Abstract: Color fringing of edges between regions of different color in an interpolated image is effectively avoided by a signal processing technique that infers the existence of a color edge in response to prescribed changes in image characteristics at sample points for which data is available, and then interpolates intermediate values that are associated with a color edge using a linear equation. A first signal value, associated with a first band containing information of a first image characteristic (e.g. the green component) is fully sampled at a first spatial frequency, and a second signal value, associated with a second band containing information of a second characteristics (e.g. a red-green differential) is sampled at a second spatial frequency, lower than the first spatial frequency, so that interpolation of the second signal values is required. Differences between first sampled signal values, associated with sampling locations of the second sampled signal values, are measured.

Abstract: An imagery data mechanism for controllably merging separate, digitally formatted and arbitrarily shaped images eliminates overlap-edge artifacts by gradually blending a paste image (40) along its border (44) with a base image (30), regardless of the shape of the paste image. The mechanism employs a `feathering` window (50) containing a plurality of neighboring pixel locations over which the pixel values of the paste image are controllably modified to achieve a tapered blending of the two images. Whether the data value for any pixel within the display is to be modified from a given database value will depend upon whether or not that pixel location is both within the paste image and a prescribed distance to the border (44) of the paste image. If the pixel location is not even part of the image, it is effectively masked, so that not only does no feathering take place, but neither the base image nor the paste image contributes to its data value.

Abstract: Apparatus and accompanying methods for use in a film to video transfer system for achieving automatic color balancing of color images for subsequent display on a color monitor. Specifically, this apparatus first transforms each pixel in an image to be transferred from logarithmic exposure ("log exposure") RGB color values into corresponding printing density values through local averaging, then utilizes a printing density based color correction method to generate a color correction offset value for each of the RGB primary colors, and finally converts the three resulting RGB color correction offset values from the printing density domain back to the log exposure domain wherein each offset value is subsequently used to correct the corresponding red, green and blue color values associated with each of the pixels that collectively forms a subsequent scan of the same image prior to the conversion of each of the color balanced RGB pixel values into analog video form.

Abstract: A method of performing interactive image processing of a digital image with a digital computer by:1. forming a sequential order operations file of each operation that is to be performed on a digital image;2. determining which pixels of the digital image must be computed (operated) upon in order to obtain a desired output image by processing through the operations log in a reverse sequential order;3. ordering the computation of the determined pixels in a way which minimizes the amount of computer memory needed by forming chunks of dependent pixels and by calling from memory only those chunks which are needed for the final result and by deleting used chunks when they are no longer needed; and4. processing through the operations file in sequential order.

Abstract: A signal processing method and apparatus for processing a sampled color image signal of the type having luminance values and chrominance values representing a highly sampled luminance component and less highly sampled chrominance components to produce interpolated chrominance values between sampled chrominance values is characterized by producing hue values at neighboring chrominance component sample locations as a function of a luminance value and the chrominance value at the neighboring locations; producing a signal representing an interpolated hue value as a function of neighboring hue values; and producing a signal representing an interpolated chrominance value as a function of the interpolated hue value and a luminance value at the interpolated location. The signal processing method reduces color fringing in an image reproduced from the sampled image signal without introducing unwanted hue shifts.

Abstract: A signal processing method and apparatus for providing interpolated values between sampled values in a sampled image signal, is characterized by providing a plurality of different interpolation routines for producing interpolated signal values for appropriately completing a respective plurality of known features, detecting which of the geometrical image features is represented by a neighborhood of sample values and applying the interpolation routine appropriate for completing the detected feature, to produce the interpolated signal value. The signal processing method and apparatus has the advantage that reconstruction errors in the reproduced image are reduced. In one mode of the invention for processing images to be viewed by humans, reconstruction errors are forced to occur in areas of the image composed of natural textures where the errors are not readily visible to the human observer. The appearance of the reconstructed image is thereby improved.

Abstract: An electronic color camera having a single-chip solid state color image sensor, includes a color dependent birefringent spatial filter that deflects red and blue light from portions of an image sampled by the neighboring green sensitive image sensing elements onto red and blue sensitive image sensing elements. Signal processing electronics produces interpolated red and blue signal values by forming red and blue hue component values at the red and blue sampling locations, interpolating the hue component values, and producing the interpolated red and blue values as a function of the interpolated hue component values, and green signal values at the interpolation locations. As a result, color fringes at monochrome edges are completely eliminated, and are substantially reduced at colored edges.