Method and apparatus for removing artifacts from aerial images

• US 9,230,308 B2
• Filed: 06/30/2011
• Issued: 01/05/2016
• Est. Priority Date: 06/30/2011
• Status: Active Grant

• Current Assignee: Weyerhaeuser NR Company (Weyerhauser)
• Original Assignee: Weyerhaeuser NR Company (Weyerhauser)
• Inventors: Ma, Zhenkui
• Primary Examiner(s): Yeh, Eueng-Nan
• Application Number:
  US13/173,089
• Publication Number:
  US 20130004065A1
• Time in Patent Office: 1,650 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/10036 Multispectral image; Hyperspectral image
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/30192 Weather; Meteorology
• CPC Class Codes: G06T 5/50 using two or more images, e.g. averaging or subtraction
• CPC Class Codes: G06T 5/77 Retouching; Inpainting; Scratch removal
• CPC Class Codes: G06T 7/90 Determination of colour characteristics

Abstract

A computer system for creating artifact-free aerial images of a region of interest. A computer system receives one or more input aerial images that are obtained at different times. A false-color image is created by applying two or more of the input aerial images to different color channel inputs of the false-color image. Based on the color of the pixels in the false-color image, pixels in the two or more input aerial images are classified as representing clear, cloud or shadow areas. An output image is created by combining pixels that are classified as representing clear areas from two or more of the input aerial images.

Claims

1. A programmed computer system for removing artifacts from aerial images of a region of interest, including:

a memory for storing program instructions that are executable by a processor;

a processor configured to execute the stored program instructions in order to;

receive two or more input aerial images of the region of interest that are obtained at different times;

create a three-channel false color image by applying pixel data for a selected band of a first input aerial image to a red color channel of an RGB monitor, applying pixel data for the selected band of a second input aerial image to a green color channel of the RGB monitor and applying pixel data for the selected band of a third input aerial image to a blue color channel of the RGB monitor;

identify pixels representing clear areas, clouds and shadows in the two or more input aerial images based on the color of the pixels in the false color image, wherein;

pixels representing clear areas in the first, second or third input aerial images are identified by gray pixels in the false color image,

pixels representing clouds in the first input aerial image are identified as pixels having a bright red color;

pixels representing clouds in the second input aerial image are identified as pixels having a bright green color;

pixel representing clouds in the third input aerial image are identified as pixels having a bright blue color;

pixels representing overlapping clouds in the first and second input aerial images are identified as pixels having a bright yellow color;

pixels representing overlapping clouds in the first and third input aerial images are identified as pixels having a bright magenta color;

pixels representing overlapping clouds in the second and third input aerial images are identified as pixels having a bright cyan color; and

wherein

pixels representing shadows in the first input aerial image are identified as pixels having a dark cyan color and that are nearby pixels having a bright red color;

pixels representing shadows in the second input aerial image are identified as pixels having a dark magenta color and that are nearby pixels having a bright green color;

pixels representing shadows in the third input aerial image are identified as pixels having a dark yellow color and that are nearby pixels having a bright blue color;

pixels representing overlapping shadows the first and second input aerial images are identified as pixels having a dark blue color and that are nearby pixels having a bright yellow color;

pixels representing overlapping shadows the first and third input aerial images are identified as pixels having a dark green color and that are nearby pixels having a bright magenta color;

pixels representing overlapping shadows the second and third input aerial images are identified as pixels having a dark red color and that are nearby pixels having a bright cyan color; and

create an output image of the region of interest by selecting one of the first, second or third input aerial images and replacing pixels that represent cloud or shadow regions in the selected aerial image with pixels representing a clear area from the non-selected first, second or third input aerial images.

2. The programmed computer system of claim 1, further comprising:

instructions that are executable by the processor to create a composite image mask having a number of pixels each of which corresponds to a pixel in the false color image, wherein the pixels in the image mask classify each pixel in the two or more input aerial images that create the false color image as representing a clear, cloud or shadow region.

3. The programmed computer system of claim 1, further comprising:

instructions that are executable by the processor to compute a mean pixel value and standard deviation for each of the two or more input aerial images for overlapping areas that are identified as being free of clouds and shadows.

4. The programmed computer system of claim 1, further comprising:

instructions that are executable by the processor to scale pixel values from the two or more input aerial images based on statistics of the output image and on statistics of the input image used to supply pixel data to the output image.

5. A non-transitory computer readable medium having instructions thereon that are executable by a, processor to remove artifacts from aerial images of a region of interest, by:

receiving two or more input aerial images of the region of interest that are obtained at different times;

creating a three-channel false color image by applying pixel data for a selected band of a first input aerial image to a red color channel of an RGB monitor, applying pixel data for the selected band of a second input aerial image to a green color channel of the RGB monitor and applying pixel data for the selected band of a third input aerial image to a blue color channel of the RGB monitor;

identifying pixels representing clear areas, clouds and shadows in the two or more input aerial images based on the color of the pixels in the false color image, wherein;

pixels representing clear areas in the first, second or third input aerial images are identified by gray pixels in the false color image,

pixels representing clouds in the first input aerial image are identified as pixels having a bright red color;

pixels representing clouds in the second input aerial image are identified as pixels having a bright green color;

pixels representing clouds in the third input aerial image are identified as pixels having a bright blue color;

pixels representing overlapping clouds in the first and second input aerial images are identified as pixels having a bright yellow color;

pixels representing overlapping clouds in the first and third input aerial images are identified as pixels having a bright magenta color;

pixels representing overlapping clouds in the second and third input aerial images are identified as pixels having a bright cyan color; and

wherein

pixels representing shadows in the first input aerial image are identified as pixels having a dark cyan color and that are nearby pixels having a bright red color;

pixels representing shadows in the second input aerial image are identified as pixels having a dark magenta color and that are nearby pixels having a bright green color;

pixels representing shadows in the third input aerial image are identified as pixels having a dark yellow color and that are nearby pixels having a bright blue color;

pixels representing overlapping shadows the first and second input aerial images are identified as pixels having a dark blue color and that are nearby pixels having a bright yellow color;

pixels representing overlapping shadows the first and third input aerial images are identified as pixels having a dark green color and that are nearby pixels having a bright magenta color;

pixels representing overlapping shadows the second and third input aerial images are identified as pixels having a dark red color and that are nearby pixels having a bright cyan color; and

create an output image of the region of interest by selecting one of the first, second or third input aerial images and replacing pixels that represent cloud or shadow regions in the selected aerial image with pixels representing a clear area from the non-selected first, second or third input aerial images.