Background
The multispectral imaging system is generally applied to the field with higher requirements on computational accuracy, but the multispectral imaging system is more complex in device and expensive in price, and the labor and material resources are more consumed when some trial and exploratory research experiments are carried out. The spectral sensitivity of the multispectral camera sensor can be acquired for simulation experiments, prior guidance can be provided for actual experiments, blind exploration in the experiments can be avoided, a user walks with less bending, and material consumption cost and energy and time of experimenters can be saved.
The spectral sensitivity of multispectral camera sensors can be measured, but requires more sophisticated and expensive instruments, and the measurement process is complex and time consuming. Meanwhile, the response value of the multispectral camera is not only related to the spectral sensitivity of the sensor, but also affected by the spectral radiation characteristic of the light source, the optical characteristic of the optical path and the spectral transmittance characteristic of the color filter, and the factors also need to be considered in a simulation experiment, and the characteristics need to be measured by corresponding instruments, so that the difficulty of the simulation experiment is increased.
Disclosure of Invention
In order to overcome the defects of difficult actual operation, high simulation experiment difficulty and multiple influence factors in the prior art, the invention provides the method for estimating the synthesized spectral sensitivity of the narrow-band multispectral camera, which can reduce the influence of noise on the estimated synthesized spectral sensitivity and improve the estimation effect of the synthesized spectral sensitivity of the narrow-band multispectral camera.
In order to achieve the purpose, the invention adopts the following technical scheme:
a method for estimating the synthesized spectral sensitivity of a narrow-band multispectral camera comprises the following steps:
(1-1) obtaining a multichannel response value of the standard color card
And correcting the multi-channel response value of the whiteboard
Wherein P and P
bThe matrix is n rows and q columns, wherein n is the number of color blocks on the standard color card, q is the number of channels of the multispectral camera, each row represents the response value of q channels corresponding to one color block, and each column represents the response value of n color blocks output by one channel;
(1-2) correcting the multichannel response value of the standard color card by using the multichannel response value of the correction white board;
(1-3) measuring the spectral reflectance of all color blocks in the standard color chart with a spectrophotometer, and obtaining the spectral reflectance of black, white and gray blocks excluding the repetition of the periphery and the center of the color chart
R is a matrix with n rows and m columns, wherein n is the number of color blocks on a standard color card, m is the dimension of spectral reflectance, and each row represents the spectral reflectance data of one color block;
(1-4) performing singular value decomposition on the spectral reflectance R of the black, white and gray blocks with repeated periphery and center of the color card removed to obtain a unitary matrix U, a unitary matrix V and a singular value matrix W;
and (1-5) solving the spectral sensitivity Q of the narrow-band multispectral camera by utilizing a least square method with non-negative restriction based on the corrected multichannel response value matrix of the standard color card and the unitary matrix and singular value matrix of the spectral reflectance ratio matrix.
Shooting a multispectral image of a standard color card by using a multispectral camera under an ambient light source, keeping the position of the multispectral camera unchanged, placing a correction white board at the standard color card, shooting the multispectral image of the correction white board and correcting a multichannel response value of the standard color card by using the multispectral image; measuring the spectral reflectance of a standard color card by using a spectrophotometer, calculating a unitary matrix and a singular value matrix of a spectral reflectance matrix consisting of all color blocks by using singular value decomposition, and calculating the synthesized spectral sensitivity of the multispectral camera by combining a multi-channel response value corrected by the standard color card and a non-negative limiting condition.
Preferably, the specific steps of step (1-1) are as follows:
(2-1) shooting a standard color card by using a narrow-band multispectral camera under an ambient light source, collecting multispectral images of the standard color card and obtaining a multichannel response value of the standard color card
(2-2) keeping the position of the narrow-band multispectral camera unchanged, and placing the correction white board at the standard color card and covering the whole area of the color card;
(2-3) shooting the correction white board by using a narrow-band multispectral camera, collecting multispectral images of the correction white board and obtaining a multichannel response value of the correction white board
Preferably, the formula for correcting the multi-channel response value of the standard color card by using the multi-channel response value of the correction white board is as follows
Pc=P./Pb
Wherein,
for the corrected multi-channel response value matrix of the standard color chart, "/" indicates the matrix P and the matrix P
bDivided by the corresponding element in (1).
Preferably, the formula for solving the spectral sensitivity Q of the narrow-band multispectral camera by using a least square method with non-negative restriction is
Q=(UWVT)+Pc=(V(W(l))-1UT)Pc (Q≥0)
Wherein
The estimated synthesized spectral sensitivity of the narrow-band multispectral camera is obtained, Q is a matrix of m rows and Q columns, each column is the synthesized spectral sensitivity of one channel, and l represents the number of adopted singular vectors; (UWV)
T)
+Representation matrix UWV
TA pseudo-inverse matrix of (d); (W)
(l))
-1A representation matrix W
(l)The inverse matrix of (d); u shape
TA transposed matrix representing the matrix U; v
TRepresenting the transpose of matrix V.
Preferably, l has a value in the range of 3 to 8.
Therefore, the invention has the following beneficial effects: the spectral response characteristics of the whole multispectral camera are estimated by using a standard color chart without measuring the spectral characteristics of all the constituent units of the multispectral imaging system but taking the multispectral imaging system as a whole; meanwhile, the influence of noise on an estimation result is reduced by utilizing singular value decomposition, the actual physical characteristics of the channel of the multispectral camera are considered, non-negative limitation is added in the calculation process, the estimation precision of the synthesized spectral sensitivity is improved, and the convenience and the precision of a simulation experiment are improved.
Detailed Description
The invention is further described in the following detailed description with reference to the drawings in which:
examples
The multispectral camera used in this embodiment is a color filter wheel type narrow-band multispectral camera, which has 16 channels formed by 16 interference color filters, and the FWHM (Full Width at Half Maximum of Half Maximum) of each of the 16 interference color filters is 20nm, and the peak transmittance wavelengths are 400nm, 420nm, 440nm, 460nm, 480nm, 500nm, 520nm, 540nm, 560nm, 580nm, 600nm, 620nm, 640nm, 660nm, 680nm and 700nm, respectively.
The standard Color card adopts a Color Checker DC (DC) Color card produced by Gretag Macbeth company, 168 Color blocks are adopted, and black, white and gray blocks which are repeated at the periphery and the center and 8 glossy Color blocks are abandoned.
The ambient Light source was D65 Light source in a Gretag Macbeth Spectra Light III Light box.
The embodiment shown in fig. 1 is a method for estimating the synthesized spectral sensitivity of a narrow-band multispectral camera, comprising the following steps:
step 100, obtaining a multi-channel response value of a standard color card
And correcting the multi-channel response value of the whiteboard
Step 101, shooting a DC color card by using a 16-channel narrow-band multispectral camera under a D65 light source, collecting a multispectral image of the DC color card and obtaining a multichannel response value of the DC color card
P is a matrix with 168 rows and 16 columns, where 168 is the number of color blocks on a standard color card, 16 is the number of multispectral camera channels, each row represents the response value of 16 channels corresponding to one color block, and each column represents the response value of 168 color blocks output by one channel;
step 102, keeping the position of the narrow-band multispectral camera unchanged, and placing a correction white board at the standard color card and covering the whole area of the color card;
step 103, shooting the correction white board by using the narrow-band multispectral camera, collecting multispectral images of the correction white board and obtaining a multichannel response value of the correction white board
P
bA matrix of 168 rows and 16 columns;
200, correcting a multichannel response value of the standard color card by using the multichannel response value of the correction white board;
correcting the multichannel response value of the standard color card by using the multichannel response value of the correction white board, wherein the correction formula is
Pc=P./Pb
Wherein,
for the corrected multi-channel response value matrix of the standard color chart, "/" indicates the matrix P and the matrix P
bIs divided by the corresponding element in (1);
step 300, measuring the spectral reflectance of 168 color blocks in the DC color card by using a spectrophotometer, and obtaining the spectral reflectance of black, white and gray blocks with the repetition of the periphery and the center of the color card removed
R is a matrix of 168 rows and 31 columns, where 168 is the number of color blocks on the DC color card, 31 is the dimension of the spectral reflectance, and each row represents the spectral reflectance data for one color block;
step 400, performing singular value decomposition on the spectral reflectance R of the black, white and gray blocks with repeated color card periphery and center removed to obtain a unitary matrix U, a unitary matrix V and a singular value matrix W;
step 500, based on the corrected multi-channel response value matrix of the standard color card and the unitary matrix and singular value matrix of the spectral reflectance ratio matrix, the least square method with non-negative restriction is used for solving the spectral sensitivity Q of the narrow-band multispectral camera, namely
Q=(UWVT)+Pc=(V(W(l))-1UT)Pc (Q≥0)
Wherein
The estimated synthesized spectral sensitivity of the narrow-band multispectral camera is obtained, Q is a matrix with 31 rows and 168 columns, each column is the synthesized spectral sensitivity of one channel, l represents the number of adopted singular vectors, l is 6, and the cumulative contribution rate of the first 6 singular vectors is larger than 95%; (UWV)
T)
+Representation matrix UWV
TA pseudo-inverse matrix of (d); (W)
(l))
-1A representation matrix W
(l)The inverse matrix of (d); u shape
TA transposed matrix representing the matrix U; v
TRepresenting the transpose of matrix V.
Fig. 2 is the synthesized spectral sensitivity estimated by the above method, and as shown in fig. 2, it can be seen that the synthesized spectral sensitivity estimated by the method proposed by the present invention is non-negative and substantially unimodal while exhibiting good narrow-band characteristics.
It should be understood that this example is for illustrative purposes only and is not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and such equivalents may fall within the scope of the present invention as defined in the appended claims.