Radiant energy – Photocells; circuits and apparatus – With circuit for evaluating a web – strand – strip – or sheet
Reexamination Certificate
2002-03-21
2004-03-16
Porta, David (Department: 2878)
Radiant energy
Photocells; circuits and apparatus
With circuit for evaluating a web, strand, strip, or sheet
C250S2140VT, C356S005100, C356S005010
Reexamination Certificate
active
06707054
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the field of three-dimensional image capture and, in particular, to image capture with a scannerless range imaging system having improved dynamic range.
BACKGROUND OF INVENTION
A means for acquiring range data of an entire scene without employing a range scanner was proposed in U.S. Pat. No. 4,935,616, entitled “Range Imaging Laser Radar”, which issued Jun. 19, 1990 in the name of Marion W. Scott (and further described in the Sandia Lab News, vol. 46, No. 19, Sep. 16, 1994). Instead of scanning a scene, this system acquires range by taking repeated images under slightly altered exposure conditions. More specifically, the scannerless range imaging system disclosed therein uses either an amplitude-modulated high-power laser diode or an array of amplitude-modulated light emitting diodes (LEDs) to completely illuminate a target scene. Conventional optics confine the target beam and image the target onto a receiver, which includes an integrating detector array sensor having hundreds of elements in each dimension.
The range to a target is determined by measuring the phase shift of the reflected light from the target relative to the amplitude-modulated carrier phase of the transmitted light. To make this measurement, the gain of an image intensifier (in particular, a micro-channel plate) within the receiver is modulated at the same frequency as the transmitter, so the amount of light reaching the sensor (a charge-coupled device) is a function of the range-dependent phase difference. A second image is then taken without receiver or transmitter modulation and is used to eliminate non-range-carrying intensity information. Both captured images are registered spatially, and a digital processor is used to operate on these two frames to extract range. Consequently, the range associated with each pixel is essentially measured simultaneously across the whole scene.
The preferred method of estimating the range in the '616 patent uses a pair of captured images, one image with a destructive interference caused by modulating the image intensifier, and the other with the image intensifier set at a constant voltage. However, a more stable estimation method uses a series of at least three images, each with modulation applied to the image intensifier, as described in commonly assigned U.S. Pat. No. 6,118,946, entitled “Method and Apparatus for Scannerless Range Image Capture Using Photographic Film” and issued Sep. 12, 2000 in the names of Lawrence A. Ray and Timothy P. Mathers. In that patent, the distinguishing feature of each image is that the phase of the image intensifier modulation is unique relative to modulation of the illuminator. If a series of n images are to be collected, then the preferred arrangement is for successive images to have a phase shift of 2&pgr;
radians (where n is the number of images) from the phase of the previous image. However, this specific shift is not required, albeit the phase shifts need to be unique. The resultant set of images is referred to as an image bundle. The range at a pixel location is estimated by selecting the intensity of the pixel at that location in each image of the bundle and performing a best fit of a sine wave of one period through the points. The phase of the resulting best-fitted sine wave is then used to estimate the range to the object based upon the wave-length of the illumination frequency, and the range from the camera to the object at a particular pixel can be readily ascertained.
Consequently, the analysis on the image bundle described by Ray et al. differs from the analysis proposed by Scott, requiring at least three images in the bundle. Additionally, the range resolution depends upon the ability to perform a least-squares-estimate of the data to a known functional form. If the data in the image is too noisy, because of under-exposure or over-exposure, the resulting range estimates will degrade. Since the system collects area-wide data in parallel it is often the case that one region will have adequate exposure levels, while other regions have less suitable exposure characteristics.
In both methods, the image is illuminated with an amplitude modulated light source. The optical path of the receiver is fitted with a optical system containing a micro-channel-plate, i.e., as in a night vision system, with the amplification of the reflected signal being modulated at the same frequency as the illuminator. In the method described by Ray et al., for the first image it is preferred not only that the illuminator and the receiver have the same frequency, but that the phase of the two devices match. Subsequent images are captured according to the method described by Ray et al. in a similar manner, but the relative phase relationship of the illuminator and the receiver are shifted by a known angle.
A digital imaging system is the preferred approach for implementing the range imaging system, since it is easier to control registration of all images within the image bundle in a digital imaging system. However, digital imagers have a limited number of exposure quantization levels, e.g., a standard consumer digital camera has 256 exposure quantization levels. In order to estimate the range, the variation of values at a given pixel within the image bundle have to be sufficient to dominate any noise within the system. In particular, if a pixel has several values that are at the maximum exposure levels, then the resulting range estimate is dubious. In general, a wider variation in the values at a pixel location in an image bundle is preferred.
It is often the situation in practice that some regions of the image bundle will have acceptable variations, while other regions will not have acceptable variations. It would be desirable to accommodate this problem. One approach is to collect multiple image bundles using a suite of exposure settings. Analysis of each image bundle is performed while maintaining a measure of performance, and then the range estimate at a particular pixel that has the best performance measure is selected. For instance, if four exposure periods are utilized, of say ¼ second, ½ second, 1 second and 2 seconds, then regions that are under-exposed with the ¼ second exposure period are likely to have better exposure characteristics in the image exposed at say 1 second. While this method does accomplish the objective, it has the limitation that the number of exposures quadruples and the time to estimate range more than quadruples. However, it is also the case that many digital imagers have more than adequate spatial resolution, and that a trade-off of spatial resolution for improved range resolution is not only feasible, but desirable as well.
Recently, an approach to improving dynamic range in visible images was proposed (see Shree K. Nayar and Tomoo Mitsunaga, “High Dynamic Range Imaging: Spatially Varying Pixel Exposures, Proceedings of Computer Vision and Pattern Recognition 2000 and International Publication No. WO 01/63914 A1, “Method and Apparatus for Obtaining High Dynamic Range Images,” which published Aug. 30, 2001 in the names of Nayer and Mitsunaga). Method and apparatus are described for obtaining relatively high dynamic range images using a relatively low dynamic range image sensor without significant loss of resolution. The image sensor has an array of light-sensing elements with different sensitivity levels in accordance with a predetermined varying sensitivity pattern for the array of light sensing elements. The predetermined varying sensitivity pattern is provided by employing a filter array prior to the imaging plane. In this case the filter array is a series of tiles, with each tile covering an area of 2×2 pixels. Each tile includes an array of transmittance filters providing a spatially varying pattern of transmittance.
SUMMARY OF THE INVENTION
It is an object of the invention to improve the range resolution of a scannerless range imaging system by using a high dynamic range imager.
The present invention is directed to overcoming one
Lee Patrick J.
Porta David
Woods David M.
LandOfFree
Scannerless range imaging system having high dynamic range does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Scannerless range imaging system having high dynamic range, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Scannerless range imaging system having high dynamic range will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3226718