A TECHNIQUE FOR THE FASTRASTER-SCANNED MOTIONOF A SMALL WINDOWFOR MILLIMETER-WAVEIMAGE ACQUISITIONJose Soto1 and J. A. Davila1´ ´1 Instituto de Fısica´Universidad Autonoma de Puebla´Puebla, Pue, 72570, Mexico

Recei�ed 17 September 2001

ABSTRACT: Here, we propose a technique that utilizes a Nipkow diskand an integrating sphere for the fast acquisition of millimeter-wa�eimages. This technique has the potential for quietly achie�ing �ideoframe rates. A demonstration of the principle is made in the opticalregion. � 2002 Wiley Periodicals, Inc. Microwave Opt Technol Lett32: 440�442, 2002.

Key words: raster scanning; integrating sphere; millimeter-wa�e imagingDOI 10.1002 � mop.10203

Millimeter-wave imaging has recently become a vigorousresearch area because of its demonstrated ability to seeunder a variety of adverse environmental conditions, such asfog, clouds, snow, smoke, rain, etc., as well as through walls,

� �doors, and most common clothing materials 1�4 . Among thecivilian applications being developed for imaging in this spec-tral range are contraband and concealed weapon detection,through-wall imaging, surveillance within buildings, remoteforest-fire detection, underground metal detection, low-visibility autonomous landing of aircrafts, and many more.

For short-range applications, both active and passive imag-ing systems are being perfected. In active imaging, the objectis illuminated with a source of millimeter waves, and a largelens or a focusing mirror is used to form an image of theobject with the power reflected by it. In passive imaging, theelectromagnetic power is thermally emitted by the object,and a focusing element forms an image of it. For long-range applications, mostly passive imaging systems are beingdeveloped.

Once the image is formed, either actively or passively,there are several methods for acquiring it. Early systems usedsingle detectors at the focal plane, the detector being me-chanically scanned across the scene of interest. Due princi-pally to the slow mechanical scan motion, these systems tooktens of minutes to take single-frame images. The present goalin millimeter-wave imaging research is to drastically reducethe image acquisition time in order to approach TV framerates. One of the important approaches being investigated toreach this goal is the use of focal plane solid-state arrays withindividualized amplification and detection. The other impor-tant approach is the combination of raster motion and asingle receiver channel. The raster motion is either of theimage in front of a fixed receiver by means of a flapping

� �reflector 1, 5 or of the receiver in front of a fixed image� �3, 6 .

Although millimeter-wave imaging may be mostly domi-nated in the future by focal plane arrays, there are, however,several problems associated with them, like high cost, large

Contract grant sponsor: CONACYTContract grant number: 25789-A

size, cryogenic cooling, complexities associated with inter-� �channel gain stability, and others 1, 3 , so that, as present,

the field remains dominated by scanned systems; however,their success awaits the discovery of a compact and efficient

� �scanner capable of achieving video rates 3 .In this work, we propose a new scheme for raster-scanned

image acquisition in which neither the image nor the singlechannel receiver has to be scanned, and whose maximumframe rate will be determined ultimately by the speed andsensitivity of the single-channel receiver, and not by mechani-cal-motion limitations. The two principal components in thisnew scheme are a Nipkow disk and an integrating sphere.The Nipkow disk, which is utilized for scanning, is providedwith a number of spirally distributed sampling apertures.Historically, it was a widely utilized tool in the early days ofmechanical television, both at the sending and receiving ends

� �of the system 4 . The integrating sphere is utilized fortransferring a fixed fraction of the sampled electromagneticpower to a stationary single-channel receiver, independentlyof the position and state of motion of the sampling aperturewithin the scanning area. We next describe these two compo-nents in more detail.

In the disk, there are N sampling apertures, each one withlinear dimensions of a few wavelengths, having a uniformangular distribution around the disk. Their radial positionsvary successively from r to r , with uniform incrementsmin max

Ž . Ž .of magnitude given by � r � r � r � N � 1 . Thus, formax minŽ .the ith aperture, its radial position is r � r � i � 1 � r,i min

where i � 1, 2, . . . , N. This is represented graphically in Fig-ure 1. The disk could be made of a reflecting material, like ametal, or of a convenient absorbing material for millimeterwaves. The sampling apertures could be real air apertures orsimply made by the removal of the reflecting or absorbingmaterial, leaving the transparent material in it. An opaquemask, again of reflective or highly absorbing material, is to beplaced in front of the rotating disk. This mask contains awindow, designated by abcd in the figure, which representsthe scanning area of interest. Its shape, a truncated circularslice, has an angular dimension such that only one of thesampling apertures appears in it at a time, and when it

Figure 1 Representation of the rotating disk with 12 samplingapertures. Only the aperture within the window abcd is active; therest of them are assumed to be covered by an opaque mask

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 32, No. 6, March 20 2002440

disappears from sight, the following one immediately appearson the opposite side of the window in the next row. Theradial dimension of the window is less critical, but it shouldnot interfere with the samplings of the innermost and outer-most sampling apertures.

The number of curved rows in the scanning is directlydetermined by, and is equal to, the number N of the sam-pling apertures, while the number M of columns is deter-mined by the number of measurements taken within theangular displacement equal to 2��N of the disk, and can beprogrammed at convenience; however, it is expected to besimilar to N. Thus, the maximum integration time per sampleallowed for the single-channel receiver is 2���NM, where �is the angular speed of the disk. In order to prevent shapedistortions in the reconstructed image, the scanning areaabcd should be mapped with the same shape on the computerscreen, and in order to avoid intensity distortions, it must betaken into consideration that the linear density of samplingby the ith aperture is proportional to r �r .min i

The other important component of our technique is anintegrating sphere for millimeter waves. In optics, integratingspheres are widely utilized instruments for radiometric mea-surements. They consist of a hollow sphere whose interior iscoated with a layer designed to have a high diffuse re-flectance. When light from a source enters an integratingsphere, it loses all memory of the point and direction ofentrance at the entrance port of the sphere, as well as of itsoriginal polarization. At the exit port, the light intensitybecomes highly uniform and diffuse. These instruments havebeen used for many years to scramble or average highlydivergent light, or light distributions too large to be accom-modated on a detector, in order to obtain meaningful powermeasurements. It thus effectively averages the input radiationsuch that any beam inhomogeneity and detector nonuniform-

� �ities do not affect the measurement 7�9 . Here, we proposeto use what would be an integrating sphere for millimeterwaves, in order to measure the relative power transmitted bythe sampling aperture of the disk with a single-channelreceiver which is much smaller in collecting area than theimage scanned area. This is expected to work since, com-pared with the speed of the electromagnetic waves, the speedof the moving aperture is too small to perturb the homogeniz-ing property of the sphere.

The best design of an integrating sphere for millimeterwaves has to be investigated, in order to have a high diffusereflectance in its interior, as well as the minimum diameter ofthe sphere which still provides a convenient level of homoge-nization for the input radiation in its interior.

In optics, an important parameter of an integrating sphereis the throughput, which is defined as the ratio of the opticalpower exiting the sphere to the power entering it. It isdetermined by the sizes of the entrance and exit ports rela-tive to the area of the sphere, and by the reflectance of theinner coating. Under ordinary conditions of operation, the

� Ž Žthroughput of the sphere is given by � R� 1 � R 1 � � �d d..�� , where R is the reflectance of the coating, and � ande d

� are the relative areas of the exit or detector port and thee� �entrance port, respectively 8, 9 .

But, if for our application, the back of the disk has thesame high reflectance as the interior of the sphere, eitherdiffuse or especular, with the exception of the samplingapertures, and the disk is brought to move in close proximityto the entrance port for minimizing the leakage of radiation,then under these circumstances, the active sampling aperture

becomes the moving entrance port of the integrating sphere.Also, if by the proper application of the high-reflectancelayer the area of the exit port is reduced to the effectivecollecting area of the receiver, and the relative area ofleakage of radiation between the sphere and the rotating diskis represented by � , and if we also define the quantity1� � 1 � R, then for � , � , � , � � 1, we obtain an approx-d e 1

Žimate expression for the throughput as � � � � � � �d d e.�� .1

Although the best design of an integrating sphere formillimeter waves is still to be determined, for the sake ofillustrating the above result, let us assume that a 60 cmdiameter sphere covered internally with a 0.98 reflectancediffusive material adequately homogenizes the radiationwithin the sphere, and that we want to scan an area of 30 cmaverage width by 30 cm radial length with a receiver of 1 cm2

effective collecting area, and that in order to have 30 rowsand 30 columns, the areas of the sampling apertures aretaken as 1 cm2. We also assume that the leakage power isbrought to negligible levels through a proper mechanicalconstruction of the system, with tolerances smaller than awavelength. Then, with these characteristics, we find that thethroughput is near 0.004. This means that near 0.4% of thesampled power would be transferred to a single-channelreceiver. This value of the throughput could even be in-creased to higher values by placing several receivers insidethe sphere and adding their signal, so that they operate as asingle-channel receiver. In Figure 2, a layout of the proposedsystem is shown.

Although we do not have the means for testing thisconcept directly for millimeter waves, we have done the testfor visible wavelengths instead, although in a rudimentaryway. For this experiment, the sampling disk was made bycovering a 21 cm diameter Plexiglas disk with a circularphotolithographic mask containing 24 transparent samplingspots, as well as 720 radial marks on its periphery. Thesemarks are optically sensed in order to read the photodetectorsignal into the computer every 0.5. Another single mark onthe periphery was used for knowing when the first reading ofthe outermost sampling spot takes place. Once the 720 valuesare read into the computer, they are assigned 30 per row in24 rows. The diameter of the sampling spots is nearly 1 mm,and the abcd window is 15 wide and is radially placed

Figure 2 Schematic of the proposed raster-scanning system. Here,the focusing element is represented as a lens, but it could be aconcave reflector or a more complex arrangement

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 32, No. 6, March 20 2002 441

Figure 3 Experimental demonstration of the workability of theoptical version technique proposed here.

between 7.4 and 9.5 cm. A very crude integrating sphere wasobtained by removing the active parts of a 8 cm diameterspherical lightbulb, and painting it both internally and exter-nally with diffusive white-matte vinyl paint. The paint wasremoved in a 2 mm diameter spot, and the light escapingthrough it was detected with a highly sensitive New-Focusmodel 2151 photoreceiver. The image projected on the scan-ning area was given the shape of a number 2, measuring16 mm high and 12 mm wide, by using another photo-lithographic mask. An expanded 20 mW green beam of anair-cooled argon laser was used to obtain this image. Theback of the sampling disk was not painted white, and re-mained black in this case. No corrections were made forshape or intensity distortions in the reconstructed image,which is shown in Figure 3. Due to the slowness of ourelectronics, the time required for acquiring and reconstruct-ing the image was near 1 min.

In spite of the crudeness and deficiencies of our experi-mental system, we see that the reconstructed image closelyresembles the original object, thus validating our proposedmethod in the optical region. From this, we can extrapolatethat a well-designed integrating sphere for millimeter waveswill also convey a fixed ratio of the sampled power to asingle-channel receiver, and that the same scheme will beoperational at these wavelengths. Since it is expected to havehigh diffuse reflection from properly rugged metallic surfaces,it remains to see if the minimum acceptable size of thesphere with respect to the degree of homogenization of theradiation in its interior is also an acceptable size fromthe portability point of view.

Although we have treated a single scanned area with afixed spatial resolution, other scanning areas with differentresolutions could be obtained on the same disk at differentradial zones, with their proper sets of sampling apertureshaving the type of distributions as described above. Also, inan alternative configuration, instead of a flat disk, a cylindri-cal drum with the sampling apertures distributed around thecurves wall of the cylinder could be used. This would lead toa more compact design of the system.

ACKNOWLEDGMENT

The authors express their gratitude to their students SantiagoCeron and Jehu Mendez for their valuable experimental´ ´collaboration.

REFERENCES

1. R.M. Smith, B.M. Sundstrom, B.W Belcher, and D. Ewen,‘‘ROSCAM: A 95-GHz radiometric one-second camera,’’ Passive

Ž .millimeter-wave imaging technology II, R. M. Smith Editor , ProcSPIE, 1998, vol. 3378, pp. 2�13.

2. L. Yujiri, H.H. Agravante, S. Fornaca, B.I. Hauss, R.L. Johnson,R.T. Kuroda, B.H. Quon, A.W. Rowe, T.K. Samec, M. Shoucri,and K.E. Yokoyama, ‘‘Passive millimeter-wave video camera,’’

ŽPassive millimeter-wave imaging technology II, R. M. Smith Edi-.tor , Proc SPIE, 1998, vol. 3378, pp. 14�19.

3. R.N. Anderton, R. Appleby, J.R. Borril, D.G. Gleed, S. Price, N.A.Salmon, G.N. Sinclair, P. Papakosta, and A.H. Lettington, ‘‘Realtime passive mm-wave imaging,’’ Passive millimeter-wave imaging

Ž .technology II, R. M. Smith Editor , Proc SPIE, 1998, vol. 3378,pp. 77�33.

4. ‘‘Broadcasting,’’ The new encyclopedia Britannica, 1994.5. A. Luukanen and V.P. Viitanen, ‘‘Terahertz imaging system based

on antenna-coupled microbolometers,’’ Passive millimeter-waveŽ .imaging technology II, R. M. Smith Editor , Proc SPIE, 1998, vol.

3378, pp. 34, 44.6. B. Blume, J. Wood, and F. Downs, ‘‘Naval special warfare MMHW

data collection results,’’ Passive millimeter-wave imaging technol-Ž .ogy II, R. M. Smith Editor , Proc SPIE, 1998, vol. 3378, pp.

86�94.7. J.A. Jackez and H.F. Kuppenhein, Theory of the integrating

Ž .sphere, J Opt Soc Amer 45 1954 , 460�470.8. G.E. Miler and A.J. Sant, Incomplete integrating sphere, J Opt

Ž .Soc Amer 48 1958 , 828�831.9. J.F. Clare, Comparison of four analytic methods for the calcula-

tion of irradiance in integrating spheres, J Opt Soc Amer A 151998, 3086�3096. See also tutorial on integrating spheres inMelles-Griot 1995�1996 catalog, Melles-Griot Electro-Optics,Boulder, CO, sect. 68.

� 2002 Wiley Periodicals, Inc.

NULL-FORMING SYNTHESIS FORCYLINDRICAL PATCH ARRAYSP. Niemand,1 J. Joubert,1 and J. W. Odendaal11 Department of Electrical, Electronic and Computer EngineeringCentre for ElectromagnetismUniversity of PretoriaPretoria 0002, South Africa

Recei�ed 25 September 2001

ABSTRACT: A synthesis method is modified to synthesize patterns withmultiple nulls in an otherwise omnidirectional pattern for cylindricalmicrostrip patch arrays. The directi�e radiation pattern of a cylindricalmicrostrip patch is incorporated into the synthesis procedure. The effectof the number of array elements and interelement spacing is in�estigated.� 2002 Wiley Periodicals, Inc. Microwave Opt Technol Lett 32:442�446, 2002.

Key words: null forming; cylindrical array; microstrip patch antenna;multiple nulls; omnidirectional patternDOI 10.1002 � mop.10204

1. INTRODUCTION

Adaptive antennas play an important role in many mobilecommunication systems, providing increased cell coverage

�through antenna gain control and interference rejection 1,�2 . The increase in the numbers of both unwanted directional

interferences and strong nearby sources in modern mobilecommunication systems emphasizes the need to include null-steering capabilities for base-station antennas. The signalquality can be improved by using multiple nulls in the direc-

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 32, No. 6, March 20 2002442