Smart Pixels with Smart Illumination
Robert B. Darling
A compact optical design is proposed that allows integrated emitter and detector arrays to illuminate and sense remote objects.Several potential sensor applications are presented.
Smart pixels, the integration of photodetector arrays and processing electronics on a single semiconductor chip, have been driven by its capability to perform parallel processing of large pixelated images and in real-time reduce a complex image into a manageable stream of signals that can be brought off-chip.1-2In recent years, optical modulators and emitters have been integrated with photodetectors and on-chip electronics.3-5The driving applications for this have been highly parallel optical interconnects and optical switching networks.In this paper, we propose and discuss the advantages of a new class of smart pixels sensors that take advantage of structured illumination to enhance their functionality: smart pixels with smart illumination (SPSI).
In order to function, photodetector sensor arrays require the objects they sense to be illuminated by a light source.The simplest arrays rely on ambient light and suffer from the effects ofspatial, spectral, and temporal variations in the illumination.Built-in illumination sources can greatly reduce the deleterious effects of ambient light, however there is still noise from the ambient background.Built-in illumination can either flood the image or supply an array of beams that match the pixel spacing. Non-integrated sources typically require bulky beamsplitters.The structured illumination has the advantage of more efficient light usage, but introduces tight imaging tolerances.These tight tolerances on uniformity, rotation, translation, and defocus lead to high design and fabrication costs.High optical efficiency (greater than 25% efficiency) requires polarization optics that are expensive, bulky, and impose limits on the sensed image’s polarization properties.Even if it were easy, illuminating the image with a fixed uniform light source or an array of uniform beams is not ideal.A fixed illumination level, set to keep the majority of the photodetectors within their dynamic range, results in bright areas and dull areas of the image becoming saturated or lost in the noise, respectively.In addition, the cumulative optical power output from the illuminator is limited by power consumption and thermal dissipation constraints.A structured illumination source that is dynamically controllable would alleviate these problems.
If an emitter array and detector array were integrated on a single chip, a compact optical sensor design with individual control of each pixel’s illumination could be constructed.Efficient use of an integrated emitter and detector sensor array requires an optical system that images the emitter array onto a remote object and then efficiently and accurately images the illuminated spots onto the detector array.(A conventional imaging system would image the illuminated spots back onto the emitters.)Various schemes employing bulk image shifters can be envisioned, but are not practicable due to there bulk and alignment tolerances.In this paper, we propose a compact and practical optical design for enabling an integrated array of emitters and detectors to illuminate and sense a remote object.The design allows each pixel to control its illumination through feedback from the detectors.The smart pixels with smart illumination (SPSI) concept has several potential applications in sensor array technology, as will be discussed.
The proposed SPSI imaging system utilizes a 1x2 binary phase grating (BPG) like those used in free space optical interconnect and for image split and shift modules.6,7The optical system is shown in Figure 1 for a single pixel.Though only an on-axis pixel and f to f Fourier imaging are shown, the design works for an array of pixels and for other imaging configurations.
The light from the emitter is split into two beams by the BPG that then illuminates two spots (A and B) on the object.The return light beam from each spot is split into two beams:one beam illuminating a detector and the other beam retracing back to the emitter.Detector PD-A (PD-B) detects the reflected spot from A (B). As the BPG can be close to 100% efficient in splitting beams, this system has an impressive maximum 50% optical efficiency.This optical design is compact, efficient, low cost, requires minimal alignment, and has no polarizing elements.It eliminates the need to tightly match the focus, translation, and rotation of the detectors and illuminators.
Since BPG are not 100% efficient, light due to the undiffracted order of the BPG will hit spot X on the object and spots Y on the chip.These stray spots do not affect the operation of PD A and B.The second and higher order diffraction of the BPG will eventually need to be considered, but can be made small by proper design of the BPG.Note that half the return light images back on the emitter.With LED arrays, such optical feedback is not a factor.For laser arrays (e.g. VCSELs), such feedback may be intolerable when the object is highly reflecting.One possible solution is an array of miniature optical isolators positioned in front of the emitters.
Figure 1:Optical System
The proposed optical design and recent developments in emitter and detector integration now allow each pixel to operate either independently or cooperatively with their neighbors and their illumination sources.The illumination of each pixel can be adjusted or modulated in accordance with local conditions and various sensing needs.The average illumination received by each detector may be adjusted to maximize the detector’s dynamic range.The illumination could be modulated to perform background subtraction.The illumination could dynamically be concentrated on the parts of the object of current interest, greatly relieving power consumption and dissipation constraints while significantly boosting the signal to noise ratio.The dual illumination of the object by each emitter allows differential detection to be performed at each pixel.
SPSI Edge Detector Array
As an example of one type of SPSI sensor, we discuss here the operation of an edge detector capable of detecting small (a few percent) local variations in reflectivity in an image whose reflectivity slowly varies by several orders of magnitude. The SPSI edge detector uses dual illumination and optical feedback to control the average illumination level of the two detectors in each pixel (PD-A and PD-B).The emitter output of each pixel is adjusted so that the sum of the signals from the pixel’s two detectors matches a user defined level.This level is likely in the middle of the detectors’ dynamic range, but could be set to exploit non-linearities in the detector’s response.The output of each pixel is the difference between the signals of the two detectors and represents the normalized spatial derivative of the reflectivity of the sample, specifically (RA-RB)/(RA+RB), where RA (RB) is the reflectivity of the spot A (spot B).
To illustrate the edge detector array operation, consider the upper plot in Figure 2 of the reflectivity of an object that has a dull region (0.1% reflectivity), a bright region (50% reflectivity), and a region in which the reflectivity ramps from dullto bright.In each region, there is an area of 10% increase or decrease in reflectivity.In addition, the region of ramped reflectivity has an area with a rounded off (sinusoidal, to be exact) increase in reflectivity, with a 20% peak increase.To detect these regions with uniform illumination would require the detectors and processors to have dynamic ranges of over 5000:1.In the lower plot of Figure 2 is the simulated output of the SPSI edge detector array. The spikes represent the edges of the sharp increases or decreases in reflectivity.The region of ramping reflectivity shows up as a small constant output.Even rounded off edges show up well.By providing the appropriate light level at each point of the object, greater signal to noise ratio is achieved with significantly less power consumption.
Figure 2Simulated Output of SPSI Edge Detector
In this paper, we propose a compact optical design for enabling an integrated array of emitters and detectors to illuminate and sense a remote object.Our compact optical design for allows integrated emitter and detector arrays to dynamically and efficiently illuminate and sense remote objects. Each pixel can control its illumination level or modulation characteristics through feedback from the pixel’s dual detectors.The smart pixels with smart illumination concept has several potential applications in sensor array technology including edge detection; dynamic spotlight tracking system8; a tracking sensor that monitors focus, translation, and rotation of a scribe line; winner-take-all neural networks that highlights the “winning” pixel; tracking of non-linear (e.g fluorescent ) objects; intensity compression algorithms; tracking, differential sensing, and error correction in page oriented memory and processing systems; pixel by pixel background subtraction; and ranging.
1.M. W. Derstine, W. R. Babbitt, R. G. A. Craig, A. L. Lentine, F. B. McCormick, R. A. P. Tooley, and A. C. Walker, Photonics Spectra 25, issue 5, 139 (May 1991).
2.A. L. Lentine and D. A. B. Miller, IEEE J. Quantum Electronics, 29, 655-669 (1993).
3. K.V. Shenoy, C.G. Fonstad, A.C.Grot, and D. Psaltis, IEEE Phot. Tech. Lett. 7, 508 (1995).
4.H. C. Liu, J. Li, Z. R. Wasilewski, and M. Buchanan, Electronics Letters 31,. 832-3 (1995).
5. K. W. Goossen et al, IEEE Photonics Technology Letters 7, 360-2 (1995).
6.Taghizadeh-M-R.Redmond-I-R.Walker-A-C.Tooley-F-A, P. Smith-S-D.,Taylor-W. Proceedings of the SPIE 883 245 (1988).
7. J. Jurgen and B. Brumback, Opt. Comm. 76, 318 (1990).
8.Developed in collaboration with Susan A. Tonkin of the University of Washington.
9.A. C. Grot, D. Psaltis, K. V. Shenoy, and C. G. Fonstad, Jr., IEEE Photonics Technology Letters 6, 819 (1994)
For more information on Smart Pixels with Smart Illumination
Smart Pixels with Smart Illumination: Concept (Slides)
Smart Pixels with Smart Illumination: Concept (Word Document)