Katherine J Wang

US Patent:

58023615, Sep 1, 1998

Filed:

Sep 30, 1994

Appl. No.:

8/315965

Inventors:

Katherine Wang - San Jose CA
James Normile - Woodside CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

G06F 314
G06F 3147

US Classification:

395600

Abstract:

A system and method for searching images includes a low level analyzer for processing image data in a plurality of images into side information files, the image data of each side information file pertaining to a particular statistical or perceptual abstraction of the images, such as motion, color, texture, segments, and the like. A user interface allows a user to graphically construct a search inquiry with icons representing image attributes corresponding to the image data in the side information files. A search inquiry may include complex boolean expressions, and custom image attributes may be created to represent useful search inquiries. The image attributes of a search inquiry may also be ranked. A high level analyzer parses a search inquiry to determine the image attributes contained therein, and analyzes the image data of the side information files to determine a similarity measure for each image with respect to the search inquiry. The high level analyzer orders retrieved images according to their similarity measures, and provides them to a image display controller for display.

US Patent:

55966594, Jan 21, 1997

Filed:

Sep 1, 1992

Appl. No.:

7/938942

Inventors:

James O. Normile - Sunnyvale CA
Katherine S. Wang - San Jose CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

G06K 936
G06K 938
G06K 946

US Classification:

382253

Abstract:

Improved proprecessing and postprocessing for vector quantization, for example, for encoding an image. In one embodiment, the luminosity of the vectors in an image are used to determine the codes for vector quantization. In another embodiment, a median filter is performed to eliminate motion artifacts. In another embodiment, temporal filtering is applied once the difference between an earlier frame and a current frame exceeds a threshold. Embodiments are also provided for adaptive temporal filtering based on temporal "no change" blocks and their errors. Embodiments are provided for different regions of images which reference different codebooks, and regions of variable size. Embodiments are also provided for shared and variable size codebooks for different images or other data.

US Patent:

56490305, Jul 15, 1997

Filed:

Mar 15, 1995

Appl. No.:

8/405448

Inventors:

James Oliver Normile - Sunnyvale CA
Katherine Shu-Wei Wang - San Jose CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

H04N 712

US Classification:

382253

Abstract:

Improved method and apparatus for vector quantization (VQ) to build a codebook for the compression of data. The codebook or "tree" is initialized by establishing N initial nodes and creating the remainder of the codebook as a binary codebook. Children entries are split upon determination of various attributes, such as maximum distortion, population, etc. Vectors obtained from the data are associated with the children nodes, and then representative children entries are recalculated. This splitting/reassociation continues iteratively until a difference in error associated with the previous children and current children becomes less than a threshold. This splitting and reassociating process continues until the maximum number of terminal nodes is created in the tree, a total error or distortion threshold has been reached or some other criterion. The data may then be transmitted as a compressed bitstream comprising a codebook and indices referencing the codebook.

US Patent:

58224653, Oct 13, 1998

Filed:

Jun 7, 1995

Appl. No.:

8/486828

Inventors:

James Oliver Normile - Sunnyvale CA
Katherine Shu-Wei Wang - San Jose CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

G06K 900

US Classification:

382253

Abstract:

Improved method and apparatus for vector quantization (VQ) to build a codebook for the compression of data. The codebook or "tree" is initialized by establishing N initial nodes and creating the remainder of the codebook as a binary codebook. Children entries are split upon determination of various attributes, such as maximum distortion, population, etc. Vectors obtained from the data are associated with the children nodes, and then representative children entries are recalculated. This splitting/reassociation continues iteratively until a difference in error associated with the previous children and current children becomes less than a threshold. This splitting and reassociating process continues until the maximum number of terminal nodes is created in the tree, a total error or distortion threshold has been reached or some other criterion. The data may then be transmitted as a compressed bitstream comprising a codebook and indices referencing the codebook.

US Patent:

58057331, Sep 8, 1998

Filed:

Dec 12, 1994

Appl. No.:

8/354367

Inventors:

Katherine Wang - San Jose CA
James Normile - Woodside CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

G06K 936
G06K 946
G06K 1500
G06F 1500

US Classification:

382232

Abstract:

A method and system summarizes scenes in a video sequence by detecting scene changes, and then comparing scenes in a moving window to determine their similarity. Similar scenes are consolidated and represented by a representative frame, a number of which are displayed to a user. Scene changes are detected by comparing average color histograms for each scene, motion compensated pixel differences or motion compensated edge maps, or a combination of these methods. Scenes in the video sequence are selected for summarizing according to their normalized time duration. Of the selected scenes, similar or related scenes are determined by comparing the average color histograms of each pair of scenes in a moving window, using a standard population error measure, such as a Chi-squared test. For each set of related scenes, a representative frame is taken, either as the medial frame from the entire time duration of the related scenes or as the first frame of the medial scene in the set. The representative frames are displayed to the user, wherein set of related scenes are retrived by selecting the representative frame for the set.

US Patent:

56232624, Apr 22, 1997

Filed:

Aug 17, 1994

Appl. No.:

8/291655

Inventors:

James O. Normile - Woodside CA
Katherine Shu-wei Wang - San Jose CA
Ke-Chiang Chu - Saratoga CA
Dulce B. Ponceleon - Palo Alto CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

H03M 740

US Classification:

341 67

Abstract:

Decoding and encoding of variable length data words and data strings is accelerated by testing for and processing more than one word or string per encoding or decoding cycle. In an encoding scheme wherein fixed length data words are encoded into variable length data strings, decoding is carried out by first receiving a data stream having a plurality of encoded data strings contained therein, and then testing at least a portion of the data stream to determine whether the portion contains one of a number of selected sets of multiple data strings. If the portion of the data stream contains one of the selected sets of multiple data strings, the multiple data strings are decoded into a corresponding set of multiple data words. This decoding procedure allows a plurality of encoded data strings to be decoded in a single decoding cycle. The procedure may be implemented using either a single lookup table or a set of split-level lookup tables.

US Patent:

58728651, Feb 16, 1999

Filed:

Feb 8, 1995

Appl. No.:

8/386308

Inventors:

James Normile - Woodside CA
Katherine Wang - San Jose CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

G06K 962

US Classification:

382224

Abstract:

A computer system and computer implemented method automatically classify video sequences into categories. A set of categories is defined either manually through the association of selected video sequences with user supplied category designations, or automatically through segregation of a set of video sequences into groups of similar sequences. Input video sequences are then classified by either pixel decomposition or primitive attribute decomposition; the former analyzing each image on a pixel basis, the latter employing extracted image information. Categories can be trained as new video sequences are input into the system, or new categories can be created to accommodate such new sequences that are dissimilar from existing categories.

US Patent:

57199612, Feb 17, 1998

Filed:

Mar 29, 1996

Appl. No.:

8/625608

Inventors:

James Oliver Normile - Woodside CA
Katherine Shu-wei Wang - San Jose CA
Ke-Chiang Chu - Saratoga CA
Dulce Beatriz Ponceleon - Palo Alto CA

Assignee:

Apple Computer, Inc. - Cupertino CA

International Classification:

G06K 936

US Classification:

382239

Abstract:

A signal processing system determines the characteristic of a signal for encoding or decoding by examining and classifying such signal, and then applies a transformation or inverse transformation to such signal. Depending on classification of the signal, various transforms or inverse transforms are applicable adaptively thereto.