Image analysis – Color image processing – Pattern recognition or classification using color
Reexamination Certificate
1998-12-30
2003-04-29
Johnson, Timothy M. (Department: 2623)
Image analysis
Color image processing
Pattern recognition or classification using color
C382S199000
Reexamination Certificate
active
06556709
ABSTRACT:
BACKGROUND
1. Field
The present invention relates to segmenting objects within an image. More particularly, the present invention relates to defining an object within an image using color and contour information.
2. Background Information
There are currently three major types of image retrieval systems which permit a user to search a database for desired images. However, these systems are unable to effectively search for images which include user specified objects. Each of these three major types of retrieval systems are described in greater detail below.
The first type of image retrieval system is available through various so-called World Wide Web sites on the Internet. Images available through these systems are manually cataloged into predefined categories. This manual cataloging process entails an individual looking at each image, identifying a predominant object within the image, and cataloging the image in a predefined category which best describes the predominant object. For example, an image which shows only a tiger against a solid colored background may be cataloged in a category entitled “animals.”
Users who access this first type of image retrieval system, select from among the predefined categories and obtain images. One of the drawbacks of this retrieval system is the subjective nature of the cataloging process. One individual's impression of which object predominates an image may differ from another individual's impression. For example, an image including mountains near a sandy beach and an ocean may be placed in a category entitled “water,” or a category entitled “mountains” depending on who catalogs the image. Thus, a user's ability to retrieve this particular image is hindered because the image may be placed in any one of at least two different categories.
A second type of image retrieval system is a software-based system which prompts the user to enter a keyword which describes objects the user wishes to appear in images retrieved by the system. For example, a user may enter “water” as a keyword to obtain images which include water. However, rather than searching for images which include the keyword specified object, these systems search for images which include the predominant color of the keyword object. Thus, if the user enters “water” as a keyword, the system will retrieve all images which include blue as the predominant color. Some of these images may include water, but some may include a vast blue sky and no water.
In an effort to enhance the system's ability to obtain images closer to the user's wishes, some of these systems enable the user to specify a region of the image in which the object should appear. Thus, if a user specifies that water should be located in a lower region of the desired image, the system is less likely to use the predominant color criteria and provide the user with images which include a vast blue sky and no water.
Another drawback of this second type of retrieval system is that these systems do not persuit the user to specify a range of color for an object which should appear in the retrieved images. For example, water in an image may appear in many different shades of blue. If the system searches for only one particular shade of blue, many images which include water of a different shade will not be retrieved for the user.
The third known type of image retrieval system builds on the second type by enabling the user to specify a range of colors for objects which should appear in the retrieved images. In addition to entering a keyword to specify objects which should appear in the retrieved images, the user is able to use pointers to specify the desired intensity of each of the three primary colors (red, green, and blue). Thus, a user may use one pointer to specify the desired shade of blue water and may use another pointer to specify the desired shade of green trees in a forest. The user may also use multiple pointers to specify the desired shade of non-primary colors.
In addition to the limitations described above with respect to the second type of retrieval system, the third type of image retrieval system operates in a manner which assumes human beings may change one of the three primary colors mutually exclusive of the other two. In reality the human eye adjusts all three primary colors simultaneously. Another limitation of the third type of image retrieval system relates to the fact that although these systems are equally sensitive to each of the primary colors, the human eye is not equally sensitive to each of the primary colors. As a result of these differences in how the retrieval system and the human eye adjust/perceive color, the color intensity the user selects by adjusting one of the three pointers and the color intensity within the images retrieved by the system will not likely match.
Accordingly, there is a need for an improved method and apparatus for indexing images in a storage medium so a user may retrieve desired images.
SUMMARY
According to an embodiment of the present invention, a method for characterizing an object within an image is provided. First, the color of the object is identified. The contour of the object is then determined by identifying regions of the image where the identified color changes to another color. Then, the name of the object is associated with the determined color and contour.
REFERENCES:
patent: 5852823 (1998-12-01), De Bonet
LandOfFree
Method and method for characterizing objects within an image does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and method for characterizing objects within an image, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and method for characterizing objects within an image will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3021594