KR101557271B1 - Method for detecting a circle-type object and approximating a substitute circle based on Image processing - Google Patents

Abstract

The prevent invention relates to a method for approximating a detected circle shape according to the detection of a circle shape on an image and a system therefor. The method for approximating the detected circle shape according to the detection of the circle shape on the image includes an image inputting step (S100), an image pre-processing step (S200), an edge detecting step (S300) of performing edge filtering, an initial coordinate detecting step (S400), an internal coordinate detecting step (S500), a circle coordinate detecting step (S600), and an approximating step (S700). The present invention easily detects the circle shape by an image processing method.

Description

FIELD OF THE INVENTION [0001] The present invention relates to a method of approximating a circle shape based on detection of an original shape in an image,

The present invention relates to a method of approximating a detection circle shape according to detection of an original shape in an image, and more particularly, to a system and method for detecting a circular object from a test image input by a camera, The present invention relates to a method of approximating a detection circle shape according to detection of an original shape in an image capable of reproducing an approximate circle, and a system therefor.

Most of the circular object detection techniques are based on the Hough Transform (HT) proposed by Hough in 1962.

Randomized Hough Transform (RHT), one of the most optimized HTs, has been proposed by Lei Xu and RHT is divided into two processes for circular object detection.

First, the center point of a circular object is estimated from the midpoint of arbitrary three points and the tangent line, and the remaining three variables are determined from the simplified circular object equation.

In the conventional HT, the 5 - dimensional space used for circular object detection is divided into 2 - dimensional space corresponding to the origin of the circular object and 3 - dimensional space corresponding to the remaining variables, thereby reducing the storage space. Also, the calculation time can be significantly reduced by using a method of selecting an arbitrary pixel when selecting a target pixel to be detected.

However, since three edge pixels are arbitrarily selected to determine a circular object in an image, there is a problem that a circular object can be erroneously found, the execution time is proportional to the total number of pixels, and the number of circular objects can not be accurately estimated .

On the other hand, the present invention can not only rapidly detect an image using an image processing technique, but also can approximate a circle that is optimal for the detected circle and extract it.

It can be applied to all applications based on detecting a circular shape or an object of an optical fiber connector or detecting a circular object such as a coin detecting device, face, pupil shape recognition, signal light recognition, There are advantages.

In this paper, we propose a new algorithm that can effectively detect iris regions in anterior segment images by using the simultaneous equations of circles made by using partial information of circumference without changing the threshold value. Lt; / RTI >

A New Circle Detection Algorithm for Iris Domain Segmentation. The Institute of Electronics Engineers of Korea Korea Advanced Institute of Science and Technology (KAIST) (2008) 16 (12) 1385-1392 1229-7771

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems of the conventional art, and it is an object of the present invention to provide a method and apparatus for detecting a circular object from a test image input to a camera, A method of approximating a detection circle shape according to detection of a circular shape in an image that can be reproduced, and a system therefor.

A method of approximating a detection circle shape according to an exemplary embodiment of the present invention includes an image input step (S100) in which a test image acquired using a camera is input, and an image input step (S100) An image preprocessing step (S200) for performing image pre-processing to remove noise included in the input test image, edge filtering (S200) for an inspection image preprocessed by the image preprocessing step (S200) edge detecting step S300 of detecting an edge of the object based on an edge of the object by performing edge filtering on the edge of the object, The boundary coordinate of the object detected by the edge detecting step S300 and the initial center-of-gravity coordinates detected by the initial coordinate detecting step S400 An inner coordinate detection step S500 of detecting an inner coordinate of the object surrounded by the boundary, a step of detecting an inner coordinate of the circle, and a step of detecting a center coordinate and a radius coordinate of the circular shape using the inner coordinate detected by the inner coordinate detection step S500. The inner coordinates of the object surrounded by the boundary detected by the coordinate detecting step S600 and the inner coordinate detecting step S500 and the coordinates of the center of the circle detected by the circular coordinate detecting step S600 And an approximation step (S700) of determining the coordinates of the center of the best approximated circle and the coordinates of the radius.

At this time, the image preprocessing step S200 includes a scaling step S210 for performing a pixel value scaling to make the brightness value of the inspection image constant, and a scaling step S210 for scaling a pixel value scaled by the scaling step S210. And a histogram smoothing step S220 for adjusting the contrast of the histogram.

In addition, the edge detection step S300 performs edge filtering using a bi-lateral filter, and detects a boundary in accordance with the circular object in the inspection image.

In addition, the inner coordinate detection step S500 may determine a straight line according to a predetermined coordinate in the inspection image using the following equation, and then calculate the inner coordinate of the object detected by the edge detection step S300 The boundary coordinates between the boundary coordinates of the object overlapping the linear coordinates and the initial center-of-gravity coordinates are calculated using the boundary coordinates, and the inner coordinates of the object surrounded by the boundary are detected.

That is, a straight line formed by the coordinates (p, q), (x _e , y _e ) in the inspection image is expressed by the following equation.

Here, (p, q), (x _e , y _e ) are predetermined coordinates.

Further, the approximation step S700 may further include an inner coordinate of the object surrounded by the boundary detected by the inner coordinate detecting step S500 and an inner coordinate of the object detected by the circular coordinate detecting step S600 The intersection between the center coordinates of the circular shape and the radial coordinates is maximized, and the coordinates of the center and the radius of the circle of the optimal approximation are determined so as to minimize the difference set.

Here, n (A) is the inner coordinate of the object surrounded by the boundary,

n (B) is the center coordinate of the detected circular shape, the radial coordinate,

S is the minimum value of the difference set.

The system for approximating a detection circular shape according to an in-image circle detection according to an embodiment of the present invention includes an image input unit 100 for receiving and transmitting an inspection image from a camera, An image correction unit 200 for detecting center coordinates and radius coordinates of a circular shape existing in the inspection image by performing image processing, and an image correction unit 200 for detecting a center coordinate and a radius coordinate of the circular shape detected by the image correction unit 200, And an approximate circle determining unit (300) for determining center coordinates and radius coordinates of the best approximated circle using the inner coordinates of the object existing in the inspection image detected by the correction unit (200) do.

The image correcting unit 200 includes an image preprocessing unit 210 for performing image preprocessing to remove noise included in the inspection image transmitted from the image input unit 100, An edge detector 220 for performing edge filtering on an inspection image subjected to image preprocessing by the image preprocessing unit 210 to detect a boundary according to an object in the inspection image, The initial coordinates of the center of gravity are detected based on the boundary of one object, the internal coordinates of the object surrounded by the boundary are detected by using the boundary coordinates of the detected object and the initial center of gravity, And a coordinate detecting unit 230 for detecting the center coordinates and the radial coordinates of the circular shape.

Also, the approximate circle determination unit 300 uses the inner coordinates of the object surrounded by the boundary detected by the coordinate detection unit 230 and the center coordinates and the radial coordinates of the circular shape to calculate the center coordinates and the radius And the coordinates are detected.

A method and an apparatus for approximating a shape of a detection circle according to the present invention detect an original shape in an image and detect a circular shape object from the inspection image input to the camera, And a system for the same.

Through this, it is possible to easily detect the circular shape by the image processing technique on the circular equipment, and the effect can be approximated to the circle that is the most suitable for this circle.

Accordingly, the present invention can be applied to all applications based on detection of circular objects such as circular shape detection and foreign object detection of an optical fiber softening device, coin detection device, face, pupil shape recognition, signal light recognition,

There is an advantage that it is applicable to a device that can detect a circular object obtained through a real image and reproduce it as an artificially approximated circle by computer graphics or the like.

FIG. 1 is a flowchart illustrating a method of approximating a detection circle shape according to an embodiment of the present invention. Referring to FIG.
2 is a block diagram of a system for approximating a detection circle shape according to an embodiment of the present invention.
FIG. 3 is a diagram for explaining a method of approximating a detection circular shape according to an embodiment of the present invention and detecting a circular boundary in a system therefor, and then searching for a center coordinate of a circular shape.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a method of approximating a detection circle shape according to the detection of an original shape in an image and a system therefor will be described in detail with reference to the accompanying drawings. The following drawings are provided by way of example so that those skilled in the art can fully understand the spirit of the present invention. Therefore, the present invention is not limited to the following drawings, but may be embodied in other forms. In addition, like reference numerals designate like elements throughout the specification.

In this case, unless otherwise defined, technical terms and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. In the following description and the accompanying drawings, A description of known functions and configurations that may unnecessarily obscure the description of the present invention will be omitted.

In addition, a system refers to a collection of components, including devices, mechanisms, and means that are organized and regularly interact to perform the required function.

A method for approximating a detection circular shape according to the present invention is a technique for inputting an inspection image, a preprocessing technique for reducing noise on an input inspection image, an edge (boundary of an object, edge detection technique for detecting the center coordinates of the circle using the coordinates of the extracted edges, a technique for detecting the optimal radius in the center coordinates of the circle, a technique for approximating the optimal center using the center and radius of the circle Use technology to find the circle.

1 is a flowchart illustrating a method of approximating a detection original image according to an original image detection in an image according to an embodiment of the present invention. Referring to FIG. 1, a method of approximating a source image according to an original image detection according to an embodiment of the present invention will be described in detail.

FIG. 2 is a block diagram of a system for approximating a source image according to an original image detection according to an embodiment of the present invention. Referring to FIG. 2, A system for a method of approximating a source image according to an image detection will be described in detail.

As shown in FIG. 2, the system for approximating the original image according to the original image detection of the present invention includes an image input unit 100, an image correction unit 200, and an approximate circle determination unit 300 .

The image input unit 100 may receive a test image from a camera that is provided in advance or is included in a terminal held by a manager, and may transmit the test image to the image correcting unit 200.

The image correcting unit 200 may perform image processing on the inspection image transmitted from the image input unit 100 to detect a center coordinate and a radius coordinate of a circle existing in the inspection image.

To be more specific,

As shown in FIG. 2, the image correction unit 200 may include an image preprocessing unit 210, an edge detection unit 220, and a coordinate detection unit 230.

The image preprocessing unit 210 may perform image preprocessing to remove noise included in the t-test image transmitted from the image input unit 100,

The edge detector 220 may perform edge filtering on the inspection image, which is pre-processed by the image preprocessing unit 210, to detect a boundary of an object in the inspection image.

The coordinate detecting unit 230 detects an initial center-of-gravity coordinate based on the boundary of the object detected by the edge detecting unit 220,

Detecting inner coordinates of an object surrounded by the boundary using the boundary coordinates of the detected object and the initial center-of-gravity coordinates,

The center coordinates and the radial coordinates of the circular phenomenon can be detected using the detected inner coordinates.

The approximate circle determination unit 300 uses the coordinates of the center coordinates and radius coordinates of the original shape detected by the image correction unit 200 and the internal coordinates of the objects existing in the inspection image detected by the image correction unit 200 To determine the center coordinates and radius coordinates of the best approximated circle.

In other words, the approximate circle determination unit 300 uses the inner coordinates of the object surrounded by the boundary detected by the coordinate detection unit 230 and the center coordinates and the radial coordinates of the circular phenomenon, The radius coordinates can be detected.

Each of the operations performed by the image input unit 100, the image correction unit 200, and the approximate circle determination unit 300 will be described in detail below with reference to a method of approximating a source image according to the present invention's original image detection .

As shown in FIG. 1, a method of approximating a source image according to the present invention includes an image input step S100, an image preprocessing step S200, an edge detection step S300, (S400), an inner coordinate detection step (S500), a circular coordinate detection step (S600) and an approximation step (S700).

To learn more about each step,

The image input step S100 may receive a test image acquired by the image input unit 100 using a camera.

The inspection image means a captured still image of an image input from a camera or one frame of a moving image.

At this time, since the maximum value and the minimum value of the brightness are different according to the illumination and the surrounding environment, it is necessary to adjust the maximum value and the minimum value to be constant.

In order to do so, the image correcting unit 200 may perform the image preprocessing step S200.

The image preprocessing step S200 may perform image preprocessing to remove noise included in the inspection image input by the image input step S100.

The image preprocessing step S200 may include a scaling step S210 and a histogram smoothing step S220.

The scaling step (S210) may perform pixel value scaling to make the brightness value of the inspection image constant.

In detail, pixel value scaling can be performed by applying Equation (1) below to adjust the maximum value and minimum value of the brightness of the inspection image.

Here, I (i, j) means a brightness value of the inspection image at an arbitrary spatial position (i, j)

I _max and I _min mean the maximum value and the minimum value of the brightness value of the inspection image,

(8비트 영상인 경우 L=256임.)를 의미한다.
L is the number of stages of the pixel value

(L = 256 in case of 8-bit image).

In this case, even if the pixel value scale change is performed by performing the scaling step S210 through Equation (1), since the quality is not visually improved, the range of the pixel value is obtained as a wide high contrast, In order to improve the quality of the image, the histogram smoothing step S220 may be performed.

The histogram smoothing step S220 may perform contrast adjustment of the inspection image that has been scaled by the scaling step S210.

The histogram smoothing step S220 may have a uniform distribution density of brightness values in order to obtain wide high contrast and improve visual quality.

That is, the pixel value of the inspection image can be matched as shown in Equation (2).

Here, N is the total number of pixels,

n _j denotes the number of pixel values j.

The edge detection step S300 may include edge filtering for the test image pre-processed by the image preprocessing step S200, i.e., the scaling step S210 and the histogram smoothing step S220, edge filtering may be performed to detect boundaries according to objects in the inspection image.

The edge detection step S300 may perform edge filtering using a bidirectional filter, and may detect an edge of the original object in the inspection image.

More specifically,

The edge detection step S300 may pass the bidirectional filter to easily detect the edge of the circular object included in the inspection image, and to detect the circular shape well.

In this case, the edge detection step S300 may perform edge detection of the circular object in the inspection image by bidirectional filtering using Equation (3).

이며,here,

Lt;

이며,

Lt;

이며,

Lt;

P1 and P2 denote the tap size of the bidirectional filter,

는 커널의 중앙으로부터 떨어진 화소의 거리의 가중평균을 위한 가우시안 함수의 표준편차이고,

Is the standard deviation of the Gaussian function for the weighted average of the distances of the pixels away from the center of the kernel,

는 화소값의 가중평균을 위한 가우시안 함수의 표준편차이다.

Is the standard deviation of the Gaussian function for the weighted average of pixel values.

In other words, the edge detection step S300 may perform edge detection of an object in the inspection image by bidirectional filtering.

The edge filtering operation is as follows,

Use the Sobel operator to calculate the size of the gradient vector,

To get a slender edge, use the 3x3 window to keep the gradient size at the maximum value in the gradient vector direction and keep the remainder to 0,

Using the two thresholds to get connected edges, we can use a hysteresis threshold scheme that traces edges until a low threshold is reached in the gradient direction with a high threshold value.

The initial coordinate detection step S400 may detect the initial center-of-gravity coordinates based on the boundary of the object detected by the edge detection step S300 in the image processing unit 200. [

That is, the edge detected in the edge detection step S300 is detected in the form of a connected sequence, and if it is assumed that the number of coordinates used to represent the connected edge is M, the edge can be expressed by the following equation (4) .

The initial center-of-gravity coordinates (x _e , y _e ) detected using the coordinates of the edge as shown in Equation (4) can be expressed by Equation (5).

If the initial gravity center coordinate is determined by the initial coordinate detection step S400, the internal coordinate detection step S500 may be performed to obtain an internal coordinate set for the object surrounded by the edge.

The inner coordinate detection step S500 may be performed by using the boundary coordinates of the object detected by the edge detection step S300 and the initial center-of-gravity coordinates detected by the initial coordinate detection step S400 in the image processing unit 200 To detect the inner coordinates of the object surrounded by the boundary.

That is, after determining a straight line according to coordinates previously set in the inspection image using Equation (6) below, using the boundary coordinates of the object detected by the edge detection step (S300) overlapping with the straight line coordinate, The distance between the boundary coordinates of the object overlapping the coordinates and the initial center-of-gravity coordinates is calculated, and the inner coordinates of the object surrounded by the boundary can be detected.

In other words, the inner coordinate detection step S500 may calculate a straight line made up of coordinates (p, q) and (x _e , y _e ) arbitrarily given in the inspection image.

If the coordinates of all the points on the straight line obtained through Equation (6) are denoted by F, it is possible to detect the inner coordinates of the object surrounded by the edges by taking into consideration the case of being included in F and included in the edge coordinates at the same time.

Accordingly, it is possible to calculate a straight line between the coordinates (p, q) and the center of gravity coordinates (x _e , y _e ) given beforehand by measuring the distances to the center of gravity while scanning the coordinates of the inspection image,

The edge coordinates on the straight line are obtained,

By calculating the distance between the edge coordinates on the straight line and the coordinates of the center of gravity, the set of internal coordinates of the object surrounded by the edge can be defined by the following equation (7).

After the inner coordinates of the object surrounded by the edge are determined, a process of determining the coordinates inside the circular shape and the circular shape, i.e., the circular coordinate detection step (S600) is required.

The circular coordinate detection step S600 can detect the center coordinates and the radial coordinate of the circular motion using the inner coordinates detected by the inner coordinate detection step S500 in the image processing unit 200. [

In other words, the original coordinate detection step (S600) is the initial center coordinate (x _e (0) of the detection coordinates the coordinates of the center of the circle-shaped through the above expression (5) and equation (6) circular-shaped, y _e (0 )).

x _e (0) = x _e , y _e (0) = y _e ,

The initial radius of the circular shape can be determined by the following equation (8).

Using this, the set of points made up of the circular inner coordinates can be determined as shown in Equation (9).

If the inner coordinates of the circle and the circle are determined, the approximate circle extractor 300 can find a circle that optimally approximates the edge through the approximation step S700.

In the approximating step S700, the inner coordinates of the object surrounded by the boundary detected by the inner coordinate detecting step s500 and the center coordinates and the radial coordinates of the circular shape detected by the circular coordinate detecting step S600 are used To determine the center coordinates and radius coordinates of the best approximated circle.

More specifically, the inner coordinates of the object surrounded by the boundary detected by the inner coordinate detecting step (S500) and the center coordinates of the circle detected by the circular coordinate detecting step (S600) are calculated using the following equation (10) The intersection between the radial coordinates is maximized, and the center and radial coordinates of the best approximated circle can be determined so that the difference set is minimized.

Here, in order to obtain the coordinates of the points that can be optimally approximated, the number of coordinates inside the circle-shaped object surrounded by the edges is denoted by n (A), the number of coordinates within the circle can be denoted by n (B) These values can be expressed by the number of pixels satisfying the above equations (7) and (9), respectively.

Also, the number of common coordinates formed by intersection of set A and set B can be expressed as n (A∩B).

The initial cost using this can be expressed as follows.

Equation (10) determines the ratio such that the difference between the inner coordinates represented by the edges and the inner coordinates expressed by the circles approximated to each other is minimized.

와 같이 변화될 경우,If the initially determined center coordinates and radius of the circular shape are varied within a certain range, that is,

If it changes,

The minimum value S determined by Equation (10) can be calculated.

라고 표현할 경우, 최적으로 근사화되는 원의 중심좌표 및 반지름 좌표는 하기의 수학식 11과 같이 나타낼 수 있다.
The minimum value S is used to calculate the variable amount

, The coordinates of the center and the radius of the circle that are optimally approximated can be expressed by Equation (11) below.

That is, in other words, a method of approximating a detection circular shape according to an embodiment of the present invention and a system therefor, detects a circular shape among objects existing in a digital image, Can be easily drawn.

In addition, an image of a cross-section of a circular device is input from a camera, a circular shape is detected through a predetermined image processing technique, and a lot of noise included therein is removed to generate an algorithm for circular detection can do.

In addition, by finding the center and the radius of the approximate circle most corresponding to the detected circle shape, an algorithm for drawing an optimal approximate circle can be easily generated.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, And various modifications and changes may be made thereto by those skilled in the art to which the present invention pertains.

Accordingly, the spirit of the present invention should not be construed as being limited to the embodiments described, and all of the equivalents or equivalents of the claims, as well as the following claims, belong to the scope of the present invention .

100:
200:
300: Approximate circle determining unit

Claims (8)

An image input step (S100) in which a test image acquired using a camera is input;
An image preprocessing step (S200) for performing image pre-processing to remove noise included in the inspection image input by the image input step (S100);
An edge detection step (S300) of performing edge filtering on the inspection image preprocessed by the image preprocessing step (S200) to detect a boundary according to the object in the inspection image;
An initial coordinate detecting step (S400) of detecting an initial center-of-gravity coordinate based on a boundary of the object detected by the edge detecting step (S300);
An inner coordinate detection unit for detecting an inner coordinate of an object surrounded by the boundary using the boundary coordinates of the object detected by the edge detection step (S300) and the initial center-of-gravity coordinates detected by the initial coordinate detection step (S400) Step S500;
A circular coordinate detection step (S600) of detecting a center coordinate and a radius coordinate of the circular shape using the inner coordinate detected by the inner coordinate detection step (S500); And
Using the inner coordinates of the object surrounded by the boundaries detected by the inner coordinate detecting step (S500) and the center coordinates and the radial coordinates of the circular shape detected by the circular coordinate detecting step (S600), an optimal approximated circle An approximation step (S700) of determining the center coordinates and the radius coordinates;
Lt; / RTI >
The internal coordinate detection step (S500)
A straight line according to a predetermined coordinate is determined in the inspection image using the following equation,
The distance information between the boundary coordinates of the object overlapping the linear coordinates and the initial center-of-gravity coordinates is calculated using the boundary coordinates of the object detected by the edge detection step (S300) overlapping with the linear coordinates, And detecting an inner coordinate of the detected circular shape.

Here, (p, q), (x _e , y _e ) are predetermined coordinates.
The method according to claim 1,
The image preprocessing step (S200)
A scaling step (S210) of performing pixel value scaling to make the brightness value of the inspection image constant; And
A histogram smoothing step (S220) of performing contrast adjustment of the inspection image in which pixel value scaling is performed by the scaling step (S210);
And an approximation method of the detection circular shape according to the circular shape detection in the image.
The method according to claim 1,
The edge detection step (S300)
Edge filtering is performed using a bi-lateral filter,
Wherein an edge of the circular object is detected in the inspection image.
delete The method according to claim 1,
In the approximation step S700,
An intersection between an inner coordinate of an object surrounded by the boundary detected by the inner coordinate detecting step (S500) and a center coordinate and a radial coordinate of a circle detected by the circular coordinate detecting step (S600) using the following equation And a center coordinate and a radius coordinate of an optimal approximated circle are determined so that the difference set is minimized, and the approximation method of the circle shape according to the circular shape detection in the image.

Here, n (A) is the inner coordinate of the object surrounded by the boundary,
n (B) is the center coordinate of the detected circular shape, the radial coordinate,
S is the minimum value of the difference set.
delete delete delete

Images