Method for identifying objects and features in an image

• US 5,640,468 A
• Filed: 04/28/1994
• Issued: 06/17/1997
• Est. Priority Date: 04/28/1994
• Status: Expired due to Term

• Current Assignee: Shin-Yi Hsu
• Original Assignee: Shin-Yi Hsu
• Inventors: Hsu, Shin-yi
• Primary Examiner(s): Couso, Jose L.
• Application Number:
  US08/234,767
• Time in Patent Office: 1,146 Days
• Field of Search: 382/103, 382/181, 382/190, 382/199, 382/240, 382/263, 382/284, 382/302
• US Class Current: 382/190
• CPC Class Codes: G06T 2207/10028 Range image; Depth image; 3D point clouds
• CPC Class Codes: G06T 2207/10036 Multispectral image; Hyperspectral image
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/20084 Artificial neural networks [ANN]
• CPC Class Codes: G06T 2207/30181 Earth observation
• CPC Class Codes: G06T 2207/30212 Military
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06T 7/174 involving the use of two or more images
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 10/754 involving a deformation of the sample pattern or of the reference pattern; Elastic matching
• CPC Class Codes: G06V 20/13 Satellite images
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

The present invention features the use of the fundamental concept of color perception and multi-level resolution to perform scene segmentation and object/feature extraction in the context of self-determining and self-calibration modes. The technique uses only a single image, instead of multiple images as the input to generate segmented images. Moreover, a flexible and arbitrary scheme is incorporated, rather than a fixed scheme of segmentation analysis. The process allows users to perform digital analysis using any appropriate means for object extraction after an image is segmented. First, an image is retrieved. The image is then transformed into at least two distinct bands. Each transformed image is then projected into a color domain or a multi-level resolution setting. A segmented image is then created from all of the transformed images. The segmented image is analyzed to identify objects. Object identification is achieved by matching a segmented region against an image library. A featureless library contains full shape, partial shape and real-world images in a dual library system. The depth contours and height-above-ground structural components constitute a dual library. Also provided is a mathematical model called a Parzen window-based statistical/neural network classifier, which forms an integral part of this featureless dual library object identification system. All images are considered three-dimensional. Laser radar based 3-D images represent a special case.

Claims

1. A method of generalizing objects or features in an image, the steps comprising:

a) retrieving an original image, each pixel of which has a value represented by a predetermined number of bits, n;

b) transforming said original image into at least two distinct bands, each of said pixels in each of the band-images comprising less than n bits;

c) transforming said original image into at least two distinct resolutions by a series of down sampling schemes;

d) projecting each of said transformed, band-images, into a composite domain;

e) expanding each down-sampled image back to previous resolution by doubling, tripling, or quadrupling the pixels in both the x and y direction;

f) creating a composite image from all of said transformed, band-images; and

g) creating edge-based images by a series of comparisons and integrating among all resultant edge-based images.

2. The method of generalizing objects or features in an image in accordance with claim 1, the steps further comprising:

h) digitizing said composite image, if necessary; and

p1 i) analyzing regions in said segmented images to identify objects.

3. The method of generalizing objects or features in an image in accordance with claim 1, the steps further comprising:

g) coordinating said composite image with independently-generated information to identify features and objects.

4. The method of generalizing objects or features in an image in accordance with claim 1, wherein said transforming step (b) comprises compressing said original image, each of said bands using a different compression factor or technique, and said transforming step (c) comprises extracting said original image, each of said bands using a different down-sampling factor or technique.

5. The method of generalizing objects or features in an image in accordance with claim 4, wherein said compressing of said original image is accomplished by extracting the square root of the pixel values thereof.

6. The method of generalizing objects or features in an image in accordance with claim 4, wherein said compressing of said original image is accomplished by extracting the log of the pixel values thereof.

7. The method of generalizing objects or features in an image in accordance with claim 4, wherein said compressing of said original image is accomplished by extracting the double log of the pixel values thereof.

8. The method of generalizing objects or features in an image in accordance with claim 4, wherein said compressing of said original image is accomplished by filtering said pixel values thereof.

9. The method of generalizing objects or features in an image in accordance with claim 1, wherein said transforming step (b) comprises bit reduction and remapping without compression.

10. A self-calibrating, self-determining method of generalizing objects or features in an image, the steps comprising:

a) retrieving an original image in pixel form;

b) transforming said original image into at least one stable structure band;

c) executing a predetermined algorithm with said transformed image to perform iterative region-growing or region-merging by interrogating said image with a set of linearly-increasing color values;

d) generating groups having a set of values indicating a number of regions in each segmented image;

e) monitoring a slope and slope change of a scene characteristic (SC) curve;

f) establishing at least one stopping point;

g) projecting each of said transformed, band-images, into a composite domain; and

h) creating a composite image from all of said transformed, band-images.

11. The self-calibrating, self-determining method of generalizing objects or features in an image in accordance with claim 10, wherein said stopping point occurs when the slope or slope change of said SC curve is greater than zero.

12. The self-calibrating, self-determining method of generalizing objects or features in an image in accordance with claim 10, the steps further comprising:

i) digitizing said composite image, if necessary; and

j) analyzing regions in said segmented images to identify objects.

13. The self-calibrating, self-determining method of generalizing objects or features in an image in accordance with claim 12, the steps further comprising:

p1 k) coordinating said composite image with independently-generated information to identify features and objects.

14. The self-calibrating, self-determining method of generalizing objects or features in an image in accordance with claim 10, wherein said transforming step (b) comprises compressing said original image, each of said bands using a different compression factor or technique.

15. The method of generalizing objects or features in an image in accordance with claim 10, wherein said library comprises data representative of full images or a combination of portions thereof.

16. The method of generalizing objects or features in an image in accordance with claim 15, wherein said library further comprises a noise model for facilitating matching of an actual image with stored library elements.