Detection of microaneurysms
AssigneeUniversity of Surrey
InventorsHongying Lilian Tang, Su Wang
Life of this patent
Estimating the term…
Abstract
An image analysis method and an image processing apparatus are disclosed. The image analysis method comprises receiving a retinal image and identifying at least one candidate object from a plurality of objects within the retinal image. Singular spectrum analysis SSA is performed on the at least one candidate object, to obtain an intensity profile along at least one cross-sectional line through the candidate object. At least one feature from the intensity profile is extracted for classification of the at least one candidate object.
Claims20
An image analysis method, comprising:
- receiving a retinal image;
- identifying at least one candidate object from a plurality of objects within the retinal image;
- for each identified candidate object, obtaining a plurality of cross-sectional profiles along a plurality of cross-sectional lines through the candidate object;
- performing singular spectrum analysis SSA on the plurality of cross-sectional profiles for the candidate object, by using singular value decomposition SVD to decompose the plurality of cross-sectional profiles into a set of principal components and reconstructing a plurality of intensity profiles based on the principal components;
- and
- extracting at least one feature from the reconstructed intensity profiles for classification.
An image processing apparatus comprising:
- a device for receiving a retinal image;
- and
- a processor configured to:
- identify at least one candidate object from a plurality of objects within the retinal image;
- for each identified candidate object, obtain a plurality of cross-sectional profiles along a plurality of cross-sectional lines through the candidate object;
- perform singular spectrum analysis SSA on the plurality of cross-sectional profiles for the candidate object, by using singular value decomposition SVD to decompose the plurality of cross-sectional profiles into a set of principal components and reconstructing a plurality of intensity profiles based on the principal components;
- and
- extract at least one feature from the reconstructed intensity profiles for classification.
An image processing apparatus comprising:
- an input configured to receive a retinal image;
- a candidate object analysis unit configured to identify at least one candidate object from a plurality of objects within the retinal image, obtain a plurality of cross-sectional profiles along a plurality of cross-sectional lines through the candidate object, for each identified candidate object, and perform singular spectrum analysis SSA on the plurality of cross-sectional profiles for the candidate object, by using singular value decomposition SVD to decompose the plurality of cross-sectional profiles into a set of principal components and reconstructing a plurality of intensity profiles based on the principal components;
- and
- a feature extraction unit configured to extract at least one feature from the reconstructed intensity profiles for classification.
Dependencies are inferred from the claim text.