Method and apparatus for image retrieval
OwnerLida Huanginventor
Life of this patent
Estimating the term…
Abstract
An efficient image retrieval method utilizing topological order distance (TOD) algorithm is disclosed. In one aspect, an image recognition method may include steps of receiving one or more input images; extracting one or more key features on the input images; determining an origin and one or more feature points on the key features on the input images; establishing a topological relationship in each dimension between the feature points on the key features; and utilizing the topological relationship in each dimension between the feature points to identify the input images in an image database. Comparing with the conventional image retrieval techniques, the TOD algorithm is advantageous because it is more efficient, scale and rotation invariant, and tolerant to affine and perspective transformation.
Claims18
An image recognition method, comprising steps of:
- receiving one or more input images with at least two dimensions;
- extracting two or more key features on the input images;
- determining an origin and two or more feature points on the key features on the input images;
- selecting a coordinate system to determine a position of each feature point in the input images, wherein the position of each feature point is determined by at least two parameters according to the dimensions of the input images;
- establishing a topological relationship in each dimension between the feature points on the key features;
- and
- utilizing the topological relationship in each dimension between the feature points to identify the input images in an image database,
- wherein the step of establishing a topological relationship in each dimension between the feature points on the key features comprises steps of calculating and sorting topological distances between each feature points and the origin to produce one or more sequences according to the dimension of the image, and the sequence at least includes a first sequence related to a first parameter between the origin and each feature points, and a second sequence related to a second parameter between the origin and each feature points in the input image.
An image retrieval system comprising:
- means for receiving one or more input images with at least two dimensions;
- a key feature extracting unit including a feature point generator that generates an origin and at least two feature points;
- and
- an image processor executing instructions to perform:
- selecting a coordinate system to determine a position of each feature point in the input images, wherein the position of each feature point is determined by at least two parameters according to the dimensions of the input images;
- establishing a topological relationship in each dimension between the feature points on the key features;
- and
- utilizing the topological relationship in each dimension between the feature points to identify the input images in an image database,
- wherein the image processor includes a sequence generator configured to measure topological distances between each feature points and the origin, and configured to produce a sequence based on the topological distances therebetween, and the sequence at least includes a first sequence related to a first parameter between the origin and each feature points, and a second sequence related to a second parameter between the origin and each feature points in the input image.
Dependencies are inferred from the claim text.