Object motion mapping using panchromatic and multispectral imagery from single pass electro-optical satellite imaging sensors
OwnerSteven McArdle, Trung Pham +2inventors
InventorsSteven McArdle, Trung Pham, Jian Yang, Mehdi Hosseini
Life of this patent
Estimating the term…
Abstract
The present invention is a semi-automated process to extract object motion and attributes from utilizing a remote sensing methodology using Earth Observation data from single pass satellite imagery. Many single pass satellite sensors collect imagery that include a panchromatic and multispectral image with a slight temporal offset. The method of the present invention performs image processing, object segmentation, object measurement, image normalization and image matching and velocity calculation to extract physical attributes of the object, object location, and object motion.
Claims19
A method to extract a set of object parameters to determine an object velocity vector and physical attributes from a single-pass satellite imagery that provides a panchromatic (PAN) image and a multispectral (MS) image having a set of spectral bands and a temporal offset, said method comprising steps of:
- a) identifying two seed points, a front and a back seed points, on the PAN and the MS images at front and back of an object,
- b) performing an image registration;
- c) performing an object segmentation comprising steps of:
- using front and back seed points to obtain an object profile as a line connecting said front and said back seed points, and identifying an object centroid as being a middle point of the object profile;
- enabling a set of perpendicular profiles perpendicular to the object profile in a mesh-like structure extending across and beyond the object;
- using the set of perpendicular profiles to detect edges of the object by identifying a set of edge-pixels;
- removing a plurality of noise pixels from the set of edge-pixels;
- refining edges of the object by refining the set of edge-pixels, and
- infilling the set of edge-pixels to create a complete edge shape and a perimeter of the object;
- d) performing an image normalization to resample the multispectral image to create a Simulated Panchromatic Image (SPI) that matches the Panchromatic image (PAN), now referred to as an Original Panchromatic Image (OPI), by performing a rescaling and a dynamic regression analysis to determine a set of grey level values (DN) for creating the SPI;
- e) selecting and clipping the OPI and the SPI for an image matching;
- f) matching the OPI and a spatially displaced SPI at a coincident location of the object, and applying a point matching of a set of target pixels in the OPI to extract corresponding pixels in the SPI;
- g) determining the object velocity vector having a speed and a trajectory by obtaining a displacement of said set of target pixels, and
- h) performing an image quality testing by a set of similarity measurements performed on the object profile and the object segmentation.
Dependencies are inferred from the claim text.