KR101428534B1 - Apparatus and method for correcting lens distortion - Google Patents

Abstract

(R ² ) of a distance between a pixel of a corrected image transmitted from an image sensor through a lens and an optical origin is divided into a plurality of segments, A similarity ratio estimating unit for estimating a similarity ratio s using an approximated straight line, and a similarity ratio estimating unit for calculating coordinates (x ', y') of pixels distorted by using the estimated similarity ratio (X, y ') -th pixel of the distorted image, and a correction pixel value calculation unit for calculating a correction pixel value for calculating the (x, y) -th pixel value of the corrected image using the coordinate value of the A part including a part is provided.
By using the apparatus and method for correcting lens distortion as described above, the optical distortion of the lens can be corrected based on the linear spline interpolation method, and the computational complexity in the correction process can be remarkably reduced.

Description

[0001] APPARATUS AND METHOD FOR CORRECTING LENS DISTORTION [0002]

The present invention relates to a lens distortion correction apparatus and method, and more particularly, to a lens distortion correction apparatus and method for correcting distortion caused by optical characteristics of a lens using a digital video signal processing technique.

2. Description of the Related Art Generally, a digital camera such as a camera incorporated in a mobile phone has a lens and an image sensor such as a CCD or CMOS for detecting an image incident through the lens.

The image transmitted through the lens differs from the image actually seen by the human eye in various aspects due to various optical characteristics of the lens itself.

Here, the difference between the image seen through the naked eye and the image detected by the image sensor is called the lens aberration, and the lens aberration is removed during the design and manufacture of the lens, thereby obtaining substantially the same image as that shown by the naked eye. .

There are various kinds of lens aberrations, and in particular, a distortion aberration of a lens (hereinafter referred to as "lens distortion" or "lens distortion") caused by a difference in geometrical form between a lens and a sensor, Is aberration of the lens which is represented by a curve after passing through the lens.

That is, image-based lens distortion is largely classified into radial distortion and tangential distortion.

A technique for correcting the double radial distortion is mainly proposed. In fact, most of the lens distortion is degraded due to radial distortion.

An example of such a technique for correcting radial distortion is disclosed in Korean Patent Registration No. 10-0456632 (published on Nov. 10, 2004, hereinafter referred to as Patent Document 1), Korean Patent Registration No. 10-0744937 (2007 Patent Document 2 ")," A pipelined architecture for real-time correction of barrel distortion in wide-angle camera images "(HT Ngo et al., IEEE Transactions on Circuits and Systems for Video Technology [IEEE Trans. Circuits Syst. Video Technol.] Vol. 15, No. 3, Mar. 2005., hereinafter referred to as "Non-Patent Document 1").

On the other hand, DIS (x ', y'), Corrected Image Space (hereinafter, referred to as 'CIS') values of the (x ', y') th pixel of the distorted image (X, y) pixel of the current pixel (hereinafter, referred to as a pixel) is referred to as CIS (x, y).

Then, the lens distortion correction method according to the related art including the patent document 1, the patent document 2 and the non-patent document 1 is a method of correcting the distortion of the pixels (x ', y') of DIS corresponding to the pixel position of (x, y) .

(X, y) of the rectangular coordinate system is expressed as (r,?) In the polar coordinate system, and r '(corrected) in the coordinate system is calculated by the following equation , r is the distance from the optical origin to (x, y), r 'is the distance from the optical origin to (x', y '), and then represented by a rectangular coordinate system, Find the location.

(O _x , O _y ) is the coordinate value of the optical origin.

Therefore, r 'can be expressed by the following equation (2).

Here, c0, c1, and c3 are distortion coefficients, f (r) is a correction polynomial, and the accuracy of correction varies depending on the degree of f (r).

Accordingly, (x ', y') can be calculated using the following equations (3) and (4).

이다.here,

to be.

DIS (x ', y') which is a pixel value of DIS corresponding to (x ', y') calculated in this manner is obtained and correction for the corresponding pixel is completed through the process of obtaining CIS (x, y).

Here, (x, y) all have integer values, but the position (x ', y') of the distorted pixel may not be an integer value.

Accordingly, the lens distortion correction method according to the related art can obtain the value of DIS (x ', y') by using neighboring pixel values based on Bilinear Transform.

Korean Patent Registration No. 10-0456632 (published on November 10, 2004) Korea Patent Registration No. 10-0744937 (issued on August 1, 2007)

 "A pipelined architecture for real-time correction of barrel distortion in wide-angle camera images" (HT Ngo et al., IEEE Transactions on Circuits and Systems for Video Technology, vol. 15 , No. 3, Mar. 2005.

As described above, in the conventional lens distortion correcting method including Patent Document 1, Patent Document 2 and Non Patent Document 1, in the process of obtaining the pixel position in the distorted image corresponding to the pixel position in the corrected image The correction of the polar coordinate system is carried out, which involves the calculation of complex polynomials.

In particular, since the correction method of the lens distortion according to the related art performs the correction process as described above for each pixel, the correction process is performed so that the calculation of f (r) and the transcendental function such as sin &thetas; Operation must be accompanied.

In recent years, high-resolution digital imaging apparatuses have become widespread, and such high-resolution digital imaging apparatuses must have a relatively high degree of f (r) relative to low resolution.

Accordingly, the conventional lens distortion correction method has a problem that the computational complexity of the lens distortion correction process is further increased in order to achieve high quality lens distortion correction in a high-resolution imaging apparatus.

On the other hand, it is possible to correct the optical distortion of the lens by introducing a more precise lens by improving the manufacture of the lens itself and the process technology. However, in this case, an expensive lens is required, there was.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a lens distortion correction apparatus and method for correcting distortion caused by optical characteristics of a lens using a digital image signal processing technique.

It is another object of the present invention to provide an apparatus and method for correcting lens distortion that can minimize computational complexity in a lens distortion correction process.

In accordance with an aspect of the present invention for achieving the above object, the present invention provides a method for correcting an error between a pixel of a corrected image transmitted from an image sensor via a lens and a square (r ² ) , An approximation ratio estimating unit for estimating a similarity ratio s using approximated straight lines, and a similarity ratio estimating unit for estimating a similarity ratio s of the distorted image by using the estimated similarity ratio x (x, y ') th pixel of the corrected image by using the distortion pixel coordinate value calculator for calculating the coordinate value of the (x, y') th pixel of the distorted image, And a correction pixel value calculation unit for calculating a pixel value.

Wherein the approximating unit comprises: a segment dividing unit dividing the square (r ² ) of the distance between the (x, y) th pixel of the corrected image and the optical origin into a plurality of segments; and a similarity ratio corresponding to each segment divided in the segment dividing unit And a linear approximation unit that approximates the linear approximation unit to a straight line.

And the likelihood ratio estimator estimates a similarity ratio s based on the linear spline interpolation.

The distortion pixel coordinate value (x ', y') is obtained by substituting the estimated similarity ratio s into the equations (1) and (2) and using the slope and slice of the approximated straight line, ) Values are calculated by multiplying one time and adding one time.

.......................[수학식 1]

... " (1)

.......................[수학식 2]

... " (2) "

Where r is the distance between the optical origin and the (x, y) th pixel of the corrected image,

r 'is the distance between the optical origin and the (x', y ') th pixel of the distorted image,

(O _x , O _y ) is the coordinate value of the optical origin.

According to another aspect of the present invention, there is provided a method of correcting an image, comprising the steps of: (a) calculating a square of a distance (r ² ) from an optical origin to a (x, y) ) the square of the distance to the optical reference point calculated in the step (based on the r ²⁾ estimating a dameumbi (s), (c) of the distorted image with the dameumbi (s) estimated by the step (b) ( (x ', y') th pixel value (x ', y') of the distorted image calculated in the step (c) (x, y) pixel value (CIS (x'y)) of the corrected image using the corrected pixel value (x ', y').

Wherein the step (a) (a1) of the corrected image (x, y) th pixel and calculating the squared (r ²⁾ between the optical origin distances and (a2) and an optical reference point calculated in the (a1) step And dividing the square of the distance (r < ² >) into a plurality of sections and approximating each divided section to a straight line.

The step (b) is characterized by estimating the similarity ratio (s = r '/ r) based on the linear spline interpolation. Where r 'is the distance between the optical origin and the (x', y ') th pixel of the distorted image.

The distance r between the optical origin and the (x, y) th pixel of the corrected image is calculated by Equation 3, and the distance between the optical origin and the (x ', y') th pixel of the distorted image (r ') is calculated by Equation (4), and the similarity ratio s is calculated by Equation (5).

...............[수학식 3]

&Quot; (3) "

.....[수학식 4]

&Quot; (4) "

...........[수학식 5]

&Quot; (5) "

In this case, c0, c1 and c3 are distortion coefficients, f (r) is a correction polynomial, s (r) is a similarity function, (O _x , O _y ) is a coordinate value of the optical origin.

The step (c) is a step of substituting the estimated similarity ratio (s) into the equations (1) and (2) to multiply the value of the similarity function s (r) by using the slope and the slice of the approximated straight line, And calculates coordinate values (x ', y') of the (x ', y') -th pixel of the distorted image through calculation once.

.......................[수학식 1]

... " (1)

.......................[수학식 2]

... " (2) "

Where r is the distance between the optical origin and the (x, y) th pixel of the corrected image,

r 'is the distance between the optical origin and the (x', y ') th pixel of the distorted image,

(O _x , O _y ) is the coordinate value of the optical origin.

As described above, the present invention can correct the optical distortion of the lens based on the linear spline interpolation method, thereby remarkably lowering the computational complexity performed in the correction process.

That is, the present invention approximates a straight line of each divided section using the similarity ratio estimated by the linear spline interpolation in the process of correcting the optical distortion of the lens, and approximates the straight line of the corrected image by using the slope and the slice of the approximated straight line The pixel value can be calculated.

Accordingly, the present invention can achieve high-quality distortion correction that can be achieved through a high-order similarity ratio function while minimizing the amount of computation in the distortion correction process.

As a result, the present invention has the effect of improving the performance of the digital imaging apparatus for correcting the optical distortion of the lens and reducing the manufacturing cost of the product.

1 is a block diagram of a lens distortion correcting apparatus according to a preferred embodiment of the present invention;
FIG. 2 is a graph showing the relationship between the similarity ratio function and the square of the distance between the pixel of the corrected image and the optical origin,
FIG. 3 is a flowchart for explaining a lens distortion correction method stepwise according to a preferred embodiment of the present invention. FIG.

Hereinafter, an apparatus and method for correcting a lens distortion according to a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram of a lens distortion correcting apparatus according to a preferred embodiment of the present invention, and FIG. 2 is a graph showing a relationship between a similarity ratio function and a square of a distance between a pixel of the corrected image and the optical origin.

The present invention uses a method of estimating the similarity ratio based on linear spline interpolation without directly calculating the similarity ratio function directly in order to correct distorted images in the lens through the lens.

For this purpose, in this embodiment, the calculation for obtaining the (x ', y') th pixel of the distorted image corresponding to the (x, y) th pixel of the corrected image in the distortion correction process is expressed by the following equations .

Here, r is the corrected image and the optical origin (x, y) distance between the second pixel, r 'is a distorted image (x' distance, (O _x, O _y) between, y ') th pixel is It is the coordinate value of the optical origin.

That is, r can be expressed by Equation (1) as described in the background art, and r 'can be expressed by Equation (2).

Accordingly, in the present embodiment, r '/ r is defined as a similarity ratio (or an expansion ratio, hereinafter referred to as' s') between the distorted image and the corrected image.

The amplitude ratio (s) can be defined as Equation (7) below.

1, a lens distortion correcting apparatus 20 according to a preferred embodiment of the present invention includes a lens 10 for correcting a distortion of an image, A similarity ratio estimating unit 22 for estimating a similarity ratio s using the approximated straight line, an estimated similarity ratio s, an estimated similarity ratio s, (x ', y') -th pixel of the calculated distorted image, which calculates the coordinate value of the (x ', y') th pixel of the distorted image using the pixel value s, (X, y) th pixel value of the corrected image by using the coordinate values of the corrected pixel value of the image.

The image corrected through the lens distortion correcting apparatus 20 configured as described above may be transmitted to the image output apparatus 12 and displayed on the screen of the image output apparatus 12. [

As shown in FIG. 2, the approximation unit 21 divides the square (r ² ) of the distance between the (x, y) th pixel and the optical origin of the corrected image into a plurality of, for example, six (Not shown), and a linear approximation unit (not shown in the figure) of each divided section as a straight line.

The likelihood ratio estimator 22 estimates the similarity ratio (s = r '/ r) based on linear spline interpolation.

The distortion pixel coordinate value calculation unit 23 substitutes the similarity ratio s estimated by the similarity ratio estimation unit 23 into the above Equations 5 and 6 to calculate the coordinates of the (x ', y') th pixel of the distorted image And calculates the value (x ', y').

That is, the distortion pixel coordinate value calculator 23 multiplies the value of the similarity function (s (r)) by the slope and the offset of the straight line of each approximated section, .

The corrected pixel value calculating unit 24 calculates the corrected pixel value (x, y) of the corrected image using the value (DIS (x ', y')) of the calculated (x ', y') th pixel of the distorted image (CIS (x, y)).

As described above, the present invention approximates the straight line of each divided section using the similarity ratio estimated by the linear spline interpolation in the process of correcting the optical distortion of the lens, and obtains the corrected corrected image simply by using the slope and the slice of the approximated straight line Can be calculated.

Accordingly, the present invention can achieve a high-quality distortion correction that can achieve the degree of distortion correction process operation through a high-order similarity ratio function.

Next, a lens distortion correction method according to a preferred embodiment of the present invention will be described in detail with reference to FIG.

FIG. 3 is a flowchart for explaining a lens distortion correction method according to a preferred embodiment of the present invention.

The method for correcting a lens distortion according to a preferred embodiment of the present invention includes calculating a square (r ² ) of a distance from an optical origin to a (x, y) -th pixel of a corrected image, as shown in FIG. 3 (S10), the square of the distance from the calculated optical origin (r ²⁾ a (x ', y') of the image distortion by using the step (S20), the estimated dameumbi (s) to estimate the dameumbi (s) according to the (X ', y') of the distorted image is calculated using the pixel value DIS (x ', y') of the distorted image, the step S30 of calculating the coordinate value (x ', y' (X, y) th pixel value (CIS (x'y)) of the pixel (S40).

More specifically, in step S10, the approximation unit 21 calculates the square (r ² ) of the distance between the (x, y) th pixel and the optical origin of the corrected image, and calculates the square of the distance r ² ) is divided into a plurality of sections, and each divided section is approximated by a straight line.

For example, each interval may be divided by the poles of the function that calculates the square (r ² ) of the distance between the (x, y) th pixel of the corrected image and the optical origin.

Of course, the present invention is not limited to this, and various numerical analytical methods may be introduced depending on the number of divided subdivisions to change the divisional interval.

Then, in step S20, the similarity ratio estimating unit 22 estimates the similarity ratio (s = r '/ r) based on the linear spline interpolation method.

In step S30, the distortion pixel coordinate value calculation unit 23 substitutes the similarity ratio s estimated by the similarity ratio estimation unit 23 into the above-described equations (5) and (6) to calculate the slope and slice of the approximated straight line (X ', y') pixel of the distorted image is calculated by multiplying the value of the similarity ratio function (s (r)) by the multiplication once and the addition one time.

Then, in step S40, the correction pixel value calculation unit 24 calculates the corrected pixel value (x, y ') of the corrected image by using the value (DIS (x', y ')) y) th pixel value (CIS (x, y)).

Through the above process, the present invention can correct the optical distortion of the lens based on the linear spline interpolation and significantly reduce the computational complexity in the correction process.

Although the present invention has been described in detail with reference to the above embodiments, it is needless to say that the present invention is not limited to the above-described embodiments, and various modifications may be made without departing from the spirit of the present invention.

The present invention is applied to a technique of improving distortion correction quality by minimizing computational complexity in a process of correcting distortion caused by optical characteristics of a lens using a digital image signal processing technique.

10: Lens
11: Image sensor
12: Video output device
20: Lens Distortion Correction Device
21: Approximate portion
22:
23: Distortion pixel coordinate value calculating section
24: Correction pixel value calculation unit

Claims (9)

The square of the distance (r ² ) between the pixel of the corrected image transmitted from the image sensor 11 through the lens 10 and the optical origin is divided into a plurality of segments and the similarity ratios of the divided segments are defined as a straight line The approximate approximation unit 21,
A likelihood ratio estimating unit 22 for estimating a similarity ratio s by using an approximated straight line,
A distortion pixel coordinate value calculator 23 for calculating the coordinate value of the (x ', y') th pixel of the distorted image using the estimated similarity ratio s, and
And a corrected pixel value calculating unit (24) for calculating the (x, y) th pixel value of the corrected image by using the coordinate value of the (x ', y') th pixel of the calculated distorted image Lens distortion correction device. The method according to claim 1,
The approximation unit 21 includes an interval dividing unit for dividing the square (r ² ) of the distance between the (x, y) th pixel of the corrected image and the optical origin into a plurality of intervals,
And a linear approximation unit for linearly approximating a similarity ratio corresponding to each section divided by the section division unit. 3. The method of claim 2,
Wherein the likelihood ratio estimator (22) estimates the similarity ratio (s) based on the linear spline interpolation. The method of claim 3,
The distortion pixel coordinate value (x ', y') is obtained by substituting the estimated similarity ratio s into the equations (1) and (2) and using the slope and slice of the approximated straight line, ) Is calculated by multiplying the value by one and the addition one time.

... " (1)

... " (2) "
Where r is the distance between the optical origin and the (x, y) th pixel of the corrected image,
r 'is the distance between the optical origin and the (x', y ') th pixel of the distorted image,
(O _x , O _y ) is the coordinate value of the optical origin. (a) calculating a square (r ² ) of the distance from the optical origin to the (x, y) -th pixel of the corrected image,
(b) estimating a similarity ratio s based on a square (r ² ) of the distance from the optical origin calculated in the step (a)
(c) calculating a coordinate value (x ', y') of the (x ', y') th pixel of the distorted image using the similarity ratio s estimated in the step (b)
(x, y) th pixel value of the image corrected using the (x ', y') th pixel value DIS (x ', y') of the distorted image calculated in the step (c) (CIS (x'y)) of the lens distortion correction value. 6. The method of claim 5, wherein step (a)
(a1) of the corrected image (x, y) and calculating the squared (r ²⁾ between the second pixel and the optical origin distance
(a2) lens distortion characterized in that it comprises the step of approximating the distance squared (r ²⁾ of each section dameumbi of the segmented by dividing into a plurality of sections of the optical reference point calculated in the (a1) step in a straight line Correction method. The method according to claim 6,
Wherein the step (b) estimates the similarity ratio (s = r '/ r) based on the linear spline interpolation.
Where r 'is the distance between the optical origin and the (x', y ') th pixel of the distorted image. The method according to claim 6,
(R) between the optical origin and the (x, y) -th pixel of the corrected image is calculated by Equation (3)
The distance r 'between the optical origin and the (x', y ') th pixel of the distorted image is calculated by Equation (4)
Wherein the similarity ratio (s) is calculated by Equation (5).

&Quot; (3) "

&Quot; (4) "

&Quot; (5) "
In this case, c0, c1 and c3 are distortion coefficients, f (r) is a correction polynomial, s (r) is a similarity function, (O _x , O _y ) is a coordinate value of the optical origin. 9. The method according to any one of claims 5 to 8,
The step (c) is a step of substituting the estimated similarity ratio (s) into the equations (1) and (2) to multiply the value of the similarity function s (r) by using the slope and the slice of the approximated straight line, Calculating a coordinate value (x ', y') of the (x ', y') th pixel of the distorted image through calculation once.

... " (1)

... " (2) "
Where r is the distance between the optical origin and the (x, y) th pixel of the corrected image,
r 'is the distance between the optical origin and the (x', y ') th pixel of the distorted image,
(O _x , O _y ) is the coordinate value of the optical origin.

Images