发明名称 |
Estimating a visible vector representation for pixels in an infrared image |
摘要 |
What is disclosed is a system and method for estimating color for pixels in an infrared image. In one embodiment, an infrared image is received which has been captured using a N-band infrared imaging system comprising a multi-spectral camera or a hyperspectral camera. The IR image is composed of an array of pixels with N intensity values having been collected for each pixel in the image. Then, for each pixel of interest, a search metric is used to search a database of vector samples to identify a visible-IR set which is closest to the intensity values of the IR band vector collected for the pixel. A visible vector representation is then estimated for the pixel based upon the visible portion corresponding to the closest visible-IR set. Thereafter, color coordinates for this pixel are computed from the visible vector. The method repeats for all pixels of interest in the IR image. |
申请公布号 |
US8810658(B2) |
申请公布日期 |
2014.08.19 |
申请号 |
US201213364835 |
申请日期 |
2012.02.02 |
申请人 |
Xerox Corporation |
发明人 |
Skaff Sandra;Bala Raja;Mestha Lalit Keshav;Xu Beilei |
分类号 |
H04N5/33 |
主分类号 |
H04N5/33 |
代理机构 |
Fleit Gibbons Gutman Bongini & Bianco P.L. |
代理人 |
Blair Philip E.;Fleit Gibbons Gutman Bongini & Bianco P.L. |
主权项 |
1. A method for estimating a visible vector representation for pixels in an infrared image, the method comprising:
receiving an infrared (IR) image captured using a N-band infrared imaging system, said IR image comprising a plurality of pixels with each pixel having an IR band vector of N intensity values collected for that pixel; and for each pixel of interest in said infrared image:
using a search metric to search a spectra database of vector samples to identify a visible-IR set with intensity values closest to this pixel's IR band vector, where said visible-IR sets and said intensity values in the IR band are narrowband spectra, said search metric comprising: where N is the number of spectral bands of said imaging system, NcIR is the number of samples in said visible-IR set, SIRxcIR(λi) is the IR reflectance of sample xcIR from the visible-IR set at wavelength λi, and SIRxIR(λi) is the IR reflectance of sample xIR from the IR set at wavelength λi, with a visible portion of xcIR being assigned to sample xIR; and
estimating a visible vector representation for said pixel based upon the visible portion corresponding to said closest visible-IR vector. |
地址 |
Norwalk CT US |