Mitsubishi Denki Kabushiki Kaisha

Mitsubishi Electric Corporation

Sisvel Technology S.r.l.

Retinascan Limited

New Luck Global Limited

Mitsubishi Electric R&d Centre Europe B.v.

Bober, Miroslaw

Paschalakis, Stavros

Stavros Paschalakis

Miroslaw Bober

Paul Brasnett

James Cooper

Nikola Sprljan

A system and method of classifying images of pathology. An image is received, normalized, and segmented normalizing the image; into a plurality of regions; A disease vector is automatically determining for the plurality of regions with at least one classifier comprising a neural network. Each of the respective plurality of regions is automatically annotated, based on the determined disease vectors. The received image is automatically graded based on at least the annotations. The neural network is trained based on at least an expert annotation of respective regions of images, according to at least one objective classification criterion. The images may be eye images, vascular images, or funduscopic images. The disease may be a vascular disease, vasculopathy, or diabetic retinopathy, for example.

System and method for automated funduscopic image analysis

A method for transforming an image descriptor, based on a gradient histogram (h) having a plurality of histogram bins (h), into a transformed gradient histogram descriptor (v) including a set of values (v), wherein it is provided the step of applying at least one criterium for calculating the set of values (v), at least one criterium having a location criterium of the histogram bins (h). The invention also relates to an image processing device suitable for carrying out the method.

Method for transforming an image descriptor based on a gradient histogram and relative image processing apparatus

A method for encoding an image descriptor (H) based on gradient histograms (h) into a transformed descriptor (V) having transformed subdescriptors (v), wherein each gradient histogram (h) includes a plurality of histogram bins (h) and each subdescriptor (v) includes a set of values (v), wherein the subdescriptors (v) are generated according to an element utilization order list specifying a subdescriptor index (p) of the set of sub descriptors (v) and an element index (j) of the set of values (v).

Image processing apparatus and method for encoding an image descriptor based on a gradient histogram

A method and apparatus for coding and decoding the fingerprint of a multimedia item such as video or audio is disclosed. A multimedia content temporal, such as a video segment or audio segment, is described by a coarse fingerprint and a plurality of fine fingerprints, each fine fingerprint corresponding to a temporal sub-interval of said temporal interval, said temporal sub-interval typically being smaller than said temporal interval. One or more fine fingerprints are encoded in a non-predictive way, with no reference to the temporally neighboring signatures, and one or more fine fingerprints are encoded in a predictive way, from the temporally neighboring signatures. The predictive encoding entails computing the difference between neighboring fine fingerprints to make up a prediction difference matrix, scanning said prediction difference matrix into a one dimensional vector by vectorising along rows or along columns or along diagonals or along any suitable scanning pattern, and performing lossless encoding on the one dimensional vector by an appropriate method, preferably selected, at least in part, based on the scanning method used.

Multimedia signature coding and decoding

A method and apparatus for processing a first sequence of images and a second sequence of images to compare the first and second sequences is disclosed. Each of a plurality of the images in the first sequence and each of a plurality of the images in the second sequence is processed by (i) processing the image data for each of a plurality of pixel neighborhoods in the image to generate at least one respective descriptor element for each of the pixel neighborhoods, each descriptor element comprising one or more bits; and (ii) forming a plurality of words from the descriptor elements of the image such that each word comprises a unique combination of descriptor element bits. The words for the second sequence are generated from the same respective combinations of descriptor element bits as the words for the first sequence. Processing is performed to compare the first and second sequences by comparing the words generated for the plurality of images in the first sequences with the words generated for the plurality of images in the second sequence.

Video identification

A method of representing at least one image comprises deriving at least one descriptor based on color information and color interrelation information for at least one region of the image, the descriptor having at least one descriptor element, derived using values of pixels in said region, wherein at least one descriptor element for a region is derived using a non-wavelet transform. The representations may be used for image comparisons.

Methods of representing and analysing images

A method of representing and analysing images comprises producing a plurality of descriptors of an image at one or more scales and for one or more color channels, said descriptors capturing color content and interrelation information within the regions, and associating the descriptors in a plurality of ways based on their characteristics such as scale, color channel, feature semantics, and region, and comparing such representations of images to assess the similarity of images.

A method of representing and analyzing images comprises producing a plurality of descriptors of an image at one or more scales and for one ore more color channels, said descriptors capturing color content and interrelation information within the regions, and associating the descriptors in a plurality of ways based on their characteristics such as scale, color channel, feature semantics, and region, and comparing such representations of images to assess the similarity of images.

A method of classifying pixels in an image comprises combining (a) a method of identifying pixels of a first content type, wherein said method may also falsely identify pixels of a second content type as belonging to said first content type, with (b) a method of distinguishing between pixels of types including said first content type and pixels of types including said second content type, wherein method (b) comprises (c) a method for distinguishing between said first content type and content including said second content type and/or (d) a method for distinguishing between said second content type and content including said first content type.

Method and apparatus for separating content types in images

A method of detecting a region having predetermined colour characteristics in an image comprises transforming colour values of pixels in the image from a first colour space to a second colour space, using the colour values in the second colour space to determine probability values expressing a match between pixels and the predetermined colour characteristics, where the probability values range over a multiplicity of values, using said probability values to identify pixels at least approximating to said predetermined colour characteristics, grouping pixels which at least approximate to said predetermined colour characteristics, and extracting information about each group, wherein pixels are weighted according to the respective multiplicity of probability values, and the weightings are used when grouping the pixels and/or when extracting information about a group.

Growing community of inventors

Guildford, United Kingdom

Stavros Paschalakis