Automated concrete/asphalt detection based on sensor time delay

A process for identifying a material in an image, comprising:

• accessing two bands of a multispectral image, the two bands are from different sensors at different positions of a focal plane;
• separately finding edges in the two bands;
• finding movement in the multispectral image based on change in position of edges between the two bands geographically aligned;
• identifying possible pavement using nominal spectral characteristics of pavement;
• filtering to remove edges that indicate movement that is not along possible pavement;
• filtering to remove edges that indicate movement that is not at possible speeds for moving vehicles;
• determining one or more spectral characteristics of sample pixels of the possible pavement near edges remaining after the filtering;
• and
• automatically identifying additional pixels in the image that have one or more spectral characteristics that are within a threshold of one or more spectral characteristics of the sample pixels, the identified additional pixels depict the material.

A non-transitory processor readable storage device having processor readable code embodied on the processor read storage device, the processor readable code for programming one or more processors to perform a method comprising:

• accessing two bands of a multispectral spacecraft image, the multispectral spacecraft image having multiple bands including a first set of bands and a second set of bands, the first set of bands are received from a first sensor at a first position on the spacecraft, the second set of bands are received from a second sensor at a second position on the spacecraft, the two bands include a first band from the first set of bands and a second band from the second set of bands, the first band is adjacent to the second band in terms of wavelength;
• determining edges in the first band using a difference of Gaussians process that results in a first edge sub-image and determining edges in the second band using the difference in Gaussian process that results in a second edge sub-image;
• finding movement in the multispectral spacecraft image based on determining change in position of edges between the first edge sub-image and the second edge sub-image by computing dissimilarity between image chips of the first edge sub-image and the second edge sub-image to identify a set of edges;
• identifying possible asphalt or concrete pixels in the multispectral spacecraft image using a spectral angle mapper, a nominal spectrum for asphalt and a nominal spectrum for concrete to create a nominal pavement mask;
• first filtering out edges of the set of edges that are not adjacent to possible asphalt or concrete pixels;
• second filtering out edges of the set of edges that are not moving at possible speeds for moving vehicles;
• identifying sample pixels of the multispectral spacecraft image that are in the nominal pavement mask and are near edges of the set of edges that were not filtered out by the first filtering and the second filtering;
• determining one or more spectral characteristics for the sample pixels;
• and
• identifying additional pixels in the multispectral spacecraft image that have one or more spectral characteristics that are within a threshold of one or more spectral characteristics of the sample pixels.