CN102663685B - Geometric correction method based on nonlinearity - Google Patents

Abstract

The invention discloses a geometric correction method based on nonlinearity. The method comprises the steps of according to a certain linear transformation relation, mapping four boundary pixel points of an image to be corrected to a correction image, determining the length-width attribute value of the correction image, ensuring the shape standardability of the correction image, optimizing the position proportion relation of the coordinate position of pixel points of the image to be corrected in the coordinate system of the image to be corrected, mapping to the correction image according to the optimized position proportion relation, according to the pixel value of each pixel point in the image to be corrected determining the pixel value of corresponding pixel points in the correction image, and filling void pixel points in the correction image to obtain the final correction image. Since the method provided by the invention linearly and nonlinearly processes the pixel points of the image and the pixel value of the pixel points, the method is not only simple to implement, but also can effectively correct nonlinear offset of the image, and can meet the accuracy requirements of image identification and image matching.

Description

A kind of based on nonlinear geometric correction method

Technical field

The present invention relates to the image rectification technology in a kind of image recognition, images match field, especially relate to a kind of based on nonlinear geometric correction method.

Background technology

Image calibration is one of key link of image recognition and images match, and the high precision of image calibration is to ensure one of basic condition that image recognition and images match are effectively carried out.But, most of geometry correction algorithm is only considered the geometry deformation of the aspects such as skew, rotation, inclination, but also there is comparatively serious nonlinear geometry distortion in a lot of images to be calibrated, be that source images and target image cannot convert with simple linear coordinate, but complicated the unknown and unsettled nonlinear coordinate transformation, this nonlinear geometry distortion has had a strong impact on the net result of image recognition and images match.Therefore, calibrate this nonlinear geometry distortion in the urgent need to a kind of new geometry correction algorithm, to ensure final engineering result precision.

Summary of the invention

Technical matters to be solved by this invention is to provide a kind of stalwartness, effectively the distortion of the nonlinear geometry of correcting image and the high geometric correction method of correction accuracy.

The present invention solves the problems of the technologies described above adopted technical scheme: a kind of based on nonlinear geometric correction method, it is characterized in that comprising the following steps:

1. the coordinate position of supposing four boundary pixel points of image to be corrected is respectively (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃), (x ₄, y ₄), to (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃) and (x ₄, y ₄) carry out linear operation, the corresponding coordinate position of four boundary pixel points that obtain image to be corrected in correcting image, correspondence is designated as (x respectively ₁, y ₁), (x ₂', y ₁), (x ₂', y ₂'), (x ₁, y ₂'), wherein, ${x_2}^{\′}=x_1+\frac{\sqrt{{(y_2-y_1)}^2+{(x_2-x_1)}^2}+\sqrt{{(y_4-y_3)}^2+{(x_4-x_3)}^2}}{2},$ ${y_2}^{\′}=y_1+\frac{\sqrt{{(y_3-y_1)}^2+{(x_3-x_1)}^2}+\sqrt{{(y_4-y_2)}^2+{(x_4-x_2)}^2}}{2};$

2. for the arbitrary pixel in image to be corrected, its coordinate position is designated as to (X, Y), then coordinates computed position (X, Y) the position proportional relation in the coordinate system of image to be corrected, again to coordinate position (X, Y) the position proportional relation in the coordinate system of image to be corrected is optimized processing, finally press coordinate position (X, Y) position proportional after optimization is related to that equal proportion is mapped in correcting image, the coordinate position obtaining in image to be corrected is (X, Y) the corresponding coordinate position of pixel in correcting image, be designated as (X ', Y '),

3. the mapping relations with pixel corresponding in image to be corrected according to any pixel in the pixel value of each pixel in image to be corrected and correcting image, determine the pixel value of pixel corresponding in correcting image, detailed process is: for the coordinate position in correcting image for (X ', Y ') pixel, if the coordinate position in correcting image be (X ', Y ') pixel in image to be corrected, only have unique respective coordinates position for (X, Y) pixel, determine that this pass is correspondence mappings relation one by one, and be directly (X by the coordinate position in image to be corrected, the pixel value assignment of pixel Y) to the coordinate position in correcting image be (X ', Y ') pixel, if the coordinate position in correcting image be (X ', Y ') pixel in image to be corrected, have the pixel of multiple correspondences, determine that this pass is one-to-many mapping relations, and the coordinate position in correcting image is the pixel value of the pixel of (X ', Y ') using the pixel value of the pixel of last mapping in image to be corrected, if the coordinate position in correcting image be the pixel of (X ', Y ') in image to be corrected without corresponding pixel, the pixel of (X ', Y ') is empty pixel to determine that the coordinate position in correcting image is.

4. the empty pixel in correcting image is carried out to local mean value and fill processing, obtain final correcting image.

Described step detailed process is 2.:

2.-1,, for the arbitrary pixel in image to be corrected, its coordinate position is designated as to (X, Y);

2.-2, the position proportional relation of coordinates computed position (X, Y) in the coordinate system of image to be corrected, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as λ ₁, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as λ ₂, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as β ₁, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as β ₂, wherein, A denotation coordination position (x ₁, y ₁), B denotation coordination position (x ₂, y ₂), C denotation coordination position (x ₄, y ₄), D denotation coordination position (x ₃, y ₃), it is (x by coordinate position that a represents in straight line x=X and image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that b represents in straight line y=Y and image to be corrected ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that c represents in straight line x=X and image to be corrected ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that d represents in straight line y=Y and image to be corrected ₁, y ₃) boundary pixel point and coordinate position be (x ₃, y ₃) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, Aa represents the line segment being formed by connecting by A and a, AB represents the line segment being formed by connecting by A and B, Dc represents the line segment being formed by connecting by D and c, DC represents the line segment being formed by connecting by D and C, and Ad represents the line segment being formed by connecting by A and d, and AD represents the line segment being formed by connecting by A and D, Bb represents the line segment being formed by connecting by B and b, and BC represents the line segment being formed by connecting by B and C;

2.-3, cross coordinate position (X, Y), the non-linear flexible longitudinal line segment ac being formed by connecting by a and c and the horizontal line segment bd being formed by connecting by b and d, by longitudinal line segment ac through non-linear flexible after with image to be corrected in be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as a ', by longitudinal line segment ac through non-linear flexible after with image to be corrected in be (x by coordinate position ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as c ', by horizontal line segment bd through non-linear flexible after with image to be corrected in be (x by coordinate position ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as b ', by horizontal line segment bd through non-linear flexible after with image to be corrected in be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as d '; Then according to a ', b ', c ' and d ', the position proportional relation after coordinates computed position (X, Y) is optimized, by λ ₁, λ ₂, β ₁, β ₂position proportional after each self-corresponding optimization is related to that correspondence is designated as λ ₁', λ ₂', β ₁', β ₂', ${{\mathrm{\λ}}_1}^{\′}={{\mathrm{\λ}}_2}^{\′}=\frac{{\mathrm{Aa}}^{\′}}{\mathrm{AB}}=\frac{{\mathrm{Dc}}^{\′}}{\mathrm{DC}}=\frac{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2}{2},$ ${{\mathrm{\β}}_1}^{\′}={{\mathrm{\β}}_2}^{\′}=\frac{{\mathrm{Ad}}^{\′}}{\mathrm{AD}}=\frac{{\mathrm{Bb}}^{\′}}{\mathrm{BC}}=\frac{{\mathrm{\β}}_1+{\mathrm{\β}}_2}{2},$ Wherein, the line segment that Aa ' expression is formed by connecting by A and a ', the line segment that Dc ' expression is formed by connecting by D and c ', the line segment that Ad ' expression is formed by connecting by A and d ', the line segment that Bb ' expression is formed by connecting by B and b ';

2.-4, be related to λ according to the position proportional after coordinate position (X, Y) optimization ₁', λ ₂', β ₁', β ₂', according to $\frac{A^{\′}{a}^{\′\′}}{A^{\′}{B}^{\′}}=\frac{{\mathrm{Aa}}^{\′}}{\mathrm{AB}}={{\mathrm{\λ}}_1}^{\′}={{\mathrm{\λ}}_2}^{\′}$ With $\frac{A^{\′}{\mathrm{dd}}^{\′}}{A^{\′}{D}^{\′}}=\frac{{\mathrm{Ad}}^{\′}}{\mathrm{AD}}={{\mathrm{\β}}_1}^{\′}={{\mathrm{\β}}_2}^{\′}$ Equal proportion is mapped in correcting image, the corresponding coordinate position of the pixel that to obtain coordinate position in image to be corrected be (X, Y) in correcting image (X ', Y '), wherein, A ' denotation coordination position (x ₁, y ₁), B ' denotation coordination position (x ₂', y ₁), D ' denotation coordination position (x ₁, y ₂'), a " represents that in straight line x=X ' and correcting image be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₂', y ₁) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, d " represents that in straight line y=Y ' and correcting image be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₁, y ₂') the intersection point of the boundary straight line that is formed by connecting of boundary pixel point.

Described step 4. in to the empty pixel in correcting image carry out local mean value fill process detailed process be: for the arbitrary empty pixel in correcting image, calculate the mean value of the pixel value of four neighbor pixels of this cavity pixel, the pixel value using the mean value calculating as this cavity pixel.

Described step 3. in the pixel of last mapping in image to be corrected be in image to be corrected with correcting image in coordinate position multiple pixels corresponding to pixel that are (X ', Y ') undertaken when 2. step is processed by priority order that last carries out the pixel that 2. step is processed.

Compared with prior art, the invention has the advantages that first according to certain linear transformation relation, four boundary pixel points of image to be corrected are mapped in correcting image, also determine the length and width property value of correcting image, ensure the shape standardization of correcting image, then the position proportional relation of the coordinate position for the treatment of the pixel in correcting image in the coordinate system of image to be corrected is optimized processing, again by the position proportional relationship map after optimization process in correcting image, the finally mapping relations with pixel corresponding in image to be corrected according to any pixel in the pixel value of each pixel in image to be corrected and correcting image, determine the pixel value of pixel corresponding in correcting image, and the empty pixel in correcting image is filled, obtain final correcting image, because the pixel of the inventive method to image and the pixel value of pixel have carried out linearity and Nonlinear Processing, therefore not only realize easy, and the effectively non-linear skew of correcting image (environment skew, photo deviation etc.), and can effectively meet the accuracy requirement of image recognition and images match.

Brief description of the drawings

Fig. 1 is the FB(flow block) of the inventive method;

Fig. 2 a is the coordinate position schematic diagram of four boundary pixel points of image to be corrected;

Fig. 2 b is four boundary pixel points of image to be corrected corresponding coordinate position schematic diagram in correcting image;

Fig. 3 a is for obtaining the schematic diagram of the position proportional relation of coordinate position (X, Y) in the coordinate system of image to be corrected;

Fig. 3 b is for optimizing the schematic diagram of the position proportional relation of coordinate position (X, Y) in the coordinate system of image to be corrected;

Fig. 3 c is the schematic diagram that obtains corresponding coordinate position in correcting image of pixel that the coordinate position in image to be corrected is (X, Y) (X ', Y ');

Fig. 4 a is image to be corrected;

Fig. 4 b is the correcting image that the image to be corrected shown in Fig. 4 a obtains after geometric correction method of the present invention is processed.

Embodiment

Below in conjunction with accompanying drawing, embodiment is described in further detail the present invention.

The one that the present invention proposes is based on nonlinear geometric correction method, and as shown in Figure 1, it comprises the following steps:

1. the coordinate position of supposing four boundary pixel points of image to be corrected is respectively (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃), (x ₄, y ₄), to (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃) and (x ₄, y ₄) carry out linear operation, the corresponding coordinate position of four boundary pixel points that obtain image to be corrected in correcting image, correspondence is designated as (x respectively ₁, y ₁), (x ₂', y ₁), (x ₂', y ₂'), (x ₁, y ₂'), wherein, ${x_2}^{\′}=x_1+\frac{\sqrt{{(y_2-y_1)}^2+{(x_2-x_1)}^2}+\sqrt{{(y_4-y_3)}^2+{(x_4-x_3)}^2}}{2},$ ${y_2}^{\′}=y_1+\frac{\sqrt{{(y_3-y_1)}^2+{(x_3-x_1)}^2}+\sqrt{{(y_4-y_2)}^2+{(x_4-x_2)}^2}}{2},$ As shown in Figure 2 a and 2 b, Fig. 2 a has provided the coordinate position of four boundary pixel points of image to be corrected, the corresponding coordinate position of four boundary pixel points that Fig. 2 b has provided image to be corrected in correcting image.

In the present embodiment, the coordinate position of supposing four boundary pixel points of image to be corrected is respectively (19,1003), (614,1053), (100,17), (700,68), the corresponding coordinate position of its obtaining after linear operation in correcting image is (19,1003), (619,1003), (19,1992), (619,1992).

2. for the arbitrary pixel in image to be corrected, its coordinate position is designated as to (X, Y), then coordinates computed position (X, Y) the position proportional relation in the coordinate system of image to be corrected, again to coordinate position (X, Y) the position proportional relation in the coordinate system of image to be corrected is optimized processing, finally press coordinate position (X, Y) position proportional after optimization is related to that equal proportion is mapped in correcting image, the coordinate position obtaining in image to be corrected is (X, Y) the corresponding coordinate position of pixel in correcting image, be designated as (X ', Y ').

In this specific embodiment, step detailed process is 2.:

2.-1,, for the arbitrary pixel in image to be corrected, its coordinate position is designated as to (X, Y).

2.-2, as shown in Figure 3 a, the position proportional relation of coordinates computed position (X, Y) in the coordinate system of image to be corrected, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as λ ₁, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as λ ₂, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as β ₁, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as β ₂, wherein, A denotation coordination position (x ₁, y ₁), B denotation coordination position (x ₂, y ₂), C denotation coordination position (x ₄, y ₄), D denotation coordination position (x ₃, y ₃), it is (x by coordinate position that a represents in straight line x=X and image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that b represents in straight line y=Y and image to be corrected ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that c represents in straight line x=X and image to be corrected ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that d represents in straight line y=Y and image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, Aa represents the line segment being formed by connecting by A and a, AB represents the line segment being formed by connecting by A and B, Dc represents the line segment being formed by connecting by D and c, DC represents the line segment being formed by connecting by D and C, and Ad represents the line segment being formed by connecting by A and d, and AD represents the line segment being formed by connecting by A and D, Bb represents the line segment being formed by connecting by B and b, and BC represents the line segment being formed by connecting by B and C.

2.-3, as shown in Figure 3 b, cross coordinate position (X, Y), the non-linear flexible longitudinal line segment ac being formed by connecting by a and c and the horizontal line segment bd being formed by connecting by b and d, by longitudinal line segment ac through non-linear flexible after with image to be corrected in be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as a ', by longitudinal line segment ac through non-linear flexible after with image to be corrected in be (x by coordinate position ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as c ', by horizontal line segment bd through non-linear flexible after with image to be corrected in be (x by coordinate position ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as b ', by horizontal line segment bd through non-linear flexible after with image to be corrected in be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as d '; Then according to a ', b ', c ' and d ', the position proportional relation after coordinates computed position (X, Y) is optimized, by λ ₁, λ ₂, β ₁, β ₂position proportional after each self-corresponding optimization is related to that correspondence is designated as λ ₁', λ ₂', β ₁', β ₂', ${{\mathrm{\λ}}_1}^{\′}={{\mathrm{\λ}}_2}^{\′}=\frac{{\mathrm{Aa}}^{\′}}{\mathrm{AB}}=\frac{{\mathrm{Dc}}^{\′}}{\mathrm{DC}}=\frac{{\mathrm{\λ}}_1+{\mathrm{\λ}}_2}{2},$ ${{\mathrm{\β}}_1}^{\′}={{\mathrm{\β}}_2}^{\′}=\frac{{\mathrm{Ad}}^{\′}}{\mathrm{AD}}=\frac{{\mathrm{Bb}}^{\′}}{\mathrm{BC}}=\frac{{\mathrm{\β}}_1+{\mathrm{\β}}_2}{2},$ Wherein, the line segment that Aa ' expression is formed by connecting by A and a ', the line segment that Dc ' expression is formed by connecting by D and c ', the line segment that Ad ' expression is formed by connecting by A and d ', the line segment that Bb ' expression is formed by connecting by B and b '.

2.-4, as shown in Figure 3 c, be related to λ according to the position proportional after coordinate position (X, Y) optimization ₁', λ ₂', β ₁', β ₂', according to $\frac{A^{\′}{a}^{\′\′}}{A^{\′}{B}^{\′}}=\frac{{\mathrm{Aa}}^{\′}}{\mathrm{AB}}={{\mathrm{\λ}}_1}^{\′}={{\mathrm{\λ}}_2}^{\′}$ With $\frac{A^{\′}{\mathrm{dd}}^{\′}}{A^{\′}{D}^{\′}}=\frac{{\mathrm{Ad}}^{\′}}{\mathrm{AD}}={{\mathrm{\β}}_1}^{\′}={{\mathrm{\β}}_2}^{\′}$ Equal proportion is mapped in correcting image, the corresponding coordinate position of the pixel that to obtain coordinate position in image to be corrected be (X, Y) in correcting image (X ', Y '), wherein, A ' denotation coordination position (x ₁, y ₁), B ' denotation coordination position (x ₂', y ₁), D ' denotation coordination position (x ₁, y ₂'), a " represents that in straight line x=X ' and correcting image be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₂', y ₁) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, d " represents that in straight line y=Y ' and correcting image be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₁, y ₂') the intersection point of the boundary straight line that is formed by connecting of boundary pixel point.

3. the mapping relations with pixel corresponding in image to be corrected according to any pixel in the pixel value of each pixel in image to be corrected and correcting image, determine the pixel value of pixel corresponding in correcting image, detailed process is: after 2. step is processed, each pixel in image to be corrected projects on correcting image, due to Nonlinear Mapping relation, the coordinate position of a pixel in the corresponding image to be corrected of coordinate position possibility of a pixel in correcting image, also the coordinate position of the multiple pixels in the corresponding image to be corrected of possibility, therefore, for the coordinate position in correcting image be (X ', Y ') pixel, if the coordinate position in correcting image be (X ', Y ') pixel in image to be corrected, only to have unique corresponding coordinate position be (X, Y) pixel, determine that this pass is correspondence mappings relation one by one, and be directly (X by the coordinate position in image to be corrected, the pixel value assignment of pixel Y) to the coordinate position in correcting image be (X ', Y ') pixel, if the coordinate position in correcting image be (X ', Y ') pixel in image to be corrected, have the pixel of multiple correspondences, determine that this pass is one-to-many mapping relations, and using the pixel value of the pixel of last mapping in image to be corrected coordinate position in correcting image be (X ', Y ') the pixel value of pixel, according to the coordinate position in correcting image be (X ', Y ') the corresponding multiple pixels of pixel in image to be corrected carry out the priority order that 2. step is processed, defining in image to be corrected last, to carry out the pixel that 2. step process be the pixel of last mapping, if the coordinate position in correcting image be the pixel of (X ', Y ') in image to be corrected without corresponding pixel, the pixel of (X ', Y ') is empty pixel to determine that the coordinate position in correcting image is.

4. the empty pixel in correcting image is carried out to local mean value and fill processing, obtain final correcting image.

In this specific embodiment, the detailed process of the empty pixel in correcting image being carried out to local mean value filling processing is: for the arbitrary empty pixel in correcting image, calculate the mean value of the pixel value of four neighbor pixels of this cavity pixel, the pixel value using the mean value calculating as this cavity pixel.As: the coordinate position of supposing the arbitrary empty pixel in correcting image for (X ", Y "), and make (X _l", Y _l"), (X _r", Y _r"), (X _u", Y _u"), (X _d", Y _d") respectively denotation coordination position be (X ", the coordinate position of the left neighbor pixel of the empty pixel of Y "), right neighbor pixel, upper neighbor pixel, lower neighbor pixel, and be (X by coordinate position _l", Y _l"), (X _r", Y _r"), (X _u", Y _u"), (X _d", Y _dthe pixel value of the pixel of ") respectively correspondence is designated as f1, f2, f3, f4, and the coordinate position in correcting image is (X ", the pixel value f of the empty pixel of Y ") (X ", Y ")=(f1+f2+f3+f4)/4.

Fig. 4 a has provided an image to be corrected, can find out the skew of image generation environment position from Fig. 4 a, has out-of-plane upset, also deposits inclination planar, in images match field, belongs to typical non linear match condition.Fig. 4 b has provided the correcting image that the image to be corrected shown in Fig. 4 a obtains after geometric correction method of the present invention is processed.Can effectively be proofreaied and correct after the inventive method is processed from finding out the image to be corrected shown in Fig. 4 a Fig. 4 b, the outer upset of plane is calibrated, plane is tilted also to be eliminated, and correcting image pixel is clear, correction accuracy is higher, applicable to the images match under non-linear drift condition.

Claims (3)

1. based on a nonlinear geometric correction method, it is characterized in that comprising the following steps:

1. the coordinate position of supposing four boundary pixel points of image to be corrected is respectively (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃), (x ₄, y ₄), to (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃) and (x ₄, y ₄) carry out linear operation, the corresponding coordinate position of four boundary pixel points that obtain image to be corrected in correcting image, correspondence is designated as (x respectively ₁, y ₁), (x ₂', y ₁), (x ₂', y ₂'), (x ₁, y ₂'), wherein, ${x_2}^{\′}=x_1+\frac{\sqrt{{(y_2-y_1)}^2+{(x_2-x_1)}^2}+\sqrt{{(y_4-y_3)}^2+{(x_4-x_3)}^2}}{2},$ ${y_2}^{\′}=y_1+\frac{\sqrt{{(y_3-y_1)}^2+{(x_3-x_1)}^2}+\sqrt{{(y_4-y_2)}^2+{(x_4-x_2)}^2}}{2};$

2. for the arbitrary pixel in image to be corrected, its coordinate position is designated as to (X, Y), then coordinates computed position (X, Y) the position proportional relation in the coordinate system of image to be corrected, again to coordinate position (X, Y) the position proportional relation in the coordinate system of image to be corrected is optimized processing, finally press coordinate position (X, Y) position proportional after optimization is related to that equal proportion is mapped in correcting image, the coordinate position obtaining in image to be corrected is (X, Y) the corresponding coordinate position of pixel in correcting image, be designated as (X', Y'),

Described step detailed process is 2.:

2.-1,, for the arbitrary pixel in image to be corrected, its coordinate position is designated as to (X, Y);

2.-2, the position proportional relation of coordinates computed position (X, Y) in the coordinate system of image to be corrected, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as λ ₁, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as λ ₂, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as β ₁, it is (x by coordinate position that coordinate position (X, Y) is projected in image to be corrected ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the boundary straight line that is formed by connecting of boundary pixel point on the position proportional relation of subpoint be designated as β ₂, wherein, A denotation coordination position (x ₁, y ₁), B denotation coordination position (x ₂, y ₂), C denotation coordination position (x ₄, y ₄), D denotation coordination position (x ₃, y ₃), it is (x by coordinate position that a represents in straight line x=X and image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that b represents in straight line y=Y and image to be corrected ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that c represents in straight line x=X and image to be corrected ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that d represents in straight line y=Y and image to be corrected ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, Aa represents the line segment being formed by connecting by A and a, AB represents the line segment being formed by connecting by A and B, Dc represents the line segment being formed by connecting by D and c, DC represents the line segment being formed by connecting by D and C, and Ad represents the line segment being formed by connecting by A and d, and AD represents the line segment being formed by connecting by A and D, Bb represents the line segment being formed by connecting by B and b, and BC represents the line segment being formed by connecting by B and C;

2.-3, cross coordinate position (X, Y), the non-linear flexible longitudinal line segment ac being formed by connecting by a and c and the horizontal line segment bd being formed by connecting by b and d, by longitudinal line segment ac through non-linear flexible after with image to be corrected in be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₂, y ₂) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as a', by longitudinal line segment ac through non-linear flexible after with image to be corrected in be (x by coordinate position ₃, y ₃) boundary pixel point and coordinate position be (x ₄, y ₄) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as c', by horizontal line segment bd through non-linear flexible after with image to be corrected in be (x by coordinate position ₂, y ₂) boundary pixel point and coordinate position be (x ₄, y ₄) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as b', by horizontal line segment bd through non-linear flexible after with image to be corrected in be (x by coordinate position ₁, y ₁) boundary pixel point and coordinate position be (x ₃, y ₃) the new intersection point of the boundary straight line that is formed by connecting of boundary pixel point be designated as d'; Then according to a', b', c' and d', the position proportional relation after coordinates computed position (X, Y) is optimized, by λ ₁, λ ₂, β ₁, β ₂position proportional after each self-corresponding optimization is related to that correspondence is designated as λ ₁', λ ₂', β ₁', β ₂', wherein, Aa' represents the line segment being formed by connecting by A and a', and Dc' represents the line segment being formed by connecting by D and c', and Ad' represents the line segment being formed by connecting by A and d', and Bb' represents the line segment being formed by connecting by B and b';

2.-4, be related to λ according to the position proportional after coordinate position (X, Y) optimization ₁', λ ₂', β ₁', β ₂', according to with equal proportion is mapped in correcting image, the corresponding coordinate position (X', Y') of the pixel that to obtain coordinate position in image to be corrected be (X, Y) in correcting image, wherein, A' denotation coordination position (x ₁, y ₁), B' denotation coordination position (x ₂', y ₁), D' denotation coordination position (x ₁, y ₂'), it is (x by coordinate position that a'' represents in straight line x=X' and correcting image ₁, y ₁) boundary pixel point and coordinate position be (x ₂', y ₁) the intersection point of the boundary straight line that is formed by connecting of boundary pixel point, it is (x by coordinate position that d'' represents in straight line y=Y' and correcting image ₁, y ₁) boundary pixel point and coordinate position be (x ₁, y ₂') the intersection point of the boundary straight line that is formed by connecting of boundary pixel point;

3. the mapping relations with pixel corresponding in image to be corrected according to any pixel in the pixel value of each pixel in image to be corrected and correcting image, determine the pixel value of pixel corresponding in correcting image, detailed process is: be (X' for the coordinate position in correcting image, Y') pixel, if the coordinate position in correcting image is (X', Y') it is (X that pixel only has unique corresponding coordinate position in image to be corrected, Y) pixel, determine that this pass is correspondence mappings relation one by one, and be directly (X by the coordinate position in image to be corrected, the pixel value assignment of pixel Y) is (X' to the coordinate position in correcting image, Y') pixel, if the coordinate position in correcting image is (X', Y') pixel has the pixel of multiple correspondences in image to be corrected, determine that this pass is one-to-many mapping relations, and the pixel value of the pixel that the coordinate position in correcting image is (X', Y') using the pixel value of the pixel of last mapping in image to be corrected, if the pixel that the coordinate position in correcting image is (X', Y') without corresponding pixel, determines that the pixel that the coordinate position in correcting image is (X', Y') is empty pixel in image to be corrected,

4. the empty pixel in correcting image is carried out to local mean value and fill processing, obtain final correcting image.

2. one according to claim 1 is based on nonlinear geometric correction method, it is characterized in that during described step 4. that the empty pixel in correcting image is carried out to local mean value fills the detailed process of processing and be: for the arbitrary empty pixel in correcting image, calculate the mean value of the pixel value of four neighbor pixels of this cavity pixel, the pixel value using the mean value calculating as this cavity pixel.

3. one according to claim 2 is based on nonlinear geometric correction method, it is characterized in that during described step 3. the pixel of last mapping in image to be corrected be in image to be corrected with correcting image in multiple pixels corresponding to the coordinate position pixel that is (X', Y') undertaken when 2. step is processed by priority order that last carries out the pixel that 2. step is processed.