Model generation and application for removing atmospheric effects in imagery based on data from an aerial platform
AssigneeAdvanced Remote Sensing, Inc.
InventorDavid Groeneveld
Life of this patent
Estimating the term…
Abstract
Systems and methods for generating and using statistical models to mitigate atmospheric effects in images are described. In some embodiments, a statistical model may be generated by selecting a vegetation type that grows in continuous healthy canopies; identifying a vegetation reference value that is a stable reflectance property of the vegetation type; in a plurality of images, selecting one or more plots of the vegetation type and obtaining top-of-atmosphere reflectance for the plots; selecting discrete areas near the plots and obtaining top-of-atmosphere reflectance for the discrete areas; obtaining image statistics for the discrete areas; and generating a statistical model based on the acquired data.
Claims16
A method for generating a statistical model for spatial representation of a multitude of atmospheric effect magnitudes distributed across a digitally recorded observation device image of the Earth's surface taken from above, to be used for scaling the removal of the atmospheric effect from a new device image, the method comprising:
- selecting a vegetation type that grows in continuous healthy canopies;
- identifying a surface reflectance vegetation reference value in a blue spectrum that is a stable property of the vegetation type to serve as a reference against which a magnitude of an atmospheric effect can be estimated, the atmospheric effect degrading image data according to its magnitude relative to what would be recorded at Earth's surface having no atmospheric effect;
- identifying the vegetation type growing in the continuous healthy canopies on a plurality of observation device images that express a range of atmospheric effect magnitudes;
- for each of the plurality of observation device images:
- selecting one or more plots of the vegetation type, each such plot expressing healthy, homogeneous and continuous vegetation cover;
- extracting a top-of-atmosphere reflectance value in a blue band for the one or more plots:
- selecting one or more discrete areas, each of the one or more discrete areas being within a 20 kilometer distance from a selected-vegetation plot;
- determining, for each discrete area of the one or more discrete areas, top-of-atmosphere reflectance statistics for two or more spectral bands;
- pairing the top-of-atmosphere reflectance statistics for the two or more spectral bands with the top-of-atmosphere reflectance value in the blue band, thereby forming one or more sample pairs, each consisting of the top-of-atmosphere reflectance value in the blue band and, for a discrete area located within the 20 kilometer distance, the top-of-atmosphere reflectance statistics for the two or more spectral bands;
- combining paired sample pairs plots and their one or more discrete areas obtained from the plurality of observation device images into a pooled sample;
- and
- from the pooled sample, generating a statistical model to predict a new top-of-atmosphere blue spectral band vegetation reference value from new spectral band statistics sampled from new discrete areas on new device images that represents the magnitude of the atmospheric effect, wherein an atmospheric effect magnitude forming a grayscale across the new device image area pursuant to additional steps for removal of the atmospheric effect from the device image,
- wherein the use of the image-sampled statistics from plots or spatially discrete areas applied for modeling can include statistics from plots and spatially discrete areas measured by spectrometry from a manned or unmanned aerial platform.
A system for generating a statistical model to map a spatial representation of atmospheric effect magnitude across an observation device image pursuant to removal of the atmospheric effect, thereby delivering estimates of surface reflectance, the system comprising:
- a computer processor, and
- a computer-readable medium storing instructions that, when executed by the processor, are configured to cause the system to perform operations, wherein the operations include:
- selecting a vegetation type that grows in continuous healthy canopies;
- obtaining spectrometer measurements to identify a blue band surface reflectance vegetation value that is a stable reflectance property of the vegetation type to serve as a reference against which the atmospheric effect can be estimated, the atmospheric effect resulting in degradation of image data relative to what would be recorded at Earth's surface;
- identifying the vegetation type growing in continuous healthy canopies on a plurality of observation device images that expresses a range of atmospheric conditions;
- for each of the plurality of observation device images:
- selecting one or more plots of the vegetation type, each such plot expressing continuous, healthy cover of the vegetation type;
- measuring a top-of-atmosphere blue spectral band vegetation reflectance value for each selected-vegetation plot;
- selecting one or more discrete areas, each of the one or more discrete areas within a 20 kilometer distance of a selected-vegetation plot;
- obtaining a top-of-atmosphere reflectance for two or more spectral bands of an observation device from the one or more discrete areas;
- determining, for each discrete area of the one or more discrete areas, a plurality of statistics based on the top-of-atmosphere reflectance of the two or more spectral bands;
- pairing the plurality of statistics for the one or more discrete areas with the top-of-atmosphere blue spectral band reflectance value of a nearest vegetation plot, thereby forming sample pairs;
- combining sample pairs from the plurality of observation device images into a pooled sample;
- from the pooled sample, generating a statistical model that predicts new top-of-atmosphere vegetation reference values as an index predicted by the model from spectral band statistics sampled from new discrete areas for the purpose of removing the atmospheric effect from the device image,
- wherein the use of the image-sampled statistics from plots or spatially discrete areas applied for modeling can include statistics from plots and spatially discrete areas measured by spectrometry from a manned or unmanned aerial platform.
Dependencies are inferred from the claim text.