Abstract: A method and apparatus for performing image quality assessment that does not require the use or availability of a reference image. The apparatus comprises a vision pre-processor, a virtual reference generator, a virtual distortion generator, and a image quality metric generator. The vision preprocessor performs a vision-like transformation upon an input image sequence to produce a vision energy map. The virtual reference generator operates upon the vision energy map to produce a virtual reference while the virtual distortion generator produces distortion estimates that are used by the virtual reference generator to produce the virtual reference. The virtual distortion and virtual reference are used by the image quality metric generator to produce a image quality metric for the input video sequence without the use of a reference video sequence.

Abstract: A method and apparatus for accurately computing parallax information as captured by imagery of a scene. The method computes the parallax information of each point in an image by computing the parallax within windows that are offset with respect to the point for which the parallax is being computed. Additionally, parallax computations are performed over multiple frames of imagery to ensure accuracy of the parallax computation and to facilitate correction of occluded imagery.

Abstract: A method and apparatus for assessing the visibility of differences between two input signal sequences, e.g., image sequences is disclosed. The apparatus comprises a perceptual metric generator having an input signal processing section, a luminance processing section, a chrominance processing section and a perceptual metric generating section.

Abstract: A signal processing apparatus and concomitant method for learning and using fidelity metric as a control mechanism and to process large quantities of fidelity metrics from a visual discrimination measure (VDM) to a manageable subjective image quality ratings. The signal processing apparatus incorporates a VDM and a neural network. The VDM receives input image sequences and generates fidelity metrics, which are received by a neural network. The neural network is trained to learn and use the fidelity metrics as a control mechanism, e.g., to control a video encoder.

Abstract: A method and apparatus for assessing the visibility of differences between two input image sequences. The apparatus comprises a visual discrimination measure having a retinal sampling section, a plurality of temporal filters and a spatial discrimination section. The retinal sampling section applies a plurality of transformations to the input image sequences for simulating the image-processing properties of human vision. The temporal filters separate the sequences of retinal images into two temporal channels producing a lowpass temporal response and a bandpass temporal response. The spatial discrimination section applies spatial processing to the temporal responses to produce an image metric which is used to assess the visibility of differences between the two input image sequences.

Abstract: An apparatus and concomitant method for decomposing an image stream into units of local contrast. A temporal contrast transform (pyramid), comprises a spatial filter, a cascade of lowpass temporal filters and a plurality of subtractors and dividers. The spatial filter is followed by a cascade of lowpass temporal filters, producing a set of temporally lowpass, spatially low pass images. Adjacent filter outputs are then subtracted, producing a set of temporally bandpass (Laplacian) images. Finally, each Laplacian image is divided by a lower level temporal lowpass image, to produce a set of temporally bandpass contrast images.

Abstract: An apparatus for assessing the visibility of differences between two input image sequences. The apparatus comprises a pair of pre-filtered processors, a pair of pyramid constructors and a JND processor. Each pre-filter processor receives a separate input image sequence and performs temporal filtering to produce a low-pass temporal response and a bandpass temporal response. In turn, the pyramid constructors receive the temporal responses and generate contrast pyramids for the temporal responses. The JND processor evaluates the contrast pyramids to produce an image metric which is used to assess the visibility of differences between the two input image sequences.

Abstract: A method and apparatus for assessing the visibility of differences between two input image sequences for improving image fidelity and visual task applications. The apparatus comprises a visual discrimination measure having a retinal sampling section, a plurality of temporal filters and a spatial discrimination section. The retinal sampling section applies a plurality of transformations to the input image sequences for simulating the image-processing properties of human vision. The temporal filters separate the sequences of retinal images into two temporal channels producing a lowpass temporal response and a bandpass temporal response. The spatial discrimination section applies spatial processing to the temporal responses to produce an image metric which is used to assess the visibility of differences between the two input image sequences.

Abstract: A spatio-temporally oriented coring method and apparatus for reducing noise in a sequence of images is disclosed. The method comprises the steps of decomposing each image into low-pass and a prefiltered sub-images, decomposing each prefiltered sub-image into a plurality of component patterns which are oriented in space, time or a combination of space and time, filtering each oriented component pattern to form a plurality of filtered patterns and reconstructing the image from the filtered patterns and the low-pass sub-image. Apparatus for reducing the noise in a sequence of images comprises means for decomposing each image into low-pass and prefiltered sub-images, means for decomposing each prefiltered sub-image into a plurality of component patterns which are oriented in space, time or a combination of space and time, means for filtering each component pattern to form a plurality of filtered patterns, and means for reconstructing the image from the filtered patterns and the low-pass sub-image.