throughout the day. In an extreme case the contrast of a target experiences a crossover event and disappears for one polarization while it presents a strong contrast (9 K) with the other polarization. Experimental results are shown along with a simulation of the scene using a ray tracing program.
©2013 Optical Society of America
Passive millimeter-wave (mmW) sensors operate in a unique part of the electromagnetic spectrum that is able to penetrate through a variety of common obscurants [1–3] while still producing resolutions that are adequate for human operators to identify targets of interest . Their passive nature can be important for military applications and they are especially suited to persistent surveillance applications due to their ability to operate independently of day/night conditions and through transient atmospheric obscurants such as clouds, rain and fog [5,6].
An important phenomenon for persistent surveillance systems based on infrared (IR) sensors is the crossover period of two objects. As the diurnal heating cycle heats objects at different rates based on their thermal inertia, the radiometric contrast between objects can become small and then eventually invert as one object warms/cools faster than another object. These periods of low contrast and contrast inversion are called crossover periods (or inversion periods ) and can occur at any time of the day but are often encountered in the early morning and afternoon. The same effect can be observed in passive mmW imagery as both imaging modalities are based on the detection of greybody emission from objects. During these periods, objects can become difficult to detect as they blend in with the surrounding background which can hamper persistent surveillance efforts.
Work by Felton et al.  has shown that the polarimetric output of mid-wave IR and long-wave IR sensors can be used to mitigate the contrast reduction of crossover periods. The IR radiometric contrast of objects can be different for various linear polarization states of radiation due to the polarization dependence of reflection and emission. This is especially relevant for passive mmW sensors since many of these sensors are only able to detect a single linear polarization state [9,10]; the reason for this is the polarization dependence of mmW components (such as waveguides and mixers) which can limit operation to a single linear polarization state. If there is a large change in the contrast of mmW imagery for different polarization states, then this effect will need to be accounted for in the design of passive mmW sensors.
1.2 Polarization dependence of crossover periods
Passive mmW sensors operate on the same principles as IR detectors in that they detect the radiation emitted by objects . One of the largest differences between passive mmW imagery and IR imagery is the effect of reflections by objects . Objects are typically more reflective at mmW frequencies compared to IR frequencies and this leads to a diminished dependence on the kinetic temperature of objects in a scene for passive mmW images. Reflections from the radiometrically cold sky  often produce large radiometric contrasts between objects in a scene that can be on the order of 100 K. For smooth dielectric materials, the radiometric temperature of an object can be described by
At normal incidence angles, the reflection coefficient of a dielectric object will be the same for all linear polarization states and there will not be a large polarization component to the radiometric temperature of the object. At greater incidence angles, various linear polarization states will have different reflection coefficients which can result in a polarization dependent radiometric temperature and contrast. The change in radiometric temperature with respect to the kinetic temperature of the object
In practice, objects will behave in a more complex manner with common background targets consisting of non-uniform objects such as rocks, leaves, debris, etc. The same principle applies to these complex objects in which the diurnal heating cycle will cause diurnal changes in radiometric temperature, although in a less predictive manner. An exception to this would be metal targets reflecting nearly constant radiometric temperatures such as the sky (assuming no transient atmospheric effects such as clouds, etc.).
This experiment looks at two situations for polarimetric diurnal contrast changes: the radiometric contrast between two different materials and the radiometric contrast between two targets of the same material orientated at two different angles. In the first case, the kinetic temperature of the objects will be different throughout the day which will cause different radiometric temperature changes between the objects. In the second case, the materials will have nearly the same kinetic temperature but the radiometric temperature will still change at a different rate from the kinetic temperature as shown in Eq. (2).
A passive polarimetric mmW sensor  operating at 77 GHz was used to image several targets over a 24 hour period in order to observe polarimetric dependent crossover periods. The test began at 18:00 on August 29th, 2012 and ended at 17:00 on August 30th, 2012. Measurements were taken every hour for most of the period specified and included the kinetic temperature of each object, ambient temperature, humidity, solar irradiance, and the mmW radiometric sky temperature at zenith. The kinetic temperatures were measured via a contact temperature probe which was placed at the same place on each target every hour. Solar irradiance was measured with a Pyranometer. The mmW radiometric sky temperature was measured by pointing the mmW sensor at zenith.
The targets are shown in Fig. 1(a) and consisted of three different materials on top of a brick background orientated either normal to the sensor or flat on the ground (~75° incident angle relative to the sensor). Metal targets are on the left side of the image and are made of 3 mm thick aluminum. The middle target is a layer of dirt and rock collected from Yuma, AZ and is suspended by a wooden sheet. On the right are two wooden targets made from 3 mm thick plywood. The two wooden targets will be referred to as Wood0 (for the wooden target orientated normal to the sensor) and Wood75 (for the wooden target lying flat on the ground). The visible camera was not bore-sighted with the mmW antenna and parallax is apparent between the visible and mmW image. The target distances ranged from 8 – 10 m from the sensor which was mounted on a tripod at a height of 2.6 m.
The metal targets are used to measure the variation in the reflected radiometric temperature throughout the experiment and are orientated at the same angles as the wooden targets. For a cloudless day the reflected radiometric sky temperature should be a near constant value as the sky temperature will not change greatly (< 5 K) during a day . The wooden targets are used to observe the role that orientation plays in changing the contrast of an object throughout a diurnal cycle. The dirt target was chosen as a complex object that is a common background material.
The passive mmW sensor used for this test is a single pixel scanning sensor based on optical up-conversion  and is capable of detecting two orthogonal linear polarization states simultaneously via an orthomode transducer. This allows the data from the two polarization channels to be co-located for each image. The two polarization states will be referred to as horizontal and vertical which correspond to linearly polarized radiation orientated horizontally and vertically relative to the earth. The field of view of the sensor was 15° x 8° and divided into 0.1° sections for an image size of 150 x 80 pixels. The sensor took approximately 8 minutes to create each image with an integration time of 40 ms per pixel. The sensor is diffraction limited with a resolution of ~8 mrad with an aperture size of 0.6 m. The noise equivalent temperature differences (NETD) of the two polarization channels are ~1 K/√Hz. The sensor was calibrated every hour using a two point calibration technique .
There were no clouds present during the testing period. The ambient temperature ranged from 17 - 31 °C and the measured solar irradiance ranged from 0 – 870 W/m2. The absolute humidity did not change significantly with an average value of 12 g/m3. Using the measured temperature and humidity values, along with the parameters of the US STD propagation model, the mmW sky temperature at zenith was calculated to be 36 K at zenith with a variation of less than 5 K throughout the day based on MPM . This value was confirmed by the mmW sensor measurements of the sky radiometric temperature although the actual variation was too low for the sensor to measure due to the noise floor and stability of the system. The variation of the reflected signal from the targets was measured by recording the radiometric temperature of the two metal targets. These measurements confirmed the stability of the reflected radiometric temperatures throughout the experiment.
The mmW sensor produces two images, one for each of the linear polarization states that it detects. The images are radiometrically calibrated so that the received mmW intensity is mapped to a radiometric temperature. For the two polarizations, the radiometric temperature of each object was measured by averaging the mmW image over the regions shown in Fig. 1(c) to reduce the effect of non-uniformity from the target surface.
In addition to the two linear polarization radiometric temperatures (Tv and Th), the S0 and S1 Stokes parameters were also calculated. The S0 parameter was calculated by adding the radiometric temperature of the two polarizations together and the S1 parameters was calculated by subtracting the two signals. This results in a total of four radiometric values for each target which are designated as Tv, Th, S0, and S1. Error bars are calculated based on the standard deviation of the measured radiometric temperature over the entire target and standard propagation of errors for the calculated contrasts.
There are two contrast values of interest in this experiment as discussed in the preceding section: the contrast between the Dirt and Brick targets and the contrast between the Wood75 and Wood0 targets. The contrast values are calculated by taking the difference between the mean radiometric temperatures of the two targets of interest for each of the four parameters. Figure 2 shows the calculated contrast between the Brick and Dirt targets for the four polarization parameters. The kinetic temperature graph in Fig. 2 shows the lower thermal inertia of the Dirt target compared to the Brick target as the Dirt target undergoes a greater change in kinetic temperature over the diurnal period.
There are three crossover events evident in the contrast plot. One occurs for the S0 parameter near 22:00 and the other two events are for the Tv parameter at 08:00 and 13:00. The maximum contrast value of the Th parameter (9 K) occurs at 13:00 while the Tv contrast at the same time is 0 K. This is relevant for sensors capable of detecting only one linear polarization state as it shows that the choice of which polarization to detect can greatly affect contrast depending on the targets of interest. The images in Fig. 2 clearly show the low contrast for Tv compared to the other parameters at 08:00.
The wooden targets present a more complex arrangement due to the relatively high transmission of thin plywood at this frequency range. The transmission of these wooden targets was measured to be ~40% at normal incidence. This means that the radiometric temperatures of the wood samples were also functions of the radiometric temperature of the brick background. Although the transmission is high, the actual effect is negligible for two reasons. The first is that the radiometric temperature of the brick is similar to the kinetic temperature of the wood targets (on average 14 K difference) while the reflected radiometric temperature of 191K is a much greater difference from the kinetic temperature of the wood targets (~110 K difference). Therefore, small changes in reflectivity will still have large changes in the radiometric temperature of the targets. The second reason is that the radiometric temperature of the brick was linked to the kinetic temperature of the wooden targets in that they both rose and fell together at similar rates. Consequently, the transmission term acts as an offset of the contrast value and does not change the overall diurnal trend. For accurate modeling purposes the transmission term must be taken into account as it does change the absolute radiometric temperature by several kelvins.
The contrast between the different wood orientations is shown in Fig. 3. As shown in Fig. 3, the kinetic temperatures of the two targets are generally equal, while the contrast changes throughout the diurnal period. In this case there are no crossover periods for any of the parameters but there is still a large difference in contrast between the parameters. The Tv parameter has 6 K greater contrast on average than the Th parameter.
Using the measured radiometric temperatures as the input to a mmW ray tracing program , a simulation of the measured scene was created. The simulated image in Fig. 4 shows what the ideal scene would look like; no sensor effects are incorporated and a constant radiometric sky temperature profile is used. Dielectric properties are based on published literature . From the embedded movie in Fig. 4 (Media 1), the differences between the two polarization states over a diurnal period are apparent with the contrast of the horizontal polarization changing less than the contrast of the vertical polarization. The reason for the large difference is due to the geometry of the situation with objects and the ground set-up parallel and perpendicular to the measured polarization states. The horizontal polarization state corresponds to s-polarization while the vertical polarization state corresponds to p-polarization which has a minimum reflectivity at the Brewster angle. Since the Brewster angle maximizes radiometric temperature coupling to the kinetic temperature of the object and minimizes reflectance from the sky, the latter is more sensitive to the object’s kinetic temperature than the former. The contrast between orthogonal linear polarization states offset from the horizontal and vertical states by 45° would be less, due to the symmetry of the experimental set-up. For different viewing aspects the differences in polarization may be less, for example an aerial view looking at nadir angles (which would have many objects normal to the sensor) would produce a smaller polarization signal.
Several targets were observed over a 24 hour period with a passive mmW sensor capable of measuring multiple polarization states of radiation. This experiment has shown that, similar to IR sensors, the contrast between objects can change over a diurnal cycle and this change is different for distinct polarization signals. This is especially important for persistent surveillance applications where the loss of contrast at certain times of the day could prove detrimental for observing/tracking targets of interest. Crossover events were observed for different polarization states which would render passive mmW sensors unable to detect a target independent of how sensitive the sensor was. This highlights the need to understand the role polarization plays in scene contrast and also the importance of polarization in the design of future passive mmW sensors.
The unpolarized contrast between objects observed during this experiment changed by over 10 K during a diurnal cycle. The polarized contrast also exhibited large changes which were often offset from the unpolarized contrast or followed different patterns. By adding polarization detection capabilities to a sensor, it may be possible to optimize the contrast of targets within a scene and avoid periods of low contrast. This benefit is countered by the increased complexity required for detection of multiple polarization parameters. For the case of several current passive mmW sensors that are capable of detecting only one linear polarization state at a time, it becomes important to understand the targets of interest before selecting which linear polarization state the sensor will be capable of detecting. In the simplest case, low contrast in an image using this type of sensor could be mitigated by rotating the sensor in order to capture a different linear polarization state which could result in greater contrast.
The contrast of objects will be very scene and application specific, while this experiment looked at only two very specific examples. Changing the materials or angles in this experiment would have resulted in different crossover periods or possibly no crossover periods. These cases were not meant to be representative of every case, but rather serve to demonstrate the potential of using polarization data to provide additional information. Further research could focus on developing general equations and simulations to further the understanding of how large a role polarization plays in contrast during diurnal cycles. This information could then be used by sensor designers to determine if the complexity of adding polarization detection capabilities would be beneficial.
This research was funded by the Department of Energy (Dr. Victoria Franques, Program Manager) under contract number DE-NA0000585.
References and links
1. H. J. Liebe, T. Manabe, and G. A. Hufford, “Millimeter-wave attenuation and delay rates due to fog/cloud conditions,” IEEE Trans. Antenn. Propag. 37(12), 1612–1617 (1989). [CrossRef]
2. W. L. Stutzman and W. K. Dishman, “A simple model for the estimation of rain-induced attenuation along earth-space paths at millimeter wavelengths,” Radio Sci. 17(6), 1465–1476 (1982). [CrossRef]
3. D. Wikner, “Millimeter-wave propagation through a controlled dust environment,” Proc. SPIE 6548, 654803, 654803-9 (2007). [CrossRef]
4. E. J. Boettcher, K. Krapels, R. Driggers, J. Garcia, C. Schuetz, J. Samluk, L. Stein, W. Kiser, A. Visnansky, J. Grata, D. Wikner, and R. Harris, “Modeling passive millimeter wave imaging sensor performance for discriminating small watercraft,” Appl. Opt. 49(19), E58–E66 (2010). [CrossRef] [PubMed]
5. L. Yujiri, M. Shoucri, and P. Moffa, “Passive millimeter wave imaging,” IEEE Microw. Mag. 4(3), 39–50 (2003). [CrossRef]
6. R. Appleby, “Passive millimetre-wave imaging and how it differs from terahertz imaging,” Philos Transact A Math Phys Eng. Sci. 362, 379–392, discussion 392–394 (2004).
7. D. L. Shumaker, J. T. Wood, and C. R. Thacker, Infrared Imaging Systems Analysis, (DCS Corporation, Alexandria, 1993), Chap. 2.
8. M. Felton, K. P. Gurton, J. L. Pezzaniti, D. B. Chenault, and L. E. Roth, “Measured comparison of the crossover periods for mid- and long-wave IR (MWIR and LWIR) polarimetric and conventional thermal imagery,” Opt. Express 18(15), 15704–15713 (2010). [CrossRef] [PubMed]
9. R. Appleby, R. N. Anderton, S. Price, N. A. Salmon, G. N. Sinclair, J. R. Borrill, P. R. Coward, V. Paraskevi Papakosta, A. H. Lettington, and D. A. Robertson, “Compact real-time (video rate) passive millimeter-wave imager,” Proc. SPIE 3703, 13–19 (1999). [CrossRef]
10. A. H. Lettington, D. Dunn, N. E. Alexander, A. Wabby, B. N. Lyons, R. Doyle, J. Walshe, M. F. Attia, and I. Blankson, “Design and development of a high-performance passive millimeter-wave imager for aeronautical applications,” Opt. Eng. 44(9), 093202 (2005). [CrossRef]
11. J. P. Wilson, D. G. Mackrides, J. P. Samluk, and D. W. Prather, “Comparison of diurnal contrast changes for millimeter-wave and infrared imagery,” Appl. Opt. 49(19), E31–E37 (2010). [CrossRef] [PubMed]
13. A. D. Sayers, “Radiometric sky temperature measurements at 35 and 89 Ghz,” Microwaves, Antennas and Propagation, IEE Proceedings H 133, 233 –237 (1986). [CrossRef]
14. J. P. Wilson, C. A. Schuetz, T. E. Dillon, P. Yao, C. E. Harrity, and D. W. Prather, “Passive 77 GHz millimeter-wave sensor based on optical upconversion,” Appl. Opt. 51(18), 4157–4167 (2012). [CrossRef] [PubMed]
15. H. J. Liebe, “MPM—An atmospheric millimeter-wave propagation model,” Int. J. Infrared Millim. Waves 10(6), 631–650 (1989). [CrossRef]
16. C. A. Schuetz, J. Murakowski, G. J. Schneider, and D. W. Prather, “Radiometric Millimeter-wave detection via optical upconversion and carrier suppression,” IEEE Trans. Microw. Theory Tech. 53(5), 1732–1738 (2005). [CrossRef]
17. W. N. Hardy, “Precision Temperature Reference for Microwave Radiometry (Short Papers),” IEEE Trans. Microw. Theory Tech. 21(3), 149–150 (1973). [CrossRef]
18. M. Murakowski, J. Wilson, J. Murakowski, G. Schneider, C. Schuetz, and D. Prather, “3D rendering of passive millimeter-wave scenes using modified open source software,” Proc. SPIE 8022, 80220B, 80220B-8 (2011). [CrossRef]
19. J. W. Lamb, “Miscellaneous data on materials for millimetre and submillimetre optics,” Int. J. Infrared Millim. Waves 17(12), 1997–2034 (1996). [CrossRef]