Camera meta-data for content categorization

• US 6,977,679 B2
• Filed: 04/03/2001
• Issued: 12/20/2005
• Est. Priority Date: 04/03/2001
• Status: Expired due to Fees

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Tretter, Daniel R., Lin, Qian
• Primary Examiner(s): Ometz, David L.
• Assistant Examiner(s): JELINEK, BRIAN J
• Application Number:
  US09/825,739
• Publication Number:
  US 20020140843A1
• Time in Patent Office: 1,722 Days
• Field of Search: 348/231.6, 348/231.2, 348/231.99, 348/231.1, 348/231.3, 348/231.5, 707/102, 707/104.1, 707/101, 382/224, 382/225, 382/226, 382/156, 382/159, 382/227, 706/20, 703/1
• US Class Current: 348/231.2
• CPC Class Codes: G06F 16/55 Clustering; Classification
• CPC Class Codes: G06F 16/58 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually

Abstract

A method and system for categorizing non-textual subject data, such as digital images, content-based data and meta-data to determine outcomes of classification tasks. The meta-data is indicative of the operational conditions of a recording device during the capturing of the content-based data. For example, the non-textual subject data may be a digital image captured by a digital camera, and the meta-data may include automatic gain setting, film speed, shutter speed, aperture/lens index, focusing distance, date and time, and flash/no flash operation. The subject image is tagged with selected classifiers by subjecting the image to a series of classification tasks utilizing both content-based data and meta data to determine classifiers associated with the subject image.

Claims

1. A method for classifying blocks of data comprising the steps of:

capturing a block of non-textual data using a recording device for which settings for data-capture attributes are indicative of characteristics of said non-textual data;

linking meta-data with said block of non-textual data, said meta-data corresponding to at least one said data-capture attribute during said capture by said recording device;

performing automated processing to assign description to contents of said block, including utilizing said meta-data in determining said description by operations within a progression of decisional nodes, said progression of decisional nodes being configured to invoke algorithms for selectively assigning descriptions to said blocks of data; and

enabling utilization of said descriptions assigned by said operations within said progression of decisional nodes to implement searches for particular said blocks of data via query matching.

2. The method of claim 1 wherein said step of capturing includes recording at least one of an image file by an image-capture device and audio file by an audio recorder.

3. The method of claim 1 wherein said step of linking includes obtaining exposure information that identifies an exposure setting of said a recording device.

4. The method of claim 1 wherein said step of capturing further includes configuring said block as a file of non-textual data in a digital format and wherein said step of linking includes forming a tag to said file, said tag being indicative of a plurality of exposure time, automatic gain, film speed, shutter speed, white balance, aperture/lens index, focusing index, and a flash/no flash operation.

5. The method of claim 1 further including a step of transmitting said block of said non-textual data and said meta-data from said recording device to a computer for performing said automated processing.

6. The method of claim 1 wherein said automated processing includes analyzing said non-textual data and said mets-data to identify content-based information and manipulating said content-based information to derive said description.

7. The method of claim 6 wherein said step of analyzing includes applying digital signal processing (DSP) to said non-textual data.

8. A system for classifying subject data comprising:

a recording device for capturing non-textual subject data and for recording mets-data, said meta-data being specific to an operational mode of said recording device during capturing of said non-textual subject data; and

a processor configured to implement a classification technique, said classification technique including a decision tree capable of invoking algorithms that utilize both of said non-textual subject data and said meta-data for identifying at least one classifier, said classifier being representative of an attribute of said subject data, said processor being further configured to implement searches for specific said non-textual subject data via query matching to classifiers identified by said classification technique.

9. The system of claim 8 wherein said recording device is a digital camera.

10. The system of claim 8 wherein said operational mode of said recording device is based on a state as determined by at least one of exposure time, auto gain setting, film speed, shutter speed, white balance, aperture/lens index, focusing distance, and flash/no flash operation.

11. The system of claim 8 wherein said classification technique is a sequential progression of decision making comprising a plurality of classification nodes, at least some of said classification nodes including algorithms for determining which of a plurality of alterative next classification nodes is to be encountered in said sequential progression of decision making.

12. The system of claim 8 wherein said classification technique is a neural network having an input stage, an output stage and at least one decision-making stage, said decision-making stage comprising a plurality of classification nodes, at least some of said classification nodes configured to receive a plurality of weighted inputs from other classification nodes within said decision-making stage and from said input stage for generating an output as a basis for identifying classifiers.

13. A method of categorizing files of non-textual data comprising the steps of:

establishing an evaluation system for decision making, including using automated processing techniques to define a plurality of algorithms, said algorithms utilizing both of content-based data and meta-data, said content-based data corresponding to content information of a file of said non-textual data and said meta-data corresponding to data-capturing settings of a data-capturing device during capture of said file of non-textual data, said establishing including a learning procedure in which said meta-data is identified for each of a plurality of learning images, said meta-data for each said learning image being indicative of operational conditions of said data-capturing device during capture of said learning image;

capturing a file of non-textual subject data;

processing said file of non-textual subject data through said evaluation system for decision making to selectively identify a plurality of classifiers associated with said file of non-textual subject data, said evaluation system including a progression of decisional nodes configured to invoke said algorithms so as to selectively identify said plurality of classifiers; and

enabling utilization of said plurality of classifiers identified by said evaluation system for decision making to implement searches for said file via query matching.

14. The method of claim 13 further comprising a step of generating a plurality of learning classifiers that are descriptive of said learning images, said step of generating including applying content-based analysis for said content-based data and meta-data analysis for said meta-data.