Method for finding representative vectors in a class of vector spaces
AssigneeSasken Communication Technologies Limited
InventorGomathi Sankar
Life of this patent
Estimating the term…
Abstract
The present invention provides a technique for retrieving pictures from a large database that is less complex and uses significantly less memory and computational resources than current techniques. This is accomplished by determining representative data vectors based on a tolerance distance that represents data vectors in a given vector space that defines the picture to extract features of the picture that facilitates in retrieving pictures.
Claims49
A method of extracting features in a picture for picture processing, comprising:
- determining representative data vectors based on a tolerance distance that represents data vectors in a class of vector spaces that define the picture to extract representative features of the picture for picture processing, wherein extracting the representative features further comprises
- computing the tolerance distance that represents the data vectors in the class of vector spaces of the picture;
- forming a histogram of the data vectors based on a predetermined number of feature values;
- forming a window based on the tolerance distance;
- computing a sum of data vectors using the data vectors falling within the window for each of the feature values as the window indexes to a next subsequent feature value in the formed histogram;
- comparing each computed sum of data vectors to a threshold value;
- and
- determining the representative data vectors to extract dominant features in the picture based on the outcome of the comparison.
A computer-readable medium having computer-executable instructions for performing a method to extract dominant colors in a picture for picture processing, comprising:
- computing a tolerance distance that represents data vectors in a class of vector spaces of the picture;
- forming a histogram of the data vectors based on a predetermined number of color values;
- computing a sum of data vectors using the data vectors falling within a window based on the tolerance distance for each color value as the window indexes to a subsequent color value in the formed histogram;
- comparing each computed sum of data vectors to a threshold value;
- and
- determining representative data vectors to extract dominant colors in the picture based on the outcome of the comparison.
A computer system to extract dominant colors in a picture for picture processing, comprising:
- a processor;
- an output device;
- and
- a storage device to store instructions that are executable by the processor to perform a method, the method comprising:
- computing a tolerance distance that represents data vectors in a class of vector spaces of the picture;
- forming a histogram of the data vectors based on a predetermined number of color values;
- computing a sum of data vectors using the data vectors falling within a window based on the tolerance distance for each color value as the window indexes to a subsequent color value in the formed histogram;
- comparing each computed sum of data vectors to a threshold value;
- and
- determining representative data vectors to extract dominant colors in the picture based on the outcome of the comparison.
A feature extraction system to extract dominant colors in a picture for picture processing, comprising:
- an input module to receive data vectors in a class of vector spaces of the picture;
- a feature extraction module coupled to the input module, wherein the feature extraction module comprises:
- an analyzer to receive the data vectors from the input module and compute a tolerance distance that represents the data vectors in the class of vector spaces in the picture, wherein the analyzer forms a histogram of the data vectors based on a predetermined number of color values, wherein the analyzer computes a sum of data vectors using the data vectors falling within a window based on the tolerance distance for each color value as the window indexes to a next color value in the formed histogram;
- a comparator to receive the computed sum of data vectors and compare the received sum of data vectors to a threshold value, wherein the analyzer determines representative data vectors to extract dominant colors in the picture based on the outcome of the comparison;
- and
- an output module, to extract and output the dominant colors in the picture using the determined representative data vectors.
A method of extracting features in a picture for picture processing, comprising:
- determining a class of feature representative data vectors in a class of vector spaces corresponding to the picture;
- computing a tolerance distance that represents the data vectors in the class of vector spaces that define the picture;
- grouping vectors within the tolerance distance;
- computing a sum of data vectors using the data vectors within the tolerance distance for each of the vector values while indexing the grouped data vectors to subsequent vector values;
- and
- comparing each computed sum of data vectors to a threshold value to extract dominant features in the image.
Dependencies are inferred from the claim text.