CN110738608A - A kind of plane image correction method and system - Google Patents

Abstract

The present invention provides a plane image correction method and system. The system includes at least an image acquisition module for acquiring original distorted image data; a geometric distortion correction module for performing geometric distortion correction on the original distorted image and obtaining size Correcting the image; an intensity correction module for performing illumination intensity correction on the size-corrected image; and an image output module for outputting the final corrected image. The invention effectively corrects the geometric distortion of the image caused by the rotation of the object in the three dimensional directions, and the problem that the pixel corresponds to the actual size of the non-uniform size, and also solves the image distortion caused by the infrared optical system and the non-uniformity of the illumination intensity. problem, the method is convenient, and the resource consumption is low.

Description

A kind of plane image correction method and system

technical field

The invention relates to the field of computer image processing, belongs to the field of plane image correction subdivision, and in particular relates to a plane image correction method and system based on projection geometric distortion and illumination inhomogeneity.

Background technique

In the field of infrared thermal imaging nondestructive testing, it is often impossible to image in one image because the object to be photographed is too large or the field of view is limited, or we want to obtain high-resolution images with more details. The object is divided into multiple areas and imaged separately, and finally these images are combined into a complete image by jigsaw puzzles. But without proper correction of the individual images, the mosaic result may be inaccurate. This is because the image captured by the thermal imager follows the rules of projection imaging, which does not guarantee the orthogonality and geometric linearity of the plane, which may cause geometric distortion of the projected image. In addition, in order to obtain more radiant energy, the infrared optical system is generally designed as a large relative aperture system, which leads to radial distortion of the infrared image. Finally, since the imaging surface usually has illumination with uneven intensity distribution, each generated image will be superimposed with this illumination intensity variation, resulting in an uneven brightness distribution of the image. In addition, due to the different distances between the thermal imager and the object to be inspected when taking pictures in different regions, the actual sizes corresponding to the images of each region are not uniform. The above problems are usually acceptable in single-image imaging, but when overlapping images in multiple regions In jigsaw synthesis, due to the existence of these problems, the puzzles cannot be accurately matched and the intensity is not uniform. However, in many scientific and engineering applications, we often need to make precise mosaics and make the flattened image look the same as if it were imaged with a single exposure under uniform illumination.

In addition, in some special usage scenarios, such as infrared optical systems, the design of infrared optical systems usually requires a large field of view, which inevitably results in geometric distortion of the image. There are two types of radial and tangential, usually radial distortion. Much larger than the tangential distortion. And because the active excitation light source is usually used in infrared nondestructive testing. For example, if a single light source is used to point at the center of the surface of the object under inspection, each generated image will be superimposed with this illumination intensity variation, resulting in a non-uniform brightness distribution of the image.

Based on the above existing problems, there is not yet an effective method or system for image distortion stitching and image correction under non-uniform illumination at the same time.

SUMMARY OF THE INVENTION

In view of the deficiencies of the prior art, the present invention provides a method and system for correcting a plane image, which can effectively eliminate the distortion of the optical system and process the image under non-uniform illumination. Specifically, the present invention provides the following technical solutions:

In one aspect, the present invention provides a plane image correction method, the method comprising:

S1. Acquire basic data of the image acquisition device, measure the distance y _c from the center of the object plane to the lens, and acquire original distorted image data, and establish an original image coordinate system with the center of the original distorted image as the origin;

S2, establish the coordinate system X-Z of the corrected image, and take the center of the corrected image as the origin;

S3. Correct the geometric distortion caused by the rotation of the X-Z plane around the z-axis and the x-axis; correct the geometric distortion caused by the rotation of the X-Z plane around the y-axis; establish the point (X, Z) on the actual plane to the physical space point (x, y, z) the relationship between;

S4. Establish a mapping from physical space points (x, y, z) to points (ξ', η') of the original distorted image;

S5. Determine the pixel point of the initial distorted image corresponding to the point (ξ', η') of the original distorted image, and use the difference algorithm to determine the position of the point in the corrected image according to the coordinates and pixel values of the adjacent points of the pixel point. pixel value; based on the y _c , by specifying the pixel size Δx of the corrected image, the image is corrected to the required size, and the size-corrected image is obtained;

S6, obtain the 2D illumination distribution estimation of the area to be corrected by least square fitting on the image after size correction, and determine the original non-uniform illumination intensity p(ξ, η);

S7. Calculate the equalized intensity pave based on the p( _ξ , η); based on the original non-uniform illumination intensity and equalized intensity, correct the intensity of each point.

Preferably, the basic data of the image acquisition device includes the size, length, width of the focal plane, and the horizontal field of view of the lens.

In addition, it should be pointed out that in this method, the steps of correcting the geometric distortion of the image and the correction of the light intensity can be adjusted before and after, that is, the light intensity of the image can be corrected first, and then the geometric distortion of the image can be corrected. The above-mentioned steps S3-S5 and steps S6-S7 can be sequentially exchanged, and the methods after the exchange should all be deemed to fall within the protection scope of the claims of the present invention.

Preferably, in the step S2, the corrected image and the original distorted image use the same image size, and more preferably, the length and width are the same as the focal plane.

Preferably, the relationship between the point (X, Z) on the actual plane and the pixel (I, J) in the corrected image is expressed as: then I=X/Δx, J=Z/Δx, where Δx is the pixel size; the relationship between the pixel (i, j) in the original distorted image and the point (ξ, η) in the original distorted image is expressed as: i=ξ/δ, and j=η/δ, where δ is the original distortion The size of each cell in the image.

Preferably, in S3, for the geometric distortion rotated around the z-axis, find the closest point ξ ₀ from the x-axis to the image acquisition device in the original distorted image plane XZ, and determine the XZ plane around the z-axis based on ξ ₀ and y _c Rotation angle a _z ; determine the three-dimensional space coordinates (x, y, z) of the rotated point (X, Z) based on y _c and a _z ;

For the geometric distortion rotated around the x-axis, find the closest point η ₀ from the z-axis to the image acquisition device in the original distorted image plane XZ, and determine the rotation angle a _x of the XZ plane around the x-axis based on η ₀ and _yc .

Preferably, the relationship between point (X', Z') and three-dimensional space coordinates (x, y, z) is established:

Preferably, in the S3, for the geometric distortion rotated around the y-axis, the correction is performed in the following manner:

where a _y is the rotation angle around the y-axis.

Preferably, in the S4, the mapping relationship between the physical space point (x, y, z) to the point (ξ', η') of the original distorted image is:

ξ′=δαx/rΔ

η′=δαz/rΔ

Wherein, r=(x ² +z ² ) ^1/2 , α=tan ⁻¹ (r/y), and Δ=tan ⁻¹ (δ/y _c ).

Preferably, in the S6, the least squares fitting adopts the following manner:

Where M is the order of the polynomial, 1 and k are the subscripts of the coefficients in the above summation polynomial and the corresponding power changes of the two variables, k increases from 0 to M, and 1 increases from 0 to k.

Preferably, in the S7, the intensity of each point is corrected in the following manner:

I _new (ξ, η) = I _old (ξ, η) p _ave /p(ξ, η)

where I _new (ξ, η) and I _old (ξ, η) are the corrected image and original image intensities at point (ξ, η), respectively.

On the other hand, the present invention also provides a plane image correction system, the system includes:

The image acquisition module is used to collect the original distorted image data;

a geometric distortion correction module for performing geometric distortion correction on the original distorted image and obtaining a size corrected image;

an intensity correction module for performing illumination intensity correction on the size-corrected image;

Image output module for outputting the final corrected image;

The geometric distortion correction module at least includes: an x-axis rotation correction unit, a y-axis rotation correction unit, a z-axis rotation correction unit, a basic data acquisition unit, and a size correction unit;

The basic data acquisition unit is used to acquire the basic data of the image acquisition module, measure the distance y _c from the center of the object plane to the lens, acquire the original distorted image data from the image acquisition module, and establish the original distorted image center as the origin. image coordinate system;

The x-axis rotation correction unit, the y-axis rotation correction unit, and the z-axis rotation correction unit are respectively used to correct the geometric distortion caused by rotation around the x, y, and z axes;

the size correction unit, configured to correct the image to a required size based on the correction results of the x-axis rotation correction unit, the y-axis rotation correction unit, and the z-axis rotation correction unit, and the pixel size Δx of the specified correction image, Obtain a size-corrected image.

Preferably, the intensity correction module includes at least: a two-dimensional illumination distribution estimation unit and an intensity correction unit;

The two-dimensional illumination distribution estimation unit is used to obtain a 2D illumination distribution estimation of the area to be corrected through least squares fitting, and determine the original non-uniform illumination intensity;

The intensity correction unit is configured to correct the intensity of each point based on the original non-uniform illumination intensity and the balanced intensity.

Preferably, the relationship between the point (X, Z) on the actual plane and the pixel (I, J) in the corrected image is expressed as: then I=X/Δx, J=Z/Δx, where Δx is the pixel size; the relationship between the pixel (i, j) in the original distorted image and the point (ξ, η) in the original distorted image is expressed as: i=ξ/δ, and j=η/δ, where δ is the original distortion The size of each cell in the image.

Preferably, the z-axis rotation correction unit finds the closest point ξ ₀ from the x-axis to the image acquisition device in the original distorted image plane XZ for the geometric distortion rotated around the z-axis, and determines the XZ plane around the XZ plane based on ξ ₀ and y _c The rotation angle a _z of the z-axis; the three-dimensional space coordinates (x, y, z) of the rotated point (X, Z) are determined based on y _c and a _z ;

Preferably, the x-axis rotation correction unit, for the geometric distortion rotated around the x-axis, finds the closest point η ₀ from the z-axis to the image acquisition device in the original distorted image plane XZ, and determines the XZ plane around the x-axis based on η ₀ and y _c the rotation angle a _x .

Preferably, the relationship between point (X', Z') and three-dimensional space coordinates (x, y, z) is established:

Preferably, the y-axis rotation correction unit corrects the geometric distortion rotated around the y-axis according to the following methods:

where a _y is the rotation angle around the y-axis.

Preferably, when the size correction unit performs distorted image correction, the mapping relationship between the physical space point (x, y, z) used to the point (ξ', η') of the original distorted image is:

ξ′=δαx/rΔ

η′=δαz/rΔ

Wherein, r=(x ² +z ² ) ^1/2 , α=tan ⁻¹ (r/y), and Δ=tan ⁻¹ (δ/y _c ).

Preferably, in the two-dimensional illumination distribution estimation unit, the least squares fitting adopts the following method:

where M is the order of the polynomial, and 1 and k are the subscripts of the coefficients in the above summation polynomial and the corresponding power changes of the two variables.

Preferably, in the intensity correction unit, the intensity of each point is corrected in the following manner:

I _new (ξ, η) = I _old (ξ, η) p _ave /p(ξ, η)

where I _new (ξ, η) and I _old (ξ, η) are the corrected image and original image intensities at point (ξ, η), respectively.

Compared with the prior art, the technical solution of the present invention has the following advantages:

It effectively corrects the geometric distortion of the image caused by the rotation of the object in three dimensions, and effectively adjusts the problem that the actual size of the pixels is not uniform due to the different shooting distances. At the same time, the image distortion caused by the infrared optical system and The non-uniform brightness distribution of the image caused by the non-uniformity of the illumination intensity has been effectively corrected, so that the image will not have problems such as distortion or uniformity after splicing or synthesizing.

Description of drawings

FIG. 1 is a schematic diagram of projection imaging according to an embodiment of the present invention;

2 is a comparison between an image before correction and an image after correction according to an embodiment of the present invention;

3 is a schematic diagram of an accurate mosaic after geometric distortion correction according to an embodiment of the present invention;

Fig. 4 is the x-y plane view after the plane X-Z of the embodiment of the present invention is rotated around the Z axis (the z axis and the Z axis both point to the reader);

FIG. 5 is an imaging plane after being rotated by an angle around the z-axis according to an embodiment of the present invention;

FIG. 6 is an imaging plane after being rotated by an angle around the x-axis according to an embodiment of the present invention;

FIG. 7 is an imaging plane after being rotated by an angle around the y-axis according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of correcting geometric distortion of an image according to an embodiment of the present invention;

9 is a schematic diagram of projection distortion correction according to an embodiment of the present invention;

FIG. 10 is the pixel intensity distribution of a straight line passing through the center of the grayscale image along the straight line according to an embodiment of the present invention;

FIG. 11 is an updated intensity curve according to an embodiment of the present invention.

specific embodiment

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Example 1

In this embodiment, in order to better reflect the idea of the present invention, a camera, especially an infrared camera, is taken as an example to illustrate the specific implementation of the method. It should be noted that although the infrared camera is taken as an example below, the application of the technical solution of the present invention It is not limited to the sub-field of infrared cameras, and should not be construed as a limitation on the protection scope of the present invention.

The imaging of cameras and other equipment follows the rules of projection imaging. In an ideal camera system, in the absence of errors (such as lens aberrations) introduced by various components, combined with Figure 1, the imaging process is in three-dimensional (3D) conditions. ) In the Cartesian coordinate system xyz (camera coordinate system), xyz is considered as the physical space for imaging, and the camera is located at the origin o(0,0,0) and points to the positive y-axis direction, also known as the camera axis. Therefore, the imaging plane (ξ, η) within the camera is perpendicular to the y-axis, and its distance from the origin is _yc , and the intersection of the imaging plane and the y-axis is the center point of the imaging plane. Due to the particularity of the formula used in projection imaging, y _c can take any value without affecting the result. Later, it is assumed that the center point of the imaging plane and the actual plane coincide, that is, it is regarded as the distance between the center point of the actual plane and the camera, As shown in Figure 1, a certain point (x, y, z) in the physical space is projected to a point (ξ, η) in the image plane, according to the projected geometric relationship:

ξ=xy _c /y (1)

η=zy _c /y (2)

Generalizes to all points (x, y, z) on a plane in physical space. Then, when "all points" in the plane are projected onto the image plane based on equations (1) and (2) above, we get an image of that plane. In real space, an object usually has six degrees of freedom of motion: three translational motions along (x, y, and z) spatial directions and three rotational motions around (x, y, and z) spatial directions. For the plane XZ, if we place it in the coordinate system shown in Figure 1, its center is on the y-axis, and its X-axis and Z-axis are parallel to the x-axis and z-axis, respectively (i.e. the XZ plane is perpendicular to the y-axis), which leaves four degrees of freedom of motion for its position in three-dimensional space: its distance from the camera and three other degrees of rotational freedom. It should be noted that the coordinate system XZ is attached to the plane, so it moves with the image surface (but its origin (X=0, Z=0) is fixed); while the coordinate systems xyz and ξ-η are in space It is fixed. In the following we assume that the plane XZ and the image plane ξ-η initially coincide with each other (ie X=ξ and Z=η). Therefore, the y _c value determines the distance between the plane XZ and the camera, and before any rotation, the 3D space coordinates of the point (X,Z) are:

x=X (3)

y=y _c (4)

z=Z (5)

When the X-Z plane changes in four degrees of freedom and its image is imaged in the camera, we observe:

(1) Rotation around the z-axis will cause linear geometric distortion of the image;

(2) Similarly, rotation around the x-axis also results in linear geometric distortion of the image;

(3) Rotation around the y-axis produces an image with image rotation but no geometric distortion;

(4) The distance change along the y-axis only changes the image size.

In addition, since the design of infrared optical system usually requires a large field of view of the system, it will inevitably cause geometric distortion of the image, including radial and tangential. Usually, the radial distortion is much larger than the tangential distortion. The technical solutions given in the embodiments are especially suitable for eliminating the barrel distortion of the optical system. Of course, the method can also be applied to other types of image distortions, including error distortions introduced by conventional devices. In addition, since an active excitation light source is usually used in infrared nondestructive testing, for example, if a single light source is used to point to the center of the surface of the inspected object, each generated image will be superimposed with the illumination intensity variation, resulting in uneven brightness distribution of the image.

In a specific embodiment, in the method proposed by the present invention, the plane imaging involves the change of 3 angular degrees of freedom in the three-dimensional direction, and the change of 1 distance degree of freedom perpendicular to the imaging plane. These 4 degrees of freedom The changes of , may cause imaging errors. Since the present invention only considers plane imaging, once the three angles and distances are determined, the dimensions in the x and z directions are uniquely determined. The problems addressed by the technical solution proposed in the present invention include the following four technical problems: one is to correct the geometric distortion of the image caused by the rotation of the object in three dimensions; The problem that the pixel corresponds to the actual size is not uniform. The third is to correct the image distortion caused by the infrared optical system. The fourth is to solve the problem of uneven brightness distribution of images caused by the unevenness of illumination intensity.

Before discussing image geometric distortion correction methods, we first define the conditions necessary for seamless stitching of multiple images. First, the corrected image should maintain the orthogonality of the straight lines on the plane. Second, the line length in the corrected image should be linearly proportional to the line length on the physical surface. Figure 2 schematically shows the format of a square flat image before correction and the format after correction. The left image is the actual imaging plane, that is, the image before correction, including the geometric distortion of the image caused by changes in four degrees of freedom. As well as the geometric distortion caused by the optical system, the coordinate system is the ξ-η coordinate system with the center of the graph as the origin. The image on the right is a 2D linear orthographic image of the actual square plate represented by the coordinate system X-Z, i.e. the corrected image. This format will eventually allow exact geometrically matched stitching of overlapping images, as schematically shown in Figure 3. The only thing missing from Figure 2 is the grayscale (or color) distribution, which can be obtained from the original image (left image in Figure 2) once the relationship between the points (X, Z) and (ξ, η) has been determined Gray value distribution. Therefore, the following discussion is devoted to establishing the mapping relationship between the points of the image before correction (input image) and the image after correction (output image). In view of this problem, in a specific embodiment, the method adopted in the present invention is as follows:

A correction method for image geometric distortion, the correction method comprises the following steps:

1) Obtain the thermal imager focal plane size, length Imax, and width Jmax; and the horizontal field of view angle α of the lens, then the vertical field of view angle can be obtained as α*Jmax/Imax;

2) Measure the distance y _c from the center of the object plane to the lens of the thermal imager;

3) Obtain the original distorted image data, and establish the original image coordinate system ξ-η with the center of the original image as the origin, the abscissa as the ξ axis, and the ordinate as the η axis;

4) Establish the coordinate system X-Z of the corrected image, with the center of the image as the origin. For simplicity, the corrected image adopts the same image size as the original image, that is, the image dimension with length Imax and width Jmax;

5) We denote the pixel in the final corrected image as (I, J), then for a certain point (X, Z), if the given pixel size is Δx, then I=X/Δx, J=Z/Δx ;

6) Denote the pixel in the original image as (i, j), corresponding to the point (ξ, η) in the original image, then i=ξ/δ, andj=η/δ, δ is each pixel in the original image The size of , is determined by the following formula:

δ=2y _c tan ^-1 (α/2)/Imax (6)

7) Correct the geometric deformation caused by the rotation of the XZ plane around the z-axis. Figure 4 shows the geometric relationship of the plane XZ on the xy-plane (at any z) after it is rotated by an angle a _z around the Z-axis. From this figure we can easily get the 3D space coordinates of the rotated point (X,Z):

x=X cosaz (7)

y=y _c +X sinaz (8)

z=Z (9)

Equation (9) shows that rotation about the Z axis has no effect on the z value.

In Figure 4, we also notice a special point in the image plane,

ξ ₀ =y _c tanaz (10)

It corresponds to the closest point from the plane XZ to the camera. This point _ξ0 is usually easy to find from the image, see Figure 5.

For a small angle of rotation, any surface line parallel to the x-axis will become a curve. For the line connecting the relative peak points AB of the two horizontal curves in the figure, the intersection with the x-axis is ξ ₀ ; using formula (10 ) can calculate the rotation angle a _z of the plane around the z-axis;

8) Correct the geometric deformation caused by the rotation of the X-Z plane around the x-axis, see Figure 6;

Similar to equation (5), rotation about the X axis does not change the value of x, so equations (7) and (10) are still valid. Find the point η ₀ on the z-axis closest to the camera: for a small angle of rotation, any surface straight line parallel to the z-axis becomes a curve, for the line connecting the relative peak points CD of the two perpendicular curves in the figure, The intersection with the z-axis is η ₀ ;

9) The rotation angle ax of the plane around the x-axis can be calculated by using the formula η ₀ =y _c tanax;

10) The geometric distortion caused by the rotation of the X-Z plane around the x-axis and the z-axis can be corrected according to the following formula;

Thereby, the mapping relationship between point (X', Z') and physical space point (x, y, z) is obtained.

11) Correct the image rotation caused by the rotation of the X-Z plane around the y-axis, see Figure 7;

12) The geometric distortion caused by rotation around the y-axis can be corrected according to the following formula:

In a preferred embodiment, it can be performed as follows: in the first execution, a _y =0 can be set to obtain a corrected image, and then a _y is estimated according to the deflection angle of the image, and program correction is performed again.

A relationship is now established between a plane point (X, Z) to a point (X', Z') to a physical space point (x, y, z).

It can be seen from the above process that the point (X', Z') is equivalent to an intermediate result here, and the formula (11) (12) establishes a plane point (X, Z) to point (X', Y') and then to The mapping relationship between physical space points (x, y, z), if there is no distortion around the y-axis shown in equation (12), then the point (X', Y') is the point (X, Z) .

13) Correct the geometric distortion of the image, as shown in Figure 8;

14) Since the camera detector is a point detector relative to the actual plane X-Z, if the imaging plane is a spherical surface centered on the origin, the imaging image will be distorted, as shown in Figure 9. Then the point on the imaging plane corresponding to the point (x, y, z) on the actual plane should be a new point (ξ', η'). This new projection formula is:

ξ′=δαx/rΔ (13)

η′=δαz/rΔ (14)

here,

r=(x ² +z ² ) ^1/2 (15)

α=tan ^-1 (r/y) (16)

Δ=tan ^-1 (δ/y _c ) (17)

From the above formula, we can finally get the mapping from the point (X, Z) on the actual plane to the point (ξ', η') on the imaging plane, that is, the mapping to the point on the original distorted image.

In the schematic diagram of projection distortion correction shown in FIG. 9, an intermediate curved surface is introduced to determine a new projection point ξ' in the xy plane, where α _x =αx/r.

15) For each pixel point (I, J) in the corrected image, input the pixel size Δx of the specified corrected image, and the corresponding point (X, Z) can be found according to step 5);

16) According to the formula (12) in step 12) and the formula (11) in step 10), the point (x, y, z) mapped to the physical space can be obtained, and then according to the formula (13)-( 17) The point (ξ', η') that can be mapped to the original distorted image, that is, the relationship between (X, Z) and (ξ', η') is finally established;

17) According to step 6), the pixel point of the initial distorted image corresponding to the point (ξ', η') can be found. , J) pixel value of the point.

18) Correct image size: According to the measured value y _c , by specifying the pixel size Δx of the corrected image, the image size is automatically corrected to the required size through the above steps.

19) Correct image brightness non-uniformity;

In many cases, artificial lighting sources are used in photography or photography. For example, if a single light source is used to point at the center of the surface, the central part of the image will show higher intensity (or brightness), which is often accompanied by higher contrast (grayscale). If we draw a line exactly through the center of the grayscale image, the pixel intensity distribution along this line will look like the solid line in Figure 10. While the variation in pixel intensity is related to the scene on the image, the smooth dashed line in Figure 10 is directly related to illumination non-uniformity. Since this inhomogeneity is an artifact superimposed on the image, it should be removed from the image and only the image content displayed.

20) Select one or more rectangular areas from the image for intensity equalization, pay attention to avoid certain areas such as defective areas, frames, and interesting scenes that do not need to be corrected, and correct them according to the following algorithm;

21) It is first necessary to obtain an estimate of the 2D illumination distribution on the surface of the area to be corrected. As shown in Figure 10, the smooth dashed line (solid curved boundary centerline) is a good approximation for the distribution of illumination intensity in one dimension. For a two-dimensional image, the median surface of the image intensity distribution can be obtained by least squares fitting of a two-dimensional polynomial function. This two-dimensional polynomial function is expressed as:

Where M is the order of the polynomial, l and k are the subscripts of the coefficients in the above summation polynomial and the corresponding power changes of the two variables, k increases from 0 to M, and l increases from 0 to k. By using the classical least squares fitting method, for example, the corresponding coefficient _akl for a fixed value M (eg M=4) can be easily obtained.

22) Once p(ξ, η) is determined, calculate the average value of p over the entire image, _pave . p _ave and p(ξ, η) denote the equalized intensity and the original non-uniform light intensity, respectively, we can use the following formula to correct the intensity of each point in the image:

I _new (ξ, η) = I _old (ξ, η) p _ave /p(ξ, η) (19)

where I _new (ξ, η) and I _old (ξ, η) are the corrected image and original image intensities (or gray levels) at point (ξ, η), respectively.

23) Applying formula (19) to the pixel intensity distribution curve in Figure 10, will obtain a new intensity curve with more uniform mean intensity and contrast, as shown in Figure 11. Therefore, equation (19) will remove the effect of illumination non-uniformity in the image. Figure 11 modifies the intensity profile of Figure 10 along a straight line through the center of the image.

24) The intensity value I _old (ξ, η) of the effective pixel point in the original image is processed through the above steps to obtain a corresponding new corrected image intensity (or gray scale) I _new (ξ, η).

Example 2

In yet another embodiment, the present invention also provides a plane image correction system, which can execute the method as described in Embodiment 1, and the structure of the system is only used as a preferred structure setting method, which is skilled in the art When the main function of the system structure after routine adjustment by personnel according to the technical solutions disclosed in the present invention is the same as that of the sub-invented system, it should be regarded as falling within the protection scope of the present invention. Preferably, the system includes:

The image acquisition module is used to collect the original distorted image data;

a geometric distortion correction module for performing geometric distortion correction on the original distorted image and obtaining a size corrected image;

an intensity correction module for performing illumination intensity correction on the size-corrected image;

Image output module for outputting the final corrected image;

The geometric distortion correction module at least includes: an x-axis rotation correction unit, a y-axis rotation correction unit, a z-axis rotation correction unit, a basic data acquisition unit, and a size correction unit;

The basic data acquisition unit is used to acquire the basic data of the image acquisition module, measure the distance y _c from the center of the object plane to the lens, acquire the original distorted image data from the image acquisition module, and establish the original distorted image center as the origin. image coordinate system;

The x-axis rotation correction unit, the y-axis rotation correction unit, and the z-axis rotation correction unit are respectively used to correct the geometric distortion caused by rotation around the x, y, and z axes;

the size correction unit, configured to correct the image to a required size based on the correction results of the x-axis rotation correction unit, the y-axis rotation correction unit, and the z-axis rotation correction unit, and the pixel size Δx of the specified correction image, Obtain a size-corrected image.

Preferably, the intensity correction module includes at least: a two-dimensional illumination distribution estimation unit and an intensity correction unit;

The two-dimensional illumination distribution estimation unit is used to obtain a 2D illumination distribution estimation of the area to be corrected through least squares fitting, and determine the original non-uniform illumination intensity;

The intensity correction unit is configured to correct the intensity of each point based on the original non-uniform illumination intensity and the balanced intensity.

Preferably, the relationship between the point (X, Z) on the actual plane and the pixel (I, J) in the corrected image is expressed as: then I=X/Δx, J=Z/Δx, where Δx is the pixel size; the relationship between the pixel (i, j) in the original distorted image and the point (ξ, η) in the original distorted image is expressed as: i=ξ/δ, and j=η/δ, where δ is the original distortion The size of each cell in the image.

Preferably, the z-axis rotation correction unit finds the closest point ξ ₀ from the x-axis to the image acquisition device in the original distorted image plane XZ for the geometric distortion rotated around the z-axis, and determines the XZ plane around the XZ plane based on ξ ₀ and y _c The rotation angle a _z of the z-axis; the three-dimensional space coordinates (x, y, z) of the rotated point (X, Z) are determined based on y _c and a _z ;

Preferably, the x-axis rotation correction unit, for the geometric distortion rotated around the x-axis, finds the closest point η ₀ from the z-axis to the image acquisition device in the original distorted image plane XZ, and determines the XZ plane around the x-axis based on η ₀ and y _c the rotation angle a _x .

Preferably, the relationship between point (X', Z') and three-dimensional space coordinates (x, y, z) is established:

Preferably, the y-axis rotation correction unit corrects the geometric distortion rotated around the y-axis according to the following methods:

where a _y is the rotation angle around the y-axis.

Preferably, when the size correction unit performs distorted image correction, the mapping relationship between the physical space point (x, y, z) used to the point (ξ', η') of the original distorted image is:

ξ′=δαx/rΔ

η′=δαz/rΔ

Wherein, r=(x ² +z ² ) ^1/2 , α=tan ⁻¹ (r/y), and Δ=tan ⁻¹ (δ/y _c ).

Preferably, in the two-dimensional illumination distribution estimation unit, the least squares fitting adopts the following method:

Where M is the order of the polynomial, 1 and k are the subscripts of the coefficients in the above summation polynomial and the corresponding power changes of the two variables, k increases from 0 to M, and 1 increases from 0 to k.

Preferably, in the intensity correction unit, the intensity of each point is corrected in the following manner:

I _new (ξ, η) = I _old (ξ, η) p _ave /p(ξ, η)

where I _new (ξ, η) and I _old (ξ, η) are the corrected image and original image intensities at point (ξ, η), respectively.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. During execution, the processes of the embodiments of the above-mentioned methods may be included. The storage medium may be a magnetic disk, an optical disk, a read-only memory (Read-Only Memory, ROM), or a random access memory (Random Access Memory, RAM) or the like.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that it can still be The technical solutions described in the foregoing embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions in the embodiments of the present invention.

Claims (12)

1. a plane image correction method, is characterized in that, described method comprises: S1. Acquire basic data of the image acquisition device, measure the distance y _c from the center of the object plane to the lens, and acquire original distorted image data, and establish an original image coordinate system with the center of the original distorted image as the origin; S2, establish the coordinate system X-Z of the corrected image, and take the center of the corrected image as the origin; S3. Correct the geometric distortion caused by the rotation of the X-Z plane around the z-axis and the x-axis; correct the geometric distortion caused by the rotation of the X-Z plane around the y-axis; establish the point (X, Z) on the actual plane to the physical space point (x, y, z) the relationship between; S4. Establish a mapping from physical space points (x, y, z) to points (ξ', η') of the original distorted image; S5. Determine the pixel point of the initial distorted image corresponding to the point (ξ', η') of the original distorted image, and use the difference algorithm to determine the position of the point in the corrected image according to the coordinates and pixel values of the adjacent points of the pixel point. pixel value; based on the y _c , by specifying the pixel size Δx of the corrected image, the image is corrected to the required size, and the size-corrected image is obtained; S6, obtain the 2D illumination distribution estimation of the area to be corrected by least square fitting on the image after size correction, and determine the original non-uniform illumination intensity p(ξ, η); S7. Calculate the equalized intensity pave based on the p( _ξ , η); based on the original non-uniform illumination intensity and equalized intensity, correct the intensity of each point. 2 . The method according to claim 1 , wherein the basic data of the image acquisition device includes the size, length, width of the focal plane, and the horizontal field of view of the lens. 3 . 3 . The method according to claim 1 , wherein in the step S2 , the corrected image and the original distorted image use the same image size. 4 . 4. The method according to claim 1, wherein the relationship between the point (X, Z) on the actual plane and the pixel (I, J) in the corrected image is expressed as: then I=X/Δx, J=Z/Δx, Among them, Δx is the pixel size; the relationship between the pixel (i, j) in the original distorted image and the point (ξ, η) in the original distorted image is expressed as: i=ξ/δ, and j=η/δ, where δ is the size of each pixel in the original distorted image. 5. The method according to claim 1, wherein, in the S3, for the geometric distortion rotated around the z-axis, find the closest point ξ ₀ from the x-axis in the original distorted image plane XZ to the image acquisition device, and based on the ξ ₀ and y _c determine the rotation angle az of the XZ plane around the z-axis; determine the three-dimensional space coordinates (x, y, _z ) of the rotated point (X, Z) based on y _c and a _z ; For the geometric distortion rotated around the x-axis, find the closest point η ₀ from the z-axis to the image acquisition device in the original distorted image plane XZ, and determine the rotation angle a _x of the XZ plane around the x-axis based on η ₀ and _yc . 6. method according to claim 5, is characterized in that, establishes the relation between point (X', Z') to three-dimensional space coordinate (x, y, z):

7. The method according to claim 1, wherein, in the S3, for the geometric distortion rotated around the y-axis, correction is performed according to the following manner:

where a _y is the rotation angle around the y-axis. 8. The method according to claim 1, wherein, in the S4, the mapping relationship between the physical space point (x, y, z) to the point (ξ', η') of the original distorted image is: ξ′=δαx/rΔ η′=δαz/rΔ Wherein, r=(x ² +z ² ) ^1/2 , α=tan ⁻¹ (r/y), and Δ=tan ⁻¹ (δ/y _c ). 9. The method according to claim 1, wherein, in the S6, the least squares fitting adopts the following manner: where M is the order of the polynomial, and l and k represent the corresponding power changes of the variables. 10. The method according to claim 1, characterized in that, in the S7, the intensity of each point is corrected in the following manner: I _new (ξ, η) = I _old (ξ, η) p _ave /p(ξ, η) where I _new (ξ, η) and I _old (ξ, η) are the corrected image and original image intensities at point (ξ, η), respectively. 11. A plane image correction system, wherein the system comprises: The image acquisition module is used to collect the original distorted image data; a geometric distortion correction module for performing geometric distortion correction on the original distorted image and obtaining a size corrected image; an intensity correction module for performing illumination intensity correction on the size-corrected image; Image output module for outputting the final corrected image; The geometric distortion correction module at least includes: an x-axis rotation correction unit, a y-axis rotation correction unit, a z-axis rotation correction unit, a basic data acquisition unit, and a size correction unit; The basic data acquisition unit is used to acquire the basic data of the image acquisition module, measure the distance y _c from the center of the object plane to the lens, acquire the original distorted image data from the image acquisition module, and establish the original distorted image center as the origin. image coordinate system; The x-axis rotation correction unit, the y-axis rotation correction unit, and the z-axis rotation correction unit are respectively used to correct the geometric distortion caused by rotation around the x, y, and z axes; the size correction unit, configured to correct the image to a required size based on the correction results of the x-axis rotation correction unit, the y-axis rotation correction unit, and the z-axis rotation correction unit, and the pixel size Δx of the specified correction image, Obtain a size-corrected image. 12 . The system according to claim 11 , wherein the intensity correction module at least comprises: a two-dimensional illumination distribution estimation unit and an intensity correction unit; 12 . The two-dimensional illumination distribution estimation unit is used to obtain a 2D illumination distribution estimation of the area to be corrected through least squares fitting, and determine the original non-uniform illumination intensity; The intensity correction unit is configured to correct the intensity of each point based on the original non-uniform illumination intensity and the balanced intensity.

Images