Process and apparatus for cross-correlating digital imagery

• US 5,719,949 A
• Filed: 02/22/1996
• Issued: 02/17/1998
• Est. Priority Date: 10/31/1994
• Status: Expired due to Term

• Current Assignee: MDA Information Systems LLC (Galileo Parent, Inc.)
• Original Assignee: MDA Information Systems, Inc. (Galileo Parent, Inc.)
• Inventors: Koeln, Gregory T., Mitchell, Roger A.
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): Kelley, Christopher S.
• Application Number:
  US08/605,625
• Time in Patent Office: 726 Days
• Field of Search: 382/100, 382/109, 382/113, 382/284, 348/143-147
• US Class Current: 382/113
• CPC Class Codes: G06T 17/05 Geographic models

Abstract

The present invention involves a system and method for processing information from high resolution data from maps, aerial photographs and other spatial data with low resolution data from limitation in order to accurately identify changes in the studied areas. The system and method relies upon a vectorized database which has been classified according to topographical features. Change is defined by a Z-statistic. The Z-statistic is calculated in a two-pass cross-correlation technique when pixel brightnesses are compared to a mean brightness for the typical topographical class and to a standard deviation for that class. The normalized difference values are accumulated across different bands, are scaled and then compared to a threshold which defines high, medium and low change values. A map is automatically produced illustrating areas of significant change.

Claims

1. A system for analyzing changes in image features comprising:

image data classification means for classifying high resolution image data into groups and summary groups;

rasterizing means for processing said high resolution image data into cells of a unitary size;

low resolution input means for retrieving and processing low resolution image data that correspond to said classified high resolution image data, comprising;

selector means for selecting spectral bands of said low resolution image data best-suited for cross-correlation with said high resolution image data,

geo-referencing means for geo-referencing said selected bands,

co-registration means for co-registering said selected bands with said high resolution image data, and

low resolution rasterizing means for processing said selected bands into cells of said unitary size;

cross-correlation means for defining a statistical value representing degree of change between said classified high resolution image data and said corresponding low resolution image data;

threshold means for comparing said cross-correlation statistical value in order to redefine thresholds representing significant, medium and low degrees of image change; and

mapping means for producing maps representing cells having maximum amounts of change.

2. The system according to claim 1, wherein said selector means selects spectral bands according to their utility in discriminating identifying characteristics of said image features.

3. The system according to claim 1, wherein said cross-correlation means includes:

means for determining a mean and standard deviation for each class code derived from said high resolution image data for each of said selected bands of said low resolution image data; and

Z-statistic calculating means for determining a Z-statistic value for each cell of said low resolution image data.

4. The system according to claim 3, wherein said threshold means includes:

an a priori threshold value for each class; and

a means for comparing said Z-statistic for each cell of said low resolution image data to said threshold value for said class in order to determine if any changes in said topographical feature has occurred.

5. The system according to claim 1, wherein said geo-referencing means includes:

means for creating correction coefficients for improving positional accuracy of said topographical features derived from said selected bands of said low resolution image data by using ephemeris data from a satellite generating said low resolution image data and ground control points from a map of said image features; and

means for testing the accuracy of said coefficients on said ground control points, and recalculating said coefficients until an a priori accuracy standard is met.

6. The system according to claim 1, wherein said co-registration means comprises:

means for merging said polygonal vectors of said high resolution image data with said geo-referenced low resolution image data;

means for measuring and correcting displacement between said high resolution image data and said low resolution image data; and

means for creating a buffered area based upon said displacement.

7. The system according to claim 3, wherein said Z-statistic calculating means comprises:

means for obtaining mean brightness values of all bands of said low resolution image data and a class value derived from said high resolution image data for a cell to be processed;

means for calculating a normalized difference between said mean brightness values and an observed brightness value for said selected bands of said cell;

means for accumulating said normalized difference for all said selected bands; and

means for multiplying said accumulated differences by a selected scalar.

8. The system according to claim 1, wherein said cells of unitary size are 25 by 25 meter-sized pixel cells.

9. An image processing system for analyzing changes in image features comprising:

a high resolution database for storing high resolution data as digital line graph ("DLG") data;

a classification scheme for topographical features represented within said high resolution data;

an attribute memory connected to said database for storing coding classifications according to said classification scheme;

a converter for converting said DLG data into polygonal vectors conforming to said topographical features;

a high resolution rasterizer for processing said high resolution data into 25 by 25 meter-sized pixel cells;

a high resolution vector memory for storing said rasterized high resolution data;

an attribute assignment device for assigning attribute data to said polygonal vectors;

a selector device for selecting optimal spectral bands of low resolution data for cross-correlation with said high resolution data;

a geo-referencing device for geo-referencing said selected bands of low resolution data;

a co-registration device for registering said low resolution data to said high resolution data;

a low resolution rasterizer for processing said low resolution data into 25 by 25 meter-sized pixel cells;

a low resolution data memory for storing said rasterized low resolution data; and

a cross-correlation device for defining a statistical value representing degree of change between said high resolution data and said associated low resolution data.