Eastman-kodak Company

Bryce E Bayer

Peter L Dillon

Ronald S Cok

Billy E Cates

A unit transformation circuit transforms three discrete input signals into a set of transform coefficient signals characteristic of a 'collapsed' Walsh-Hadamard transform. The unit transformation circuit includes two tiers of arithmetic networks. In the first tier, a pair of arithmetic networks generates (A) first sum and difference signals from the first and second input signals and (B) second sum and difference signals from the second and third input signals. Arithmetic networks in the second tier generate a set of coefficient signals from (A) the sum of the first and second sum signals (B) the sum of the first and second difference signals and (C) the difference between the first and second difference signals. The unit transformation circuit forms a fundamental circuit element from which more complex circuits are constructed capable of transforming larger numbers of discrete input signals.

Transformation circuit for implementing a collapsed Walsh-Hadamard

An improved image processing method prevents unwanted processing artifacts from degrading the reproduction of an image by stressing the interrelationship of various image gradients in the image. Image signals are generated representative of the light value of elements of the image. A local gradient signal is generated in response to a combination of image signals representative of an image gradient over a local portion of the image. An extended gradient signal is also generated representative of an image gradient over an image portion greater than the local portion. The image gradients are interrelated in a composite signal generated from a combination of the local and extended gradient signals. As an example of the combination, the composite signal is made to vary as a function of the difference between the gradient signals. The image signals are then modified subject to a characteristic of the composite signal, e.g., the magnitude of the composite signal.

Image processing method based on processing of interrelated image

An improved image processing method reduces noise in a sampled image while minimizing unintended distortion of image features. Image signals are generated representative of the light value of elements of the image. These signals are formed into signal arrays aligned to blocks of image elements. The signal arrays are transformed by a set of 4 by 4 Walsh-Hadamard functions into a corresponding set of coefficient signals. Certain of these coefficient signals represent the difference between the light value of each image element and an average light value over an image region smaller than the block being transformed. By modifying--i.e., coring or clipping--and inverting only these selected coefficient signals, artifacts related to the introduction of 'false' edge-like structure are reduced in the reconstructed image. In addition, in a multi-stage processing method, the excluded coefficient signals may represent the input signals to the next stage.

Transform processing method for reducing noise in an image

An improved image processing method uses a modified Walsh-Hadamard transform to remove noise and preserve image structure in a sampled image. Image signals representative of the light value of elements of the image are grouped into signal arrays corresponding to blocks of image elements. These signals are mapped into larger signal arrays such that one or more image signals appear two or more times in each larger array. The larger arrays are transformed by Walsh-Hadamard combinations characteristic of the larger array into sets of coefficient signals. Noise is reduced by modifying--i.e., coring or clipping--and inverting selected coefficient signals so as to recover processed signals--less noise--representative of each smaller signal array. The results exhibit acceptable rendition of low contrast detail while at the same time reducing certain processing artifacts characteristic of the unimproved Walsh-Hadamard block transform.

Image processing method using a collapsed Walsh-Hadamard transform

Processing circuitry for discrete-sample-type-color-video signals performs an interpolation along a scan row or line to define intermediate signal levels between signal 'updates' for the individual primary colors. In a preferred implementation, using green, red, and blue as primary colors, green samples occur more frequently than red or blue and interpolated green samples are specially combined to produce a 'slow' green signal that is matched to the frequency ranges of the red and blue signals. The difference between the full green signal and the slow green signal is then used for producing a signal to represent high frequency luminance detail.

Signal processing for discrete-sample-type-color-video signal

Processing circuitry for discrete-sample-type-color-video signals provides signal enhancement by algebraically summing properly weighted delayed, doubly delayed, and undelayed versions of the same signal.

A sensing array for color imaging includes individual luminance- and chrominance-sensitive elements that are so intermixed that each type of element (i.e., according to sensitivity characteristics) occurs in a repeated pattern with luminance elements dominating the array. Preferably, luminance elements occur at every other element position to provide a relatively high frequency sampling pattern which is uniform in two perpendicular directions (e.g., horizontal and vertical). The chrominance patterns are interlaid therewith and fill the remaining element positions to provide relatively lower frequencies of sampling. In a presently preferred implementation, a mosaic of selectively transmissive filters is superposed in registration with a solid state imaging array having a broad range of light sensitivity, the distribution of filter types in the mosaic being in accordance with the above-described patterns.

Growing community of inventors

Rochester, NY, United States of America

Bryce E Bayer