Image classification based on a mixture of elliptical color models
First Claim
1. A method of classifying images, comprising:
- a. generating one or more color models from one or more first images;
b. selecting one or more optimum color models from the one or more color models, wherein the optimum color models are representative of color in the one or more first images; and
c. comparing one or more color distributions from one or more second images with the one or more optimum color models.
1 Assignment
0 Petitions
Accused Products
Abstract
A method of classifying images based on elliptical color models is utilized in a number of applications. One or more color models are generated from a set of images with a region of interest. Then, sets of images are utilized for training. One set of images has regions of interest, and the other set of images is without regions of interest. By utilizing the two sets of images, a maximum difference between the sets is achieved, so that a color model is most representative of the object desired. Then using the optimal color model, a collection of images are able to be searched, and images are retrieved based on the probability that the images contain the desired object.
39 Citations
38 Claims
-
1. A method of classifying images, comprising:
-
a. generating one or more color models from one or more first images;
b. selecting one or more optimum color models from the one or more color models, wherein the optimum color models are representative of color in the one or more first images; and
c. comparing one or more color distributions from one or more second images with the one or more optimum color models. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11)
-
-
12. A method of classifying images, comprising:
-
a. generating one or more color models from a first set of images with a region of interest;
b. training the one or more color models utilizing a second set of images with one or more regions of interest and a third set of images without regions of interest;
c. comparing the one or more color models with one or more color distributions from a fourth set of images; and
d. retrieving one or more images from the fourth set of images based on the comparison between the one or more color models and the one or more color distributions. - View Dependent Claims (13, 14, 15, 16, 17, 18, 19, 20)
-
-
21. A method of optimizing color models for classifying images, comprising:
-
a. generating a color model for each of one or more first images;
b. searching for the color model for maximizing a statistical distance between the one or more first images and one or more second images;
c. updating a color model set by adding the color model for maximizing the statistical distance to the color model set; and
d. repeating searching for the color model and updating a color model until the statistical distance is maximized. - View Dependent Claims (22, 23, 24, 25)
-
-
26. A system for comparing a plurality of images, comprising:
-
a. one or more first images;
b. one or more color models generated from the one or more first images;
c. one or more second images with one or more regions of interest for training the one or more color models;
d. one or more third images without regions of interest for training the one or more color models;
e. one or more fourth images;
f. one or more color distributions generated from the one or more fourth images; and
g. a program to compare the one or more color models with the one or more color distributions. - View Dependent Claims (27, 28, 29, 30, 31, 32, 33, 34)
-
-
35. A capture and display device, comprising:
-
a. a receiving unit for receiving image data;
b. a display unit coupled to the receiving unit for displaying image data; and
c. a program coupled to the receiving unit and the display unit to compare the image data by i. generating one or more color models from one or more selected images;
ii. selecting one or more optimum color models from the one or more color models, wherein the one or more optimum color models are representative of color in the one or more selected images; and
iii. comparing one or more color distributions from the image data with the one or more optimum color models. - View Dependent Claims (36, 37, 38)
-
Specification