Image analysis – Color image processing – Pattern recognition or classification using color
Reexamination Certificate
2001-05-14
2003-06-17
Tran, Phuoc (Department: 2621)
Image analysis
Color image processing
Pattern recognition or classification using color
C382S228000
Reexamination Certificate
active
06580824
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to digital image database and classification systems, and in particular to classification of sepia tone photographs that have been converted to digital formats.
2. Background Information
Automatic image classification has many important applications, particularly in classifying a special class of images: black and white photos with sepia tones. Some of these images have been scanned from antique photos, which were originally in black and white. Over the years they gradually turn to yellow or brown due to the chemical reactions of the film paper. Still, many other sepia-tone images were specially produced either chemically or digitally to generate an antique appearance.
Classification of photos with sepia tones is useful in several applications. Because of the special color tones, many general-purpose color image processing techniques would not work well on these images. For example, a face-detection algorithm in an imaging system that detects faces purely based on the skin color is unlikely to generate the correct results on these photos because the skin color has been changed. By detecting images with sepia tones, the system can handle these images appropriately to avoid possible errors. For instance, an alternative algorithm can be used to detect the face that uses features that are invariant to color changes, such as eye corners. Therefore, the detection accuracy of the image recognition system may be significantly improved.
Sepia-toned images can also cause problems in automatic color balance or enhancement processes. Because the colors are narrowly concentrated in the sepia tones, the images could be mistakenly considered as color imbalanced. Color corrections to these images could cause undesirable color artifacts. If a color imaging process knows beforehand what kind of image is being processed, it can react appropriately and achieve better results.
Another application of sepia-tone image classification is in image indexing. Large image databases or collections require good indexing mechanisms so that images can be categorized effectively, browsed efficiently, and retrieved quickly. Conventional systems store and retrieve specific information from the database using, for example, descriptive information regarding the image file, such as file creation date, file name, file extension and the like. This form of image classification is not significantly different from the classification of any other digital information.
By relying on the file information, only cursory information can be obtained about the file and nothing at all specifically related to the image. For example, an image file could have a name that had no relation to the type or content of the image, such as a black and white image could have the file name “color_image”. Other systems provide classification based on the content of the images, such as flowers, dogs, and the like. In practice, this is usually done by keyword annotation, which is a laborious task.
Image classification techniques have been proposed in the past years that are designed for use in image databases. However, none of these prior techniques have addressed identifying sepia-toned images. Examples of these prior techniques include the following articles. S.F. Chang, W. Chen and H. Sundaram, “Semantic visual templates: linking visual features to semantics,”
Proc. of IEEE Intl. Conf. on Image Processing
, vol. 3, p. 531-35, 1998. S. Paek and S. -F. Chang, “A knowledge engineering approach for image classification based on probabilistic reasoning systems,”
Proc. of Intl. Conf. On Multimedia
&
Expo
, 2000. R. Qian, N. Haering and I. Sezan, “A computational approach to semantic event detection,”
Proc. of IEEE Conf. Computer Vision and Pattern Recognition
, vol. 1, p.200-06, 1999. M. Szummer and R. W. Picard, “Indoor-outdoor image classification,”
Proc. of IEEE Workshop on Content
-
Based Access of Image and Video Libraries
, p.42-51, 1998. A. Vailaya, M. Figueiredo, A. Jain and H.-J. Zhang, “A Bayesian framework for semantic classification of outdoor vacation images,”
Proc. of SPIE: Storage and Retrieval for Image and Video databases VII
, vol. 3656, p. 415-26, 1999. A. Vailaya and A. Jain, “Detecting sky and vegetation in outdoor images,”
Proc. of SPIE: Storage and Retrieval for Image and Video databases
2000, vol. 3972, p. 411-20, 2000. N. Vasconcelos and A. Lippman, “A Bayesian framework for semantic content characterization,”
Proc. of IEEE Conf. on Computer Vision and Pattern Recognition
, p. 566-71, 1999.
Image databases frequently contain images having different characteristics, such as color images of different color resolutions (e.g., 16, 256, 16 bit, and 24 bit color), gray scale images, black and white images, sepia-toned images, and the like. Prior image classification techniques do not address automatic sepia-toned image classification. Since sepia-toned images can cause difficulties when used with conventional image processing techniques and many sepia tone images are valuable antique photos, it would be desirable to have an image classification system that can analyze the properties of the images themselves and classify the images according to whether or not the image is a sepia-toned image.
SUMMARY OF THE INVENTION
The present invention is directed to methods and systems that classify digital images containing sepia-tones. An exemplary method comprises converting values of a first color space of a digital image to hue saturation intensity (HSV) values; removing any pixel of the digital image below at least one of a saturation threshold and an intensity threshold; analyzing, after the step of removing, remaining pixels of the digital image; and classifying the digital image as a sepia-toned image based on the analysis of the remaining pixels of the digital image.
Alternate embodiments provide for estimating the probability that each of the remaining pixels is sepia-toned based on a predetermined color distribution of sepia-toned images and determining a probability that the digital image is a sepia-toned image.
An exemplary method of training a system to detect sepia-toned images comprises converting values of a first color space of a plurality of training images to HSV values, wherein the training images are sepia-toned images; removing any pixel of the training images below at least one of a saturation threshold and an intensity threshold for each of the plurality of training images; and estimating, after the step of removing, the color distribution of HSV values for remaining pixels over the plurality of training images.
REFERENCES:
patent: 5253302 (1993-10-01), Massen
patent: 5341226 (1994-08-01), Shiau
patent: 5345320 (1994-09-01), Hirota
patent: 6282330 (2001-08-01), Yokota et al.
patent: 6477269 (2002-11-01), Brechner
patent: 0935216 (1999-08-01), None
S.-F., Chang, W. Chen and H. Sundaram, “Semantic Visual Templates; Linking Visual Features to Semantics,” Proc. of IEEE Intl. Conf. on Image Processing, vol. 3, p. 531-35, 1998.
S. Paek and S.-F. Chang, “A Knowledge Engineering Approach for Image Classification Based on Probabilistic Reasoning Systems”, Proc. of Intl. Conf. on Multimedia & Expo, 2000.
R. Qian, N. Haering and I. Sezan, “A Computational Approach to Semantic Event Detecation,” Proc. of IEEE Conf. Computer Vision and Pattern Recognition,” vol. 1, p. 200-06, 1999.
M. Szummer and R.W. Picard, “Indoor-outdoor image classification,” Proc. of IEEE Workshopon Content-Based Access of Image and Video Libraries, p. 42-51, 1998.
A. Vailaya, M. Figueiredo, A, Jain and H.-J. Zhang, “A Bayesian framework for semantic classification of outdoor vacation images,” Proc. of SPIE: Storage and Retrieval for Image and Video databases VII, vol. 3656, p. 415-26, 1999.
A. Vailaya and A. Jain, “Detecting sky and vegetation in outdoor images,” Proc. of SPIE: Storage and Retrieval for Image and Video databases 2000, vol. 3972, p. 411-20, 2000.
N. Vasconcelos and A. Lippman, “A BAyesian framework for semantic content characterization,” Proc. of
Deng Yining
Tretter Daniel
LandOfFree
Classification of photos with sepia tones does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Classification of photos with sepia tones, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Classification of photos with sepia tones will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3138447