CN116029938A - Image flat field correction method and device based on linear fitting - Google Patents
Image flat field correction method and device based on linear fitting Download PDFInfo
- Publication number
- CN116029938A CN116029938A CN202310160209.3A CN202310160209A CN116029938A CN 116029938 A CN116029938 A CN 116029938A CN 202310160209 A CN202310160209 A CN 202310160209A CN 116029938 A CN116029938 A CN 116029938A
- Authority
- CN
- China
- Prior art keywords
- image
- matrix
- value
- signal
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000003705 background correction Methods 0.000 title claims abstract description 58
- 239000011159 matrix material Substances 0.000 claims abstract description 130
- 238000012937 correction Methods 0.000 claims abstract description 31
- 238000012546 transfer Methods 0.000 claims description 4
- 238000004519 manufacturing process Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000004075 alteration Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000903 blocking effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000005669 field effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses an image flat field correction method and device based on linear fitting, wherein the method comprises the following steps: collecting a plurality of images of a photographed object; acquiring an original pixel value of each pixel point in each image; establishing a signal gain matrix and a signal offset matrix for each pixel point in each image; establishing a linear equation set of corrected pixel values for each image according to the original pixel values, the signal gain matrix and the signal offset matrix; performing linear fitting based on a least square method to obtain each signal gain value in a signal gain matrix and each signal offset value in a signal offset matrix; and calculating and obtaining the correction pixel value of each image according to the signal gain matrix, the signal offset matrix and the linear equation set which are obtained by fitting. Performing flat field correction on each image according to the corrected pixel values; the method can effectively reduce the production cost.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an image flat field correction method and device based on linear fitting.
Background
In general, in an image acquired by an actual camera, due to factors such as uneven illumination, inconsistent response of the lens center and the lens edge, inconsistent response of each pixel of an imaging device, fixed image background noise and the like, each pixel value in the image often has larger error compared with the actual situation, which is very unfavorable for the subsequent use of the picture, so that flat field correction is performed first, and the problem of nonuniform response of each pixel is solved. The flat field correction is to make the response lines of all pixels identical by changing the slope and offset of the response line of each pixel, i.e. the signal gain and signal offset. Generally, two steps of dark field correction and open field correction are needed for flat field correction, firstly, a camera performs one-time exposure on the dark field to obtain the offset of each pixel; then, imaging the gray level uniform object under the uniform illumination condition for one time to obtain a uniform field image, and preferably enabling all points in the image to approach the maximum gray level value; and finally subtracting the dark field image from the uniform light field image, and correcting the image gain by a relative calibration method. This method is time consuming and cannot be used without dark field images, since two images under different parameters need to be acquired.
Patent document CN109660736a discloses a flat field correction method and apparatus, and an image verification method and apparatus, first, adjusting exposure time and camera gain to obtain a dark field image and a standby bright field image, then performing mean filtering on the standby bright field image to obtain a bright field image, and finally correcting the image by using a flat field correction algorithm in combination with the dark field image and the bright field image, so as to improve correction accuracy of flat field correction, but because camera parameters need to be adjusted to obtain the dark field image and the bright field image, efficiency is lower, and the experimental apparatus and environment are also depended.
Patent document CN106204492B discloses a real-time flat field correction method of an area array camera based on an FPGA, performing a blocking operation on a sensor array, and performing an operation on each point by combining an interpolation algorithm on the basis of the blocking. The invention provides higher output image quality for high-resolution high-frame rate application occasions, but can not cope with ultra-high resolution pictures due to the limitation of the storage capacity of the FPGA, and the additional use of an area-array camera and the FPGA also increases the cost of flat-field correction.
Patent document CN115037922a discloses a uniform light source platform for camera calibration, a non-uniformity measurement, correction method, and a flat field correction of a camera using a field dark field image obtained by the platform. The platform has simple structure and low cost, realizes uniformity of a light source, has high requirements on the precision of the arrangement position of devices such as a camera, a baffle plate, an integrating sphere and the like, and can give larger error to a correction result in actual operation.
In summary, the image flat field correction method in the prior art is time-consuming in calculation and high in equipment requirement, so that the production cost is high.
Disclosure of Invention
The invention provides an image flat field correction method and device based on linear fitting, which can realize image flat field correction on the algorithm level without dark field images or other devices, and can effectively reduce production cost.
An image flat field correction method based on linear fitting, comprising:
collecting a plurality of images of a photographed object;
acquiring an original pixel value of each pixel point in each image;
establishing a signal gain matrix and a signal offset matrix for each pixel point in each image;
establishing a linear equation set of corrected pixel values for each image according to the original pixel values, the signal gain matrix and the signal offset matrix;
performing linear fitting based on a least square method to obtain each signal gain value in a signal gain matrix and each signal offset value in a signal offset matrix;
calculating and obtaining a correction pixel value of each image according to the signal gain matrix, the signal offset matrix and the linear equation set which are obtained by fitting;
and carrying out flat field correction on each image according to the correction pixel value.
Further, in the plurality of images, the number of row pixel points is the same, the number of column pixel points is the same, and gray values are different between different images.
Further, the number of the plurality of images is K, and each image includes n×m pixels; the signal gain matrix is:
the signal offset matrix is:
wherein i=0, 1,2, … … n-1, j=0, 1,2, … … m-1, a ij Signal gain value, b, representing pixel point of ith row and jth column in image ij And the signal offset value of the pixel point in the ith row and the jth column in the image is represented.
Further, the linear equation of the ith row and jth column pixel points of the kth image is:
；
wherein x is ij_k Representing the original pixel value, y, of the ith row and jth column pixel points in the kth image k Corrected pixel values representing the kth image, i=0, 1,2, … … n-1, j=0, 1,2, … … m-1, a ij Signal gain value, b, representing pixel point of ith row and jth column in image ij A signal offset value representing an ith row and jth column of pixels in the image;
the linear equation set comprises a linear equation of each pixel point of each image.
Further, performing linear fitting based on a least square method to obtain each signal gain value in the signal gain matrix, including:
subtracting the linear equations of two adjacent columns of pixel points in the same row in each image in the linear equation set to obtain a first equation set for eliminating correction pixel values;
sequentially subtracting the equations in the first equation set to obtain a second equation set for eliminating the signal offset value;
setting a signal gain value of a 0 th column in a second equation set, and converting the second equation set into a matrix form to obtain a first matrix;
performing linear fitting on elements in the first matrix as coordinate points to obtain a first linear equation about a signal gain value to be solved;
transpose the first matrix to obtain a first transpose matrix;
and solving a signal gain value to be solved in the first linear equation according to the first transfer matrix.
Further, the range of the signal gain value of the 0 th column in the second equation set is (0, 1).
Further, performing linear fitting based on a least square method to obtain each signal offset value in the signal offset matrix, including:
subtracting the linear equations of two adjacent columns of pixel points in the same row in each image in the linear equation set to obtain a first equation set for eliminating correction pixel values;
designating a signal offset value of a 0 th column in the first equation set, and converting the first equation set into a matrix form to obtain a second matrix;
performing linear fitting on elements in the second matrix as coordinate points to obtain a second linear equation about the signal offset value to be solved;
transposing the second matrix to obtain a second transposed matrix;
and solving a signal offset value to be solved in the second linear equation according to the second transposed matrix.
Further, the value of the signal offset value of the 0 th column in the first equation set is the average value of the original pixels of the 0 th image, and the product of the signal gain value of the 0 th column and the original pixel value of the 0 th row and the 0 th column in the second equation set is subtracted;
or,
the value of the signal offset value of the 0 th row in the first equation set is the preset correction pixel value of the 0 th image, and the product of the signal gain value of the 0 th row and the pixel value of the 0 th row in the second equation set and the pixel value of the 0 th row and the 0 th column in the 0 th image is subtracted.
Further, performing flat field correction on each image according to the corrected pixel value, including:
the pixel value of each image is set as the calculated corrected pixel value.
An image flat field correction device based on linear fitting, comprising:
the acquisition module is used for acquiring a plurality of images of the shot object;
the original pixel acquisition module is used for acquiring an original pixel value of each pixel point in each image;
a matrix establishing module for establishing a signal gain matrix and a signal offset matrix about each pixel point in each image;
the equation building module is used for building a linear equation set of corrected pixel values of each image according to the original pixel values, the signal gain matrix and the signal offset matrix;
the fitting module is used for carrying out linear fitting based on a least square method to obtain each signal gain value in the signal gain matrix and each signal offset value in the signal offset matrix;
the calculating module is used for calculating and obtaining a correction pixel value of each image according to the signal gain matrix, the signal offset matrix and the linear equation set which are obtained by fitting;
and the correction module is used for carrying out flat field correction on each image according to the correction pixel value.
The image flat field correction method and device based on linear fitting provided by the invention at least comprise the following beneficial effects:
(1) The built-in parameters of the camera are not required to be obtained or changed, so that the flat field correction can be performed only by obtaining a group of images shot by the camera, and the universality is strong;
(2) All parameters can be fitted through a group of pictures without additionally shooting dark field images by using a camera, and the efficiency is high.
(3) Only a least square method linear fitting method is used, the method is easy to understand and realize, and accumulated errors superposed by a plurality of methods are avoided, so that the image precision after flat field correction is high;
(4) And the flat field correction can be completed only at the algorithm level without additional experiment platforms and devices, the operation is convenient and simple, and the production cost is effectively reduced.
Drawings
Fig. 1 is a flowchart of an embodiment of an image flat field correction method based on linear fitting according to the present invention.
Fig. 2 is a schematic diagram of an embodiment of an original image pixel in the linear fitting-based image flat field correction method according to the present invention.
Fig. 3 is a flowchart of an embodiment of obtaining a signal gain by fitting in the image flat field correction method based on linear fitting according to the present invention.
Fig. 4 is a flowchart of an embodiment of obtaining a signal offset by fitting in the image flat field correction method based on linear fitting according to the present invention.
Fig. 5 is a schematic diagram of linear fitting under an application scenario of the image flat field correction method based on linear fitting provided by the invention.
Fig. 6 is a schematic diagram of a signal gain value obtained by fitting in an application scenario according to the image flat field correction method based on linear fitting provided by the invention.
Fig. 7 is a schematic diagram of a signal offset obtained by fitting in an application scenario according to the image flat field correction method based on linear fitting provided by the invention.
Fig. 8 is a schematic structural diagram of an embodiment of an image flat field correction device based on linear fitting according to the present invention.
Description of the embodiments
In order to better understand the above technical solutions, the following detailed description will be given with reference to the accompanying drawings and specific embodiments.
Referring to fig. 1, in some embodiments, an image flat field correction method based on linear fitting is provided, comprising:
s1, collecting a plurality of images of a shot object;
s2, acquiring an original pixel value of each pixel point in each image;
s3, establishing a signal gain matrix and a signal offset matrix which act on each pixel point in each image;
s4, establishing a linear equation set of corrected pixel values of each image according to the original pixel values, the signal gain matrix and the signal offset matrix;
s5, performing linear fitting based on a least square method to obtain each signal gain value in a signal gain matrix and each signal offset value in a signal offset matrix;
s6, calculating and obtaining a correction pixel value of each image according to the signal gain matrix, the signal offset matrix and the linear equation set which are obtained by fitting;
s7, carrying out flat field correction on each image according to the correction pixel value.
Further, in step S1, the camera captures a certain object to obtain a set of images for flat field correction, and two conditions need to be satisfied: firstly, the shot object needs to have uniform gray value in the whole image range; secondly, the exposure time of the camera needs to be adjusted so that the gray values of different images are different. In the multiple images, the number of row pixel points is the same and the number of column pixel points is the same for different images.
Further, in step S2, an original pixel value of each pixel in each image is obtained, and the original pixel values of each pixel in each image form an original pixel matrix.
For example, the number of the plurality of images is K, each image includes n (row) ×m (column) pixels, where the original pixel matrix of the 0 th image is:
wherein x is ij_0 Representing the original pixel value of the ith row and jth column of the 0 th image;
the original pixel matrix of the 1 st image is:
wherein x is ij_1 Representing the original pixel value of the ith row and jth column of the 1 st image;
the original pixel matrix for the K-1 th image is:
wherein x is ij_（K-1） Representing the original pixel values of the ith row and jth column of the K-1 th image, i=0, 1,2, … … n-1, j=0, 1,2, … … m-1.
Further, the pixel values of any position of an image should be equal everywhere, but are affected by different camera exposure at different moments, and the pixel value error of each image is relatively large, as shown in fig. 2. In order to eliminate the dark field and the field effect of the camera, a flat field correction is required to be performed on each image, each image is n (row) x m (column) pixel points, then a signal gain matrix a formed by n x m signal gain values and a signal offset matrix b formed by n x m signal offset values, which are subjected to flat field correction, of each pixel point are fitted by using the pixel values of the original K images, and then y=ax+b transformation is performed on all pixel values x of the image to obtain a corrected pixel value y, namely a flat field correction result. For all pixels of the same image, it is desirable that the corrected pixel values are substantially equal, i.e., in an ideal state, the same image has only 1 corrected pixel y value. For different images, it is desirable to correct with the same model, i.e. all images use the same set of signal gain matrix a and signal offset matrix b.
Specifically, in step S3, the number of the plurality of images is K, and each image includes n×m pixels; the signal gain matrix is:
the signal offset matrix is:
wherein i=0, 1,2, … … n-1, j=0, 1,2, … … m-1, a ij Signal gain value, b, representing pixel point of ith row and jth column in image ij And the signal offset value of the pixel point in the ith row and the jth column in the image is represented.
Specifically, in step S4, the linear equation of the pixel point in the ith row and the jth column of the kth image is:
；
wherein x is ij_k Representing the original pixel value, y, of the ith row and jth column pixel points in the kth image k A corrected pixel value representing the kth image;
the linear equation set comprises a linear equation of each pixel point of each image.
The linear system of equations for image 0 can be expressed as:
line 0:
；
line 1:
;
line (n-1):
；
the system of linear equations for image 1 can be expressed as:
line 0:
；/>
line 1:
；
line (n-1):
；
the linear system of equations for the k-1 th image can be expressed as:
line 0:
；
line 1:
；
line (n-1):
；
wherein:
The linear equations for each pixel of all images are listed as follows:
table 1: linear equation set for 0 th row and m column of k images
Table 2: flat field correction linear equation system for 1 st row and m column of k images
Table 3: flat field correction linear equation set for (n-1) th row and m column of k images
Further, as shown in fig. 3, in step S5, linear fitting is performed based on a least square method, to obtain each signal gain value in the signal gain matrix, including:
s501, subtracting the linear equations of two adjacent columns of pixel points of the same row in each image in the linear equation set to obtain a first equation set for eliminating correction pixel values;
s502, sequentially subtracting equations in the first equation set to obtain a second equation set for eliminating signal offset values;
s503, setting a signal gain value of a 0 th column in a second equation set, and converting the second equation set into a matrix form to obtain a first matrix;
s504, performing linear fitting by taking elements in the first matrix as coordinate points to obtain a first linear equation about a signal gain value to be solved;
s505, transpose the first matrix to obtain a first transpose matrix;
s506, according to the first transfer matrix, solving a signal gain value to be solved in the first linear equation.
Specifically, the signal gains are fitted first:
according to the linear equation set of tables 1-3, the first equation set for eliminating the corrected pixel value is obtained by first subtracting the 0 st column from the 1 st column by using the same signal offset b of the same column of different images and the same corrected pixel value y of the same image and different columns, as follows:
then subtracting equation (1) from equation (2), subtracting equation (2) from equation (3.) equation (k) subtracting equation (k-1), a second system of equations may be obtained:
from the above, it can be seen that a 00 、a 01 ...a 0（m-1） The values of (a) cannot be obtained by fitting all the above formulae, and a must be specified 00 The value a can be calculated in turn 01 ...a 0（m-1） Is a value of (2). Designation a 00 After the values, the second equation set is written in the form of a matrix, and a first matrix is obtained:
the handle is arranged on the side of the handle,
，
，...
viewed as coordinate points, these pointsObtaining a linear equation y=a about the signal gain value to be solved through linear fitting 01 x. Fitting a 01 The least squares method can be used: transpose the first matrix to obtain a first transpose matrix:
order the
，
The above can be written as a 01 X=Y，a 01 The expression of (2) is: a, a 01 =YX T (XX T ) -1 ；
Similarly, all adjacent two columns of the table 1 are subjected to the operation, namely, k equations are obtained by subtracting the two columns, then adjacent two rows of the equations are sequentially subtracted, and then a is obtained by using a least square fitting method 02 ，a 03 ，...a 0(m-1) . The same operation is performed on table 2, table 3. All n x m a values are finally solved to obtain each signal gain value in the signal gain matrix.
The range of the signal gain value of the 0 th column in the second equation set is (0, 1). Further, as shown in fig. 4, in step S5, linear fitting is performed based on a least square method, to obtain each signal offset value in the signal offset matrix, including:
s507, subtracting the linear equations of two adjacent columns of pixel points of the same row in each image in the linear equation set to obtain a first equation set for eliminating correction pixel values;
s508, designating a signal offset value of a 0 th column in the first equation set, and converting the first equation set into a matrix form to obtain a second matrix;
s509, performing linear fitting by taking elements in the second matrix as coordinate points to obtain a second linear equation about the signal offset value to be solved;
s510, transposing the second matrix to obtain a second transposed matrix;
s511, according to the second transposed matrix, solving a signal offset value to be solved in the second linear equation.
Specifically, after the signal gain matrix is obtained, the signal offset is fitted:
to find b 00 ，b 01 ...b 0(m-1) From the system of equations in table 1, subtracting the k equations for the 1 st column image from the k equations for the 0 th column image may result in a first system of equations:
and find a 00 , a 01 ...a 0(m-1) In the same way as above, a b must be specified first 00 Value b can be calculated sequentially 01 ...b 0(m-1) Is a value of (2). Designation b 00 After the value, the first equation set performs the term transfer:
wherein C is 00 , C 01 ... C 0(k-1) Are all constant.
Writing the first equation set into a matrix form to obtain a second matrix:
handle
，
...
Seen as coordinate points, which are linearly fitted to obtain a linear equation y=b for the signal gain value to be solved 01 x. Fitting b 01 The least squares method can be used: and transposing the second matrix to obtain a second transposed matrix:
order the
，
，b 01 Can pass through b 01 =YX T (XX T ) -1 And (5) solving.
And so on, subtracting the k equations of the jth image from the (j=1, 2..m-1) equation of the jth image from the (j-1) image to obtain a new equation set, and fitting by using a least square method to obtain b 02 ，b 03 ，...b 0(m-1) . Table 2, table 3. The same operation was performed and finally all n x m b values were solved.
The value of the signal offset value of the 0 th column in the first equation set is the average value of the original pixels of the 0 th image, and the product of the signal gain value of the 0 th column and the original pixel value of the 0 th row and the 0 th column in the second equation set is subtracted.
In some embodiments, the signal offset value of the 0 th column in the first equation set may be a preset correction pixel value of the 0 th image, which is subtracted from the product of the signal gain value of the 0 th column in the second equation set and the pixel value of the 0 th column and the 0 th row of the 0 th image.
In some application scenarios, there may be a requirement for the corrected pixel value of the image, for example, the pixel value after the correction of the 0 th image is required to be 200, and then the preset corrected pixel value may be set to be 200, and the product of the signal gain value of the 0 th column in the second equation set multiplied by the pixel value of the 0 th row and the 0 th column in the second equation set is subtracted from 200, so that the final calculated corrected pixel value is ensured to be about 200.
The original pixel value of each image is:
the corrected pixel values after calibration are:
the corrected pixel value is obtained according to the following formula:
further, performing flat field correction on each image according to the corrected pixel value, including:
the pixel value of each image is set as the calculated corrected pixel value.
The method provided in this embodiment is further described below through a specific application scenario.
Assume that a camera shoots 12 original images, and each image is composed of 1000 x 15552 pixels.
1. Acquiring 12X 1000X 15552 pixel values, and marking a pixel matrix of the ith picture as X i Then:
2. a linear system of equations is listed for all pixels to perform flat field correction, resulting in 12 x 1000 x 15552 equations, as shown in tables 4-6 below:
table 4: linear equation set of line 0, 15552, column for 12 images
Table 5: linear equation set of line 1, 15552 column for 12 images
Table 6: linear equation set of 999 th row 15552 column for k images
3. Fitting signal gain:
designation a 00 Taking the first 2 columns of 12 x 15552 equations of table 4, subtracting the 1 st column equation set from the 0 th column equation set, and subtracting the i-1 th row equation from the i-1 th row equation in turn (i=1, 2,..11), 11 equations can be obtained:
handle
，
...
Seen as coordinate points, these points are subjected to a linear fit, as shown in fig. 5, by the least squares method: writing the above equation set in the form of a matrix can result in:
the above can be written as a 01 X=Y，a 01 The expression of (2) is: a, a 01 =YX T (XX T ) -1 。
Other a were obtained by fitting in the same manner.
4. Fitting signal offset:
calculating the pixel mean value of the 0 th image as x avg Designation b 00 =x avg -a 00 *x 00_0 Then, taking the first 2 columns of 12 x 15552 equations of table 4, subtracting the 1 st column equation set from the 0 th column equation set, we can get 12 equations:
fitting a to 01 The method is the same, and the equation set is written in a matrix form, so that the following steps are obtained:
Other b values were obtained by fitting in the same manner.
Taking a and b of row 0 as an example, fig. 6 and 7 respectively show values of a and b corresponding to 15552 columns of row 0 of each image, wherein the abscissa shows the values of a and b respectively from row 0 to column 0 and the ordinate shows the values of b and the abscissa shows the values of a, and it can be seen that the values of a obtained by fitting are relatively large and the values of b obtained by fitting relatively small pixel points on row 0 of the same photo.
5. Performing flat field correction on an image:
the original pixel value of each image is
，
According to the formula:
the corrected pixel value of each point after the flat field correction can be obtained:
；
the image after flat field correction based on linear fitting has substantially uniform pixel values at each pixel point.
Referring to fig. 8, in some embodiments, there is also provided an image flat field correction device based on linear fitting, comprising:
an acquisition module 201, configured to acquire a plurality of images of a photographic subject;
an original pixel obtaining module 202, configured to obtain an original pixel value of each pixel point in each image;
a matrix creation module 203 for creating a signal gain matrix and a signal offset matrix for each pixel point in each image;
an equation building module 204 for building a linear equation set for the corrected pixel values for each image based on the raw pixel values, the signal gain matrix and the signal offset matrix;
a fitting module 205, configured to perform linear fitting based on a least square method, to obtain each signal gain value in the signal gain matrix and each signal offset value in the signal offset matrix;
a calculation module 206, configured to calculate a correction pixel value of each image according to the signal gain matrix, the signal offset matrix, and the linear equation set obtained by fitting;
a correction module 207 for performing a flat field correction on each image based on the correction pixel values.
The image flat field correction method and device based on linear fitting provided by the embodiment at least comprise the following beneficial effects:
(1) The built-in parameters of the camera are not required to be obtained or changed, so that the flat field correction can be performed only by obtaining a group of images shot by the camera, and the universality is strong;
(2) All parameters can be fitted through a group of pictures without additionally shooting dark field images by using a camera, and the efficiency is high.
(3) Only a least square method linear fitting method is used, the method is easy to understand and realize, and accumulated errors superposed by a plurality of methods are avoided, so that the image precision after flat field correction is high;
(4) And the flat field correction can be completed only at the algorithm level without additional experiment platforms and devices, the operation is convenient and simple, and the production cost is effectively reduced.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention. It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. An image flat field correction method based on linear fitting, comprising:
collecting a plurality of images of a photographed object;
acquiring an original pixel value of each pixel point in each image;
establishing a signal gain matrix and a signal offset matrix for each pixel point in each image;
establishing a linear equation set of corrected pixel values for each image according to the original pixel values, the signal gain matrix and the signal offset matrix;
performing linear fitting based on a least square method to obtain each signal gain value in a signal gain matrix and each signal offset value in a signal offset matrix;
calculating and obtaining a correction pixel value of each image according to the signal gain matrix, the signal offset matrix and the linear equation set which are obtained by fitting;
and carrying out flat field correction on each image according to the correction pixel value.
2. The method of claim 1, wherein the number of row pixels and the number of column pixels are the same for different ones of the plurality of images, and wherein the gray scale values are different between different ones of the plurality of images.
3. The method according to claim 1 or 2, wherein the number of the plurality of images is K, each image comprising n x m pixels; the signal gain matrix is:
the signal offset matrix is:
wherein i=0, 1,2, … … n-1, j=0, 1,2, … … m-1, a ij Signal gain value, b, representing pixel point of ith row and jth column in image ij And the signal offset value of the pixel point in the ith row and the jth column in the image is represented.
4. A method according to claim 3, wherein the linear equation for the ith row and jth column pixels of the kth image is:
wherein x is ij_k Representing the original pixel value, y, of the ith row and jth column pixel points in the kth image k Corrected pixel values representing the kth image, i=0, 1,2, … … n-1, j=0, 1,2, … … m-1, a ij Signal gain value, b, representing pixel point of ith row and jth column in image ij A signal offset value representing an ith row and jth column of pixels in the image;
the linear equation set comprises a linear equation of each pixel point of each image.
5. The method of claim 4, wherein obtaining each signal gain value in the signal gain matrix based on a linear fit by least squares method comprises:
subtracting the linear equations of two adjacent columns of pixel points in the same row in each image in the linear equation set to obtain a first equation set for eliminating correction pixel values;
sequentially subtracting the equations in the first equation set to obtain a second equation set for eliminating the signal offset value;
setting a signal gain value of a 0 th column in a second equation set, and converting the second equation set into a matrix form to obtain a first matrix;
performing linear fitting on elements in the first matrix as coordinate points to obtain a first linear equation about a signal gain value to be solved;
transpose the first matrix to obtain a first transpose matrix;
and solving a signal gain value to be solved in the first linear equation according to the first transfer matrix.
6. The method of claim 5, wherein the signal gain value in column 0 of the second set of equations is in the range of (0, 1).
7. The method of claim 5, wherein obtaining each signal offset value in the signal offset matrix based on a linear fit based on a least squares method comprises:
subtracting the linear equations of two adjacent columns of pixel points in the same row in each image in the linear equation set to obtain a first equation set for eliminating correction pixel values;
designating a signal offset value of a 0 th column in the first equation set, and converting the first equation set into a matrix form to obtain a second matrix;
performing linear fitting on elements in the second matrix as coordinate points to obtain a second linear equation about the signal offset value to be solved;
transposing the second matrix to obtain a second transposed matrix;
and solving a signal offset value to be solved in the second linear equation according to the second transposed matrix.
8. The method of claim 7, wherein the value of the signal offset value of column 0 in the first system of equations is the average value of the original pixels of the 0 th image, and the product of the signal gain value of column 0 and the original pixel value of column 0 of the 0 th image in the second system of equations is subtracted; or,
the value of the signal offset value of the 0 th row in the first equation set is the preset correction pixel value of the 0 th image, and the product of the signal gain value of the 0 th row and the original pixel value of the 0 th row in the second equation set are subtracted.
9. The method of claim 1, wherein performing a flat field correction on each image based on the corrected pixel values comprises:
the pixel value of each image is set as the calculated corrected pixel value.
10. An image flat field correction device based on linear fitting, comprising:
the acquisition module is used for acquiring a plurality of images of the shot object;
the original pixel acquisition module is used for acquiring an original pixel value of each pixel point in each image;
a matrix establishing module for establishing a signal gain matrix and a signal offset matrix about each pixel point in each image;
the equation building module is used for building a linear equation set of corrected pixel values of each image according to the original pixel values, the signal gain matrix and the signal offset matrix;
the fitting module is used for carrying out linear fitting based on a least square method to obtain each signal gain value in the signal gain matrix and each signal offset value in the signal offset matrix;
the calculating module is used for calculating and obtaining a correction pixel value of each image according to the signal gain matrix, the signal offset matrix and the linear equation set which are obtained by fitting;
and the correction module is used for carrying out flat field correction on each image according to the correction pixel value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310160209.3A CN116029938B (en) | 2023-02-24 | 2023-02-24 | Image flat field correction method and device based on linear fitting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310160209.3A CN116029938B (en) | 2023-02-24 | 2023-02-24 | Image flat field correction method and device based on linear fitting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116029938A true CN116029938A (en) | 2023-04-28 |
| CN116029938B CN116029938B (en) | 2023-06-23 |
Family
ID=86081429
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310160209.3A Active CN116029938B (en) | 2023-02-24 | 2023-02-24 | Image flat field correction method and device based on linear fitting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116029938B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116709046A (en) * | 2023-07-03 | 2023-09-05 | 深圳市度申科技有限公司 | Fixed pattern noise calculation and compensation method |
| CN116777999A (en) * | 2023-06-28 | 2023-09-19 | 深圳市度申科技有限公司 | Multi-adaptability high-level flat field correction method for area array camera |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1469312A (en) * | 2002-05-30 | 2004-01-21 | ������������ʽ���� | Apparatus and method for correcting image data obtained from image detector |
| US20070177032A1 (en) * | 2006-01-27 | 2007-08-02 | Nethra Imaging | Automatic color calibration of an image sensor |
| CN101442608A (en) * | 2008-12-31 | 2009-05-27 | 中国资源卫星应用中心 | Method for improving relative radiation correction of CCD camera |
| CN102521797A (en) * | 2011-11-04 | 2012-06-27 | 华中科技大学 | Scene non-uniform correction method for scanning type infrared imaging system |
| US20130182934A1 (en) * | 2012-01-13 | 2013-07-18 | Carestream Health, Inc. | Self correcting portable digital radiography detector, methods and systems for same |
| CN105929340A (en) * | 2016-06-30 | 2016-09-07 | 四川普力科技有限公司 | Method for estimating battery SOC based on ARIMA |
| CN108986170A (en) * | 2018-07-03 | 2018-12-11 | 武汉精测电子集团股份有限公司 | A kind of line-scan digital camera method for correcting flat field suitable for field working conditions |
| CN109660736A (en) * | 2017-10-10 | 2019-04-19 | 凌云光技术集团有限责任公司 | Method for correcting flat field and device, image authentication method and device |
| CN109767403A (en) * | 2019-01-23 | 2019-05-17 | 西安科技大学 | A kind of infrared focal plane array strip noise cancellation method based on scene |
| CN113808046A (en) * | 2021-09-18 | 2021-12-17 | 凌云光技术股份有限公司 | Method and device for acquiring flat field correction parameters |
| CN115361494A (en) * | 2022-10-09 | 2022-11-18 | 浙江双元科技股份有限公司 | High-speed stroboscopic image processing system and method of multi-line scanning image sensor |
| CN115665579A (en) * | 2022-10-17 | 2023-01-31 | 长春长光辰芯光电技术有限公司 | Optical response nonuniformity correction method on COMS image sensor |
-
2023
- 2023-02-24 CN CN202310160209.3A patent/CN116029938B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1469312A (en) * | 2002-05-30 | 2004-01-21 | ������������ʽ���� | Apparatus and method for correcting image data obtained from image detector |
| US20070177032A1 (en) * | 2006-01-27 | 2007-08-02 | Nethra Imaging | Automatic color calibration of an image sensor |
| CN101442608A (en) * | 2008-12-31 | 2009-05-27 | 中国资源卫星应用中心 | Method for improving relative radiation correction of CCD camera |
| CN102521797A (en) * | 2011-11-04 | 2012-06-27 | 华中科技大学 | Scene non-uniform correction method for scanning type infrared imaging system |
| US20130182934A1 (en) * | 2012-01-13 | 2013-07-18 | Carestream Health, Inc. | Self correcting portable digital radiography detector, methods and systems for same |
| CN105929340A (en) * | 2016-06-30 | 2016-09-07 | 四川普力科技有限公司 | Method for estimating battery SOC based on ARIMA |
| CN109660736A (en) * | 2017-10-10 | 2019-04-19 | 凌云光技术集团有限责任公司 | Method for correcting flat field and device, image authentication method and device |
| CN108986170A (en) * | 2018-07-03 | 2018-12-11 | 武汉精测电子集团股份有限公司 | A kind of line-scan digital camera method for correcting flat field suitable for field working conditions |
| CN109767403A (en) * | 2019-01-23 | 2019-05-17 | 西安科技大学 | A kind of infrared focal plane array strip noise cancellation method based on scene |
| CN113808046A (en) * | 2021-09-18 | 2021-12-17 | 凌云光技术股份有限公司 | Method and device for acquiring flat field correction parameters |
| CN115361494A (en) * | 2022-10-09 | 2022-11-18 | 浙江双元科技股份有限公司 | High-speed stroboscopic image processing system and method of multi-line scanning image sensor |
| CN115665579A (en) * | 2022-10-17 | 2023-01-31 | 长春长光辰芯光电技术有限公司 | Optical response nonuniformity correction method on COMS image sensor |
Non-Patent Citations (2)
| Title |
|---|
| KUANGMING YAP 等: "Gain Error Correction for CMOS Image Sensor Using Delta-Sigma Modulation", 《IEEE》 * |
| 闫苏琪 等: "基于最小二乘拟合的科学级CMOS传感器非均匀性校正", 《电子设计工程》, pages 151 - 155 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116777999A (en) * | 2023-06-28 | 2023-09-19 | 深圳市度申科技有限公司 | Multi-adaptability high-level flat field correction method for area array camera |
| CN116709046A (en) * | 2023-07-03 | 2023-09-05 | 深圳市度申科技有限公司 | Fixed pattern noise calculation and compensation method |
| CN116709046B (en) * | 2023-07-03 | 2023-12-15 | 深圳市度申科技有限公司 | Fixed pattern noise calculation and compensation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116029938B (en) | 2023-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116029938B (en) | Image flat field correction method and device based on linear fitting | |
| CN107071234B (en) | Lens shadow correction method and device | |
| US8659685B2 (en) | Method and apparatus for calibrating and correcting shading non-uniformity of camera systems | |
| CN108871588A (en) | A kind of infrared imaging system various dimensions joint asymmetric correction method | |
| CN110738707A (en) | Distortion correction method, device, equipment and storage medium for cameras | |
| CN108495101B (en) | A kind of method for correcting image, device, image capture device and readable storage medium storing program for executing | |
| CN110533618B (en) | Lens distortion correction method and photographic device | |
| TWI638149B (en) | Luminance measurement method, luminance measurement device, and image quality adjustment technique using the method and device | |
| US20180063454A1 (en) | Method and apparatus for correcting fixed pattern noise of an infrared image | |
| CN110501025B (en) | On-orbit calibration method of rolling shutter area array CMOS camera | |
| CN110766615B (en) | Picture correction method, device, terminal and computer readable storage medium | |
| GB2442050A (en) | Image pixel value correction | |
| CN107403410B (en) | Splicing method of thermal infrared images | |
| US9143756B2 (en) | Camera module, photographing method, and electronic apparatus | |
| CN105869129B (en) | For the thermal infrared images residue non-uniform noise minimizing technology after nonuniformity correction | |
| US20080278613A1 (en) | Methods, apparatuses and systems providing pixel value adjustment for images produced with varying focal length lenses | |
| CN112489137A (en) | RGBD camera calibration method and system | |
| CN111652815B (en) | Mask plate camera image restoration method based on deep learning | |
| US8542919B2 (en) | Method and system for correcting lens shading | |
| JP2012204856A (en) | Image processing method and camera module | |
| CN113808046B (en) | Flat field correction parameter acquisition method and device | |
| US20100110241A1 (en) | Multi illuminant shading correction using singular value decomposition | |
| US8331722B2 (en) | Methods, apparatuses and systems providing pixel value adjustment for images produced by a camera having multiple optical states | |
| CN107203150B (en) | Non-uniformity correction method based on infrared semi-physical simulation system | |
| CN110536066B (en) | Panoramic camera shooting method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |