Compression method for a standardized image library

• US 5,574,573 A
• Filed: 10/29/1993
• Issued: 11/12/1996
• Est. Priority Date: 10/29/1993
• Status: Expired due to Term

• Current Assignee: Eastman Kodak Company
• Original Assignee: Eastman Kodak Company
• Inventors: Ellson, Richard N., Ray, Lawrence A.
• Primary Examiner(s): Vu, Kim Yen
• Assistant Examiner(s): Williams, Kimberly A.
• Application Number:
  US08/145,284
• Time in Patent Office: 1,110 Days
• Field of Search: 358/403, 358/166, 358/452, 358/453, 358/426, 358/447, 358/448, 358/454, 358/455, 358/456, 358/261.1, 358/261.2, 358/261.3, 382/44, 382/45, 382/50, 382/54, 382/2, 382/115, 382/116, 382/117, 382/118, 382/201, 348/77, 348/422, 348/429, 348/649, 348/652
• US Class Current: 358/452
• CPC Class Codes: G06T 9/008 Vector quantisation
• CPC Class Codes: G07C 9/25 using biometric data, e.g. fingerprints, iris scans or voice recognition

Abstract

A technique involving the standardization of the subject matter of an image and the compression of the image using features of the standardization can provide for a high compression ratio. This technique can be used to improve the quality of a restored image for a predetermined amount of storage space.

Claims

1. A method of increasing the compression ratio for a library of facial images, the method comprising the steps of:

standardizing spatial features and characteristics, including the steps of;

selecting image features in an order of importance, and

positioning eyes and facial orientation in each image in a standard locational configuration,

identifying regions of increased informational significance of the images;

compressing each image; and

providing an increased data density in each compressed image responsive to the standardizing and identifying including the step of providing compressed data in a region of a face having a higher compression ratio than non-facial regions.

2. A method of increasing the compression ratio for a library of facial images, the method comprising the steps of:

automatically standardizing spatial features and characteristics and identifying regions of increased informational significance of the images; and

automatically compressing each image and providing an increased data density in each compressed image responsive to the standardizing and identifying;

wherein said standardizing step includes the steps of;

selecting image features in an order of importance;

providing set of images from said library which enhances the selected features;

determining image feature constraints for said selected features;

automatically processing said image to meet said image feature constraints, wherein said image feature constraints comprising compression and image quality requirements; and

positioning eyes and facial orientation in each image in a standard locational configuration; and

wherein said providing step includes the step of;

providing compressed data in a region of a face having a higher compression ratio than non-facial regions.

3. A method for providing a higher compression ratio for each image in a library of images, said method comprising the steps of:

automatically spatially standardizing the images forming said image library to provide spatial correlations between features of the standardized images;

automatically compressing said standardized images using the spatial correlations between said standardized images; and

automatically decompressing said compressed standardized image to provide a visual image;

wherein said standardizing step includes the steps of;

identifying image features in an order of importance;

processing images to enhance said image features;

identifying constraints of said image features;

processing said image to meet the constraints; and

automatically assigning image regions based on a presence of a desired image quality and a desired image compression;

.

wherein said processing images step includes the steps of;

processing said images using vector quantization; and

providing a plurality of codebooks, each codebook comprising representative images corresponding to said vector quantization, wherein each image region has a selected codebook for processing each image region; and

wherein said decompressing step further include the step of decompressing each image region using said selected codebook which used in said compressing step.

4. The method of claim 3 wherein said images are facial images, said standardizing step including a step of positioning eyes and orienting facial features in a predetermined manner for each image.

5. The method of claim 4, wherein said compressing step includes the step of varying a number of codebook representative images as a function of spatial position.

6. The method of claim 5, wherein said compressing step includes the step of providing reduced number of codebook representative images in the peripheral region of said image.

7. A method of automatically compressing images, comprising:

standardizing the spatial location of features in the images;

assigning an image quality level to regions comprising the spatial location of the features in the images;

determining image-to-image correlations between the regions of the images and reflection transforming the images;

allocating a storage capacity for each region of the images;

constructing a codebook for each region of the images;

partitioning each image into the regions;

determining an entry in each codebook closest to a content of each region; and

storing the entry as a compressed image.