Change detection and change monitoring of natural and man-made features in multispectral and hyperspectral satellite imagery

• US 9,946,931 B2
• Filed: 04/20/2016
• Issued: 04/17/2018
• Est. Priority Date: 04/20/2015
• Status: Active Grant

• Current Assignee: TRIAD National Security, LLC
• Original Assignee: Los Alamos National Security LLC (Government of the United States of America)
• Inventors: Moody, Daniela Irina
• Primary Examiner(s): Park, Edward
• Application Number:
  US15/133,387
• Publication Number:
  US 20160307073A1
• Time in Patent Office: 727 Days
• CPC Class Codes: G06F 18/23 Clustering techniques
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G06V 20/13 Satellite images

Abstract

An approach for land cover classification, seasonal and yearly change detection and monitoring, and identification of changes in man-made features may use a clustering of sparse approximations (CoSA) on sparse representations in learned dictionaries. A Hebbian learning rule may be used to build multispectral or hyperspectral, multiresolution dictionaries that are adapted to regional satellite image data. Sparse image representations of pixel patches over the learned dictionaries may be used to perform unsupervised k-means clustering into land cover categories. The clustering process behaves as a classifier in detecting real variability. This approach may combine spectral and spatial textural characteristics to detect geologic, vegetative, hydrologic, and man-made features, as well as changes in these features over time.

Claims

1. A computer-implemented method, comprising:

learning representative land features, by a computing system, from multi-band patches or pixels of image data to form a learned feature dictionary;

applying representative features from the learned feature dictionary, by the computing system, to image patches or pixels from an image in a sparse fashion to form a sparse representation;

clustering the sparse representation of the features in the image, by the computing system, into land cover categories; and

performing land cover classification and change detection in the sparse domain, by the computing system, after the image is clustered and outputting results.

2. The computer-implemented method of claim 1, wherein the learning of the representative land features further comprises applying a Hebbian algorithm to learn representative land features directly from the multi-band patches or pixels of image data.

3. The computer-implemented method of claim 2, wherein the learning of the representative land features further comprises using available spectral bands in the image data in combinations to form normalized band difference indices.

4. The computer-implemented method of claim 1, wherein the learned feature dictionary is learned from at least one million overlapping image patches or pixels, and the applying of the representative features comprises using a pursuit search.

5. The computer-implemented method of claim 1, wherein the clustering of the features in the image comprises using unsupervised k-means clustering.

6. The computer-implemented method of claim 1, wherein the forming of the learned feature dictionary comprises:

initializing the learned feature dictionary, by the computing system, by imprinting;

learning the learned feature dictionary, by the computing system, using a batch/online Hebbian algorithm that minimizes a cost function E for sparse representation of an input vector x_i given by;

7. The computer-implemented method of claim 6, wherein an approximate solution for a sparse coefficient vector is obtained via a greedy matching pursuit algorithm.

8. The computer-implemented method of claim 6, wherein a dictionary update for every dictionary element is obtained by performing gradient descent on E for a batch of training vectors.

9. The computer-implemented method of claim 1, wherein a plurality of different spatial resolutions is used to learn distinct feature dictionaries.

10. The computer-implemented method of claim 1, further comprising:

extracting, by the computing system, values for each pixel in a given cluster for available band indices R, comprising a normalized difference vegetation index (NDVI), a normalized difference wetness index (NDWI), a normalized difference soil index (NDSI), and a nonhomogeneous feature distance (NHFD) band ratio; and

representing the given cluster, by the computing system, in R-dimensional space, such that the representation is defined by (NDVI, NDWI, NDSI, NHFD); and

using a distance metric in a band difference index R-dimensional space as a separability and performance metric.

11. The computer-implemented method of claim 1, wherein the change detection is performed by calculating a relative change given by:

relative_change=(cluster_size_t2−

cluster_size_t1)/cluster_size_t1

where time t₂ is subsequent to time t₁.

12. The computer-implemented method of claim 1, wherein the change detection is determined by relative percent changes in label count comprising a change in area, given by: