KR101440847B1 - 3D distortion apparatus and method using fisheye lens - Google Patents

Abstract

A three-dimensional distortion apparatus using a fisheye lens is disclosed. The input part is input the target image. The virtual space creating unit creates a 3D virtual space for the input target image. The distortion unit generates a distorted 3D virtual space by distorting the generated 3D virtual space based on the camera lens specification. According to the present invention, a three-dimensional simulation of a fisheye lens distortion for a rearview camera for a vehicle can be easily performed without going directly to the scene.

Description

Technical Field [0001] The present invention relates to a 3D distortion apparatus and method using a fisheye lens,

The present invention relates to a three-dimensional distortion apparatus and method using a fisheye lens, and more particularly, to a fisheye lens distortion three-dimensional simulation apparatus and method for a rearview camera for a vehicle.

In recent years, automobiles have been equipped with various cameras for the safety and convenience of the driver. As the demand for automobile cameras increases, research on lane detection, image capture, and related fields is becoming more and more intense. Among these cameras, a rear camera for a vehicle displays necessary rear information, thereby preventing fatal rear collisions. In this respect, a fisheye lens is a natural choice as a rear camera capable of obtaining as wide a view as possible.

A detailed description of prior art related thereto will be made in Korean Laid-Open Publication No. 2011-0103724 entitled " augmented reality realization system using fisheye lens and its implementation method " Discloses an apparatus and method for displaying a part of an image to a user in accordance with a tilt by correcting a distorted shape through a distortion correction algorithm using spherical geometry of an ellipsoid after securing all data of surrounding time using a fisheye lens .

In addition, Korean Patent Laid-Open Publication No. 2008-0053834 (entitled " image distortion conversion method and parking trajectory expression method of a rear camera of a vehicle ") discloses a camera distortion function model of a vehicle rear- (Field Of View) model to improve the realism of the image provided to the driver by eliminating the error of the distortion correction of the camera, and applying the rear camera distortion correction algorithm to the predicted trajectory of the vehicle in the rear image display .

SUMMARY OF THE INVENTION It is an object of the present invention to provide a three-dimensional distortion apparatus and method using a fisheye lens which can be easily performed without directly going to the field, without a three-dimensional simulation of a fisheye lens distortion for a rear camera for a vehicle.

Another object of the present invention is to provide a computer-readable recording medium storing a program for causing a computer to execute a three-dimensional distortion method using a fisheye lens, which can be easily performed without going directly to the scene, The present invention also provides a recording medium that can be read by a computer.

According to an aspect of the present invention, there is provided a three-dimensional distortion apparatus using a fish-eye lens, including: an input unit for inputting a target image; A virtual space generating unit for generating a 3D virtual space for the input target image; And a distortion unit that generates a distorted 3D virtual space by distorting the generated 3D virtual space based on a camera lens specification.

According to another aspect of the present invention, there is provided a three-dimensional distortion method using a fish-eye lens, the method comprising: (a) inputting a target image; An input step; (b) a virtual space creation step of creating a 3D virtual space for the input target image; And (c) a distortion step of distorting the generated 3D virtual space based on the camera lens specification to generate a distorted 3D virtual space.

According to another aspect of the present invention, there is provided a computer readable medium storing a program for causing a computer to execute any one of the above methods.

According to the three-dimensional distortion apparatus and method using the fisheye lens according to the present invention, it is possible to easily perform the three-dimensional distortion of the fisheye lens distortion for the rear camera of the vehicle without directly going to the scene.

1 is a block diagram showing the configuration of a three-dimensional distortion apparatus using a fish-eye lens according to the present invention;
Fig. 2 is a diagram showing radial distortion occurring in a wide-angle lens, Fig.
Figure 3 shows a test environment proposed by NHTSA (National Highway Traffic Safety Administration)
Figure 4 shows a perspective projection camera model,
5 is a view showing a lens on a spherical coordinate system and an image plane on a polar coordinate system,
FIG. 6 is a diagram showing a distorted image result using a virtual space created using an open software library and a three-dimensional distortion method using a fish-eye lens according to the present invention;
FIG. 7 is a view comparing an input image with an image obtained by distorting an input image,
Fig. 8 is a view comparing the results of the fisheye lens distortion simulation.

Hereinafter, preferred embodiments of a three-dimensional distortion apparatus and method using a fisheye lens according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing the configuration of a three-dimensional distortion apparatus 100 using a fish-eye lens according to the present invention. The 3D distortion apparatus 100 using a fish-eye lens according to the present invention has an input unit 110, a virtual space generation unit 120, and a distortion unit 130. A target image is input to the input unit 100. The virtual space creating unit 120 creates a 3D virtual space for the target image input through the input unit 100. [ At this time, an algorithm for creating a 3D virtual space uses an open software library, which is well known to those skilled in the art, and thus a detailed description thereof will be omitted.

The distortion unit 130 generates a 3D virtual space by distorting the 3D virtual space generated based on the lens specifications of the camera that captured the input target image. Here, the lens specification of the camera may be, for example, a height or an angle of the lens, but is not limited thereto. The lens used here is preferably a wide-angle lens such as a fish-eye lens, but is not limited thereto. The angular range of the lens used in the actual simulation is between 44 ° and 55 °, which is most similar to the FOV (Field Of View) of human vision.

A wide-angle lens generates a radial distortion phenomenon which causes a pixel point to move in a radial direction with respect to an input image as shown in FIG. There are two models for the phenomenon of radial distortion.

First, the fish-eye model is based on the following. An image taken with a fisheye lens has a high resolution in the foveal area of the foveal area and a low resolution in the peripheral area thereof.

는 입력 이미지 상의 특정 픽셀이 왜곡된 위치에서의 반경,

는 입력 이미지 상의 특정 픽셀의 반경,

는 스케일링 상수(scaling constant), 그리고, 는 이미지의 왜곡되는 양을 조절하는 변수이다. here,

Is the radius at a location where a particular pixel on the input image is distorted,

Is the radius of a particular pixel on the input image,

Is a scaling constant, Is a variable that controls the amount of distortion in the image.

Next, the FOV (Field Of View) model is based on a simple optical model of a fisheye lens as shown in Equation (2) below.

는 어안 렌즈 카메라의 겉보기 각도(apparent angular) FOV를 의미한다. here,

Means the apparent angular FOV of a fisheye lens camera.

The three-dimensional distortion apparatus 100 using the fisheye lens according to the present invention starts the process of generating a virtual space by detecting the angle of each pixel based on a given lens specification. It detects the intersection of poles in 3D virtual space and infinite lines that are distorted radially. FIG. 3 is a diagram showing a test environment proposed by NHTSA (National Highway Traffic Safety Administration). 3 (a) shows a top view of a test area in which a rear camera and pole location are shown, and Fig. 3 (b) shows a top view of a test area shown at positions 4, 5, 6, And Fig. 3 (c) shows the pawl shown at positions 1, 2, and 3 shown in (a). The unit of the values shown in Fig. 3 is cm.

As described above, the virtual space creation unit 120 for creating a 3D virtual space with respect to an input target image uses an open graphic library. By simulating a regular lens, we inserted the same grid and pole as the actual test environment to create a virtual space. Here, the position and size of the pole are determined based on NHTSA regulations as shown in FIG. The NHTSA test environment standard does not include any conditions for the grid, but is inserted to measure distances.

In the image acquisition process, the specification of the lens determines the amount of refraction, and the distorted image is formed according to the degree of refraction. A simple image generation model based on the perspective projection technique is as shown in FIG.

While the input image plane is displayed in the 2D rectangular coordinate system, the fisheye lens can be represented in a spherical coordinate system because it is preserved as part of the hemisphere. The distortion factor of the fisheye lens is determined by the field number of the lens, so that a distorted angle can be obtained from each pixel of the image from the lens specification data.

)를 아래 수학식 3과 같이 구할 수 있다. The process of creating a 3D virtual space begins by transforming the rectangular coordinates of the image plane in a polar coordinate system. It can be seen that the converted image plane is radially symmetric at the lens center as shown in Fig. At this time, using the coordinates of each pixel on the image plane,

) Can be obtained as shown in the following equation (3).

는 이미지 평면 상의 뷰잉 각도(viewing angle), 그리고,

및

는 이미지 평면 상의 좌표 값을 의미한다. here,

A viewing angle on the image plane,

And

Quot; means a coordinate value on the image plane.

으로 표시되는 필드 수(field number)는 동심 지역(concentric region)의 누적 부분(cumulative portion)으로 정의된다. 통상적으로 렌즈는 중심부터 외부 둘레(outer circumference)까지 10~20개 필드(field)로 나뉜다.

는 렌즈 시야(view)의 최대 각도를 나타낸다. 센서의 크기는

로 정의된다.

로 나타내는 각 픽셀의 뷰잉 각도(viewing angle)를 계산하기 위해 렌즈의 필드 수(

)와 각도(

)를 사용한다. 더욱 구체적으로, 각 렌즈 필드(

)의 각도는 아래 수학식 4를 통해 감지하며, 각 픽셀의 각도(

)는 아래 수학식 5를 통해 결정된다.

The field number is defined as the cumulative portion of the concentric region. Typically, the lens is divided into 10 to 20 fields from the center to the outer circumference.

Represents the maximum angle of the lens view. The size of the sensor

.

To calculate the viewing angle of each pixel represented by the number of fields of the lens (

) And angle (

) Is used. More specifically, each lens field (

) Is detected through the following expression (4), and the angle of each pixel

) Is determined by the following equation (5).

와

를 이용하는 렌즈의 표면 상에 위치한 도 3에 도시된 P 포인트

의 좌표를 계산하기 위해 아래 수학식 6을 이용한다.

Wow

Lt; RTI ID = 0.0 > P-point < / RTI &

The following equation (6) is used.

은 통상적으로 1로 세팅될 수 있다. here,

Lt; RTI ID = 0.0 > 1. ≪ / RTI >

를 감지할 수 있다. 또한, 도 4에 도시된 바와 같이,

는 3D 공간 상에 정의된 선이므로,

과

는 아래 수학식 7 및 수학식 8과 같이 선형 방정식으로 나타낼 수 있다. Then, as shown in Fig. 5, the crossing points of the virtual 3D space and

Can be detected. Further, as shown in Fig. 4,

Is a line defined in the 3D space,

and

Can be represented by a linear equation as shown in Equations (7) and (8) below.

와

는 아래 수학식 9 및 수학식 10으로 표현되는 선의 변수들이다. here,

Wow

Are the parameters of the line expressed by the following equations (9) and (10).

,

)이 교차한다면, 아래 수학식 11을 만족하게 된다. As described above, two lines (

,

), The following equation (11) is satisfied.

를 만족하게 된다. 3D 왜곡된 이미지는 교차 지점에서 발생한 픽셀에서 지정된 색상을 삽입하여 생성된다. Here, the validity of the intersection checking of two lines can be checked using Equations (10) and (11) described above. However, since the grid and the pole are not infinite straight lines

. A 3D distorted image is created by inserting a specified color from a pixel originating at an intersection point.

In order to evaluate the performance of the 3D distortion apparatus 100 using the fisheye lens according to the present invention, the image is evaluated using a 640 x 480 image photographed with a camera having a fisheye lens used as an automobile rear camera. As a result of the simulation, the geometric distortion of each image was generated according to the camera position and viewing angle. 30 cm X 30 cm 3D virtual grid was created and the location of each pillar was selected based on the NHTSA standard. The three-dimensional distortion method using a fish-eye lens according to the present invention is applied to pre-generated virtual space data, and a virtual space (ac) generated by applying an open software library and the corresponding three-dimensional distortion method according to the present invention are applied The resulting distorted result (df) is as shown in Fig.

On the other hand, referring to FIG. 7, the simulated distorted result does not exactly match the actual distorted image because there is no lens that perfectly meets the standard requirement. For this reason, an error occurs when comparing the corresponding feature points between the distorted virtual space and the input image obtained by the fisheye lens camera. That is, as shown in FIG. 7, the three-dimensional distortion method using a fish-eye lens according to the present invention simulates an actual image accurately captured by a fish-eye lens camera. The results of the 3D distortion method using the fisheye lens according to the present invention can be compared with the optical simulation results shown in FIG. During image generation using a number of simulation methods, the 3D distortion method using a fish-eye lens according to the present invention generates an image using 3D information. At this time, the viewing angle used is 170 °.

The present invention can also be embodied as computer-readable codes on a computer-readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like, and a carrier wave (transmission via the Internet). In addition, the computer-readable recording medium may be distributed to a computer system connected to a wired / wireless communication network, and a computer-readable code may be stored and executed in a distributed manner.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation in the embodiment in which said invention is directed. It will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the scope of the appended claims.

100: a three-dimensional distortion apparatus using a fish-eye lens 110:
120: virtual space generation unit 130: distortion unit

Claims (7)

An input unit for inputting a target image;
A virtual space generation unit for generating a three-dimensional virtual space with respect to the target image; And
And a distortion unit for generating a distorted three-dimensional virtual space by distorting the generated three-dimensional virtual space on the basis of a camera installation position and a lens specification for the three-dimensional virtual space,
Wherein the distortion unit calculates a viewing angle of each pixel constituting the three-dimensional virtual space on the basis of a lens specification of the camera based on the installed position of the camera with respect to the three-dimensional virtual space, Dimensional distortion using the fisheye lens. The method according to claim 1,
Wherein the lens specification of the camera is an angle with the number of fields of the lens. 3. The method according to claim 1 or 2,
Wherein the camera lens is a wide-angle lens. In a three-dimensional distortion method performed by a three-dimensional distortion apparatus using a fisheye lens,
(a) receiving a target image;
(b) generating a three-dimensional virtual space for the target image; And
(c) generating a distorted three-dimensional virtual space by distorting the generated three-dimensional virtual space based on a camera installation position and a lens specification for the three-dimensional virtual space,
In the step (c), a viewing angle of each pixel constituting the three-dimensional virtual space is calculated based on the lens specification of the camera based on the installed position of the camera with respect to the three-dimensional virtual space, Dimensional virtual space is generated by using the fisheye lens. 5. The method of claim 4,
Wherein the lens specification of the camera is an angle with the number of fields of the lens. 5. The method of claim 4,
Wherein the camera lens is a wide-angle lens. A computer-readable recording medium having recorded thereon a program for causing a computer to execute a three-dimensional distortion method using the fish-eye lens according to any one of claims 4 to 6.

Images