Abstract: An image processing system and method which may be used for standalone deconvolution or, alternatively, may be employed as a staring point for very rapid convergence with subsequent use of prior art deconvolution methods. Processing speed is improved because the sequential requirement of the CLEAN method is relaxed. Fractional removal of noise is accomplished for multiple features within the image during the processing of a single subtractive iteration. Thus the number of iterations can be significantly reduced allowing dramatic reduction of the subtractive stage and a resulting increase in overall processing speed.

Abstract: A method of embedding digital data in a source image includes the steps of: a) generating a multi-level data image representing the digital data; b) convolving the multi-level data image with an encoding carrier image to produce a frequency dispersed data image; and c) adding the frequency dispersed data image to the source image to produce a source image containing embedded data. The data is recovered from the image by: a) cross correlating the source image containing embedded data with a decoding carrier image to recover the data image; and b) extracting the digital data from the recovered data image.

Abstract: Apparatus and an accompanying method for generating a semantically based, linguistically searchable, numeric descriptor of a pre-defined group of input images and which is particularly useful in a system for automatically classifying individual images, on a numerical basis, in, e.g., an image database, and, through a query-by-example paradigm, retrieving a desired image(s) therefrom. Specifically, a signature is computed for each image in a set using multi-level iterative convolution filtering, with pixel values supplied as input to each filtering level being separately convolved with each one of a set of predefined Gaussian kernels. Average and variance vectors, as collective numeric descriptors of all the images in the set, are separately computed across corresponding elements in all the image signatures for the set. A linguistic term, semantically descriptive of all the images in the set, is associated with the numeric descriptors of this set and, with this set, is stored in a database.

Abstract: A two dimensional video convolver generates visually acceptable images on a standard television receiver and includes first and second adder stages and two pixel time period delays. The first adder stage simultaneously receives binary data from three adjacent pixels in a predetermined line in a first direction of a picture to be reproduced on a display during each pixel scan period. The binary data of each of three pixels is processed with a predetermined weight to generate a binary first stage output signal representative of a weighted sum of the plurality of the three pixels in the predetermined line. The second adder stage is responsive to the binary first adder stage output signal during a time period for each of three predetermined adjacent lines in a second orthogonal direction of a picture to be reproduced on a display.

Abstract: An apparatus and a method are provided for smoothing out high frequency information in an interlaced computer generated display. A computer generated RGB signal is converted into a YUV signal, which is separated into its Y, U, and V components. The Y component is convolved by averaging a current scan line with a line above and a line below the current scan line so that black lines are lightened and white lines are darkened, thus avoiding flicker. Since convolution is only performed on the Y part of the YUV signal, only two line buffers, each having a width equal to the number of bits in the Y component only, are required.

Abstract: A method for detecting an amount of rotation or magnification in a modified image, includes the steps of: embedding a marker image having a pair of identical features separated by a distance d and oriented at an angle .alpha. in an original image to produce a marked image, the marked image having been rotated and/or magnified to produce the modified image; performing an autocorrelation on the modified image to produce a pair of autocorrelation peaks corresponding to the location of the features of the marker image in the modified image; and comparing the separation d' and orientation .alpha.' of the autocorrelation peaks with the separation d and orientation .alpha. of the features in the marker image to determine the amount of rotation and magnification in the modified image.

Abstract: The invention relates to a method of spatial filtering a point distribution representing a structure buried in noise and each of the points of which are defined by n coordinate values of an n-dimensional space, wherein for each point the function representing the number of the points in an area around the point concerned as a function of a dimension of the area is established, and a scaling coefficient equal to the exponent (a) of a simple power function is determined which approximates as near as possible the function determined in step a) in a certain area having the dimension (d), furthermore the scaling coefficients of the points of a n-dimensional stochastic point distribution of a type equivalent to the point distribution to be filtered are determined, the difference between the scaling coefficients of the point distribution to be filtered and the scaling coefficients of the stochastic point distribution are determined, and the points, the scaling coefficients of which are contained in the difference an

Abstract: An embodiment of the present invention locates facial features in an image by bandpass filtering the image and then performing morphological operations followed by a thresholding operation. This initial processing identifies candidate areas where facial features may be located. The candidate areas are evaluated by classifiers to determine if a facial feature, such as an eye or mouth, has been located.

Abstract: A technique, referred to as area-based interpolation, performs image interpolation. The system determines a curve by the pixel value at a location by taking the integral of a curve over a small area, where the size of the area is determined by the sampling size of a sampling cell. When the image is resampled with respect to a sampling cell that has a finer spacing, the system integrates the polynomial using a finer integration area. In accordance with the invention, the relation between the reintegrated, resampled high resolution image and the low resolution image is a function of an up-sampler, followed by a linear filter. The coefficients of the filter are independent of the data, but are dependent on the family of curves used. If the system models by, for example, a third degree or fourth degree polynomial everywhere in the image, then that model determines the number of coefficients that are in the filter.

Abstract: A method is described for utilizing discrete orthogonal basis to restore signal system, such as radio or sound waves and/or image system such as photographs or medical images that become distorted while being acquired, transmitted and/or received. The signal or image systems are of the linear type and may be represented by the equation ?B! ?o!=?i! wherein ?o! is an original signal or image, ?i! is a degraded signal or image and ?B! is a system transfer function matrix. The method involves estimating a signal-to-noise ratio for a restored signal or image. Next, is the selecting of a set of orthogonal basis set functions to provide a stable inverse solution based upon the estimated signal-to-noise ratio. This is followed by removing time and/or spatially varying distortions in the restored system and obtaining an appropriate inverse solution vector.

Abstract: An image forming apparatus for creating an image of a motion blur, a trail of a rapidly moving object, by sequentially creating each pixel of the motion blur from pixel values of an original image, the apparatus including an image memory, a pixel position specifying unit, an original image data storage, an original image pixel value generating unit, a pixel value calculating unit, and a pixel value writing unit.

Abstract: A system for the model-based compression of speckle images in which the pulse characteristics of the known coherent imaging system which generated the speckle image to be compressed are used together with a scatterer image model to modify that scatterer image to contain impulse information such that when the impulse information is convolved with the pulse, the speckle image is reproduced. The modified scatterer image is compressed and outputted together with the pulse characteristics so that the speckle image can be reproduced.

Abstract: An object movement estimator (400) predicts optical flow by estimating changes in sensed images using one-dimensional correlations. Sensed image portions are stored at a first instant of time and compared with sensed image portions at a second instant of time. An optimal flow is estimated by performing a one-dimensional correlation of the sensed image portions at the first and second instants of time, and a time-to-contact is estimated based on the predicted optical flow thus obtained.

Abstract: Noisy images of graphical or pictorial records, such as hand-written postal addresses, are cleaned by a method of analyzing the shapes of pixel clusters. According to such method, a set of structuring kernels is used to extract primitive shapes. The outcome of this extraction process is used to determine whether each pixel, in turn, is to be treated as part of a image element such as an alphanumeric character.

Abstract: Cell samples, stained with a fluorescent dye, taken up by DNA in the individual cells, are scanned with a cytometer, which measures the integrated value of fluorescent light/cell. The integrated values of all of the cells are compiled to create an histogram of cell counts versus integrated fluorescent light, representing a cell population of (a) cells having a complement of DNA, but not in the process of division (G.sub.0 phase), (b) cells having two full compliments of DNA, but which have not actually divided into two cells (G.sub.2 phase) and (c) cells which are in the process of replicating their DNA (S, separation phase). The percentages of cells in each of the phases, represented in the histogram as separated peaks of sizes proportional to the G.sub.0 and G.sub.2 populations, and separation S phase population, aids in the prognosis of a patient's cancer development. More serious malignancy is indicated by increased S and G.sub.2 phase populations. Errors, e.g.

Abstract: An object movement estimator (400) predicts optical flow by estimating changes in sensed images using one-dimensional correlations. Sensed image portions are stored at a first instant of time and compared with sensed image portions at a second instant of time. An optimal flow is estimated by performing a one-dimensional correlation of the sensed image portions at the first and second instants of time, and a time-to-contact is estimated based on the predicted optical flow thus obtained.

Abstract: The present invention pertains to a process which can be fully automated for accurately determining the alleles of STR genetic markers. More specifically, the present invention is related to performing PCR amplification on a location of DNA, labelling the PCR products, converting the labels into a signal, removing a reproducible PCR stutter pattern from the signal by means of a computational device, and then determining the genotype of the location of the DNA. The invention also pertains to genetics applications and systems which can effectively use this genotyping information.

Abstract: A method and apparatus for automating the design of morphological or template-based filters for print quality enhancement. A plurality of different phase, but same resolution, subsampled images are generated from training documents. Statistical data derived therefrom is then employed in an automated process to generate filters. The filters may be used for resolution enhancement and/or conversion of bitmap images. Furthermore, the statistical data is used to produce filters that are intended to not only optimize image structure, but image density as well.

Abstract: A circuit for recognizing a human face in a sequence of video is provided. The circuit includes a block structuring circuit in which the video is structured block-wise and wherein two consecutive frames are subtracted from one another. The circuit also includes a unit for block-wise post-processing the difference frames. The post-processing unit includes a comparator for determining the areas of the difference picture which may correspond to a moving object, a mapping unit for forming a status picture by mapping each block of the difference picture which corresponds to a moving object onto a binary "one", all other blocks being mapped to a binary "zero".The status picture is transformed by a transforming unit into a unidimensional feature function. The unidimensional feature function is then fed to a convolution unit to determine whether the feature function could include a "head-shoulder" scene.

Abstract: A noisy signal obtained from an acquisition system, such as a conventional or confocal microscope, is reconstructed in a manner which simultaneously estimates the true or ideal signal as well as the response function of the acquisition system. The restored signal can be a function of one or more variables, e.g., time or space. Prior knowledge of the response function of the acquisition system is not required, there is no the need to calibrate the acquisition system before acquiring a signal. The true signal and the response function of the acquisition system are estimated in iterative manner. The actual signal data and the frequency content of the actual signal data may be used to determine constraints to be applied to the estimates of the true signal and the impulse response of the acquisition system. These constraints include constraints on the signal (in the spatial domain) and constraints in the frequency domain.

Abstract: A technique is provided to remove noise from images and to enhance their visual appearance through the utilization of a technique which converts an image into a set of coefficients in a multi-scale image decomposition process, followed by modification of each coefficient based on its value and the value of coefficients of related orientation, position, or scale, which is in turn followed by a reconstruction or synthesis process to generate the enhanced image. Also contributing to the improved enhancement is a set of orientation tuned filters of a specialized design to permit steering, with the analysis and synthesis filters also having a self-inverting characteristic. Additionally, steerable pyramid architecture is used for image enhancement for the first time, with the steering being provided by the above orientation tuned filters.

Abstract: An object position detecting method and apparatus whereby even an image subjected to deformation such as shift, rotation, scaling, etc., can be recognized with high accuracy. Also disclosed is an image processing system that uses the object position detecting method and apparatus.

Abstract: A systolic array of processing elements is connected to receive weight inputs and multiplexed data inputs for operation in two dimension convolution mode, or fully-connected neural network mode, or in cooperative, competitive neural network mode. Feature vector or two-dimensional image data is retrieved from external data memory and is transformed via input look-up table to input data for the systolic array. The convoluted image or outputs from the systolic array are scaled and transformed via output look-up table for storage in the external data memory. The architecture of the system allows it to calculate convolutions of any size within the same physical systolic array, merely by adjusting the programs that control the data flow.

Abstract: A system for simultaneously convolving first and second digital binary images in parallel with a single convolver having a binary mask. A first window is selected from the first digital binary image and a second window is selected from the second digital binary image. A first binary pixel is selected from the first window and a second binary pixel is selected from the second window. The first binary pixel is bit-shifted by concatenating a plurality of trailing zero bits to the first binary pixel. A multi-bit input pixel is formed by combining the bit-shifted first binary pixel and the second binary pixel. A single multi-bit output value is generated by simultaneously applying a plurality of multi-bit input pixels to a single convolver.