KR102143918B1 - Apparatus and method for detecting LED edge based on adaptive threshold - Google Patents

Abstract

The present invention relates to a method for extracting an LED boundary by receiving an external LED visual light signal in visible light communication between vehicles. The method for extracting an LED boundary comprises the following steps of: generating an LED image frame including a plurality of all points based on reception of an external led visible light signal; extracting gradient values of the plurality of all points and mapping the gradient values for each of the all points; extracting points having a gradient value greater than a first threshold value among the all points as boundary candidate points; setting the gradient value of a point having a maximum between-class variance value among the extracted boundary candidate points as a second threshold value; and classifying points having a gradient value greater than the second threshold value among the boundary candidate points as boundary points.

Description

BACKGROUND OF THE INVENTION 1. A method and apparatus for detecting LED edge between vehicles based on adaptive thresholding.

Hereinafter, a technology for detecting an LED boundary by receiving an external LED visible light signal in LED visible light communication between vehicles is provided.

Visible light communication (VLC) is a newly emerging technology applied to traffic systems and is called V2LC (Vehicular Visible Light Communication). V2LC uses Light Emitting Diode (LED) to transmit data from vehicle-to-vehicle or vehicle-to-infrastructure communications to photodiodes or high-speed vehicle cameras.

V2LC offers greater bandwidth, lower installation cost, lower complexity and higher data rates compared to other short-range vehicle technologies (e.g. Dedicated Short Range Communication). In addition, V2LC can prevent mutual signal interference, especially in high-density traffic. These advantages allow V2LC to play an important role in the Intelligent Transportation System (ITS), an innovative advanced application that effectively reduces traffic accidents and congestion and provides a safer and safer transportation network.

Korean Patent Registration Publication No. 10-0736356 (Registration date: June 29, 2007) Korean Patent Publication No. 10-2014-0096880 (Publication date: August 06, 2014)

The LED boundary extraction method according to an embodiment comprises the steps of generating an LED image frame including a plurality of all points based on reception of an external LED visible light signal, extracting a gradient value of the plurality of all points, and Mapping the gradient value for each of the points, extracting points having a gradient value greater than the first threshold among all points as boundary candidate points, the largest inter-class variance among the extracted boundary candidate points Setting a gradient value of a point having a (between-class variance) value as a second threshold value, including classifying points having a gradient value greater than the second threshold among the boundary candidate points as boundary points can do.

According to another embodiment, the LED boundary extraction method may further include classifying points having a gradient value that is greater than the first threshold value and less than the second threshold value among the boundary candidate points as boundary uncertain points. have.

In addition, the LED boundary extraction method comprises the step of reclassifying the boundary uncertain point into the boundary point in response to the boundary point being at a position adjacent to the boundary uncertain point, and the non-border at a position adjacent to the boundary uncertain point In response to the case where there are only points, the step of reclassifying the boundary uncertain point into the non-boundary point may be further included.

Setting the second threshold value according to one side may include sequentially calculating the inter-class variance values for the boundary candidate points.

The calculating of the inter-class variance value according to the other side may include sequentially selecting a target gradient value among gradient values of the boundary candidate points, and calculating an inter-class variance value for the selected target gradient value. I can.

In addition, calculating the inter-class variance value for the selected target gradient value may include classifying a temporary boundary point and a temporary boundary uncertainty point from the boundary candidate points based on the target gradient value, and an average value of the temporary boundary points , A weight, an average value of the temporary boundary uncertainty point, and a step of calculating the inter-class variance value based on at least one of the weight.

According to an embodiment, sequentially calculating the interclass variance values for the boundary candidate points may include calculating a distribution of the gradient values of the boundary candidate points.

The LED boundary extraction method according to one side may further include setting a median value as the first threshold value in the gradient intensity range of the entire points.

1 is a flowchart illustrating a method of extracting an LED area including a method of extracting an LED boundary according to an exemplary embodiment.
2 is a flowchart illustrating a method of extracting an LED boundary according to an embodiment.
3 is a flowchart illustrating a method of setting a second threshold value according to an exemplary embodiment.
4 is a graph showing a distribution of boundary candidate points for setting a second threshold value according to an embodiment.
5 is a diagram illustrating an LED boundary extracted based on a first threshold value and a second threshold value according to an exemplary embodiment.
6 is a block diagram showing an outline configuration of an LED boundary extraction apparatus according to an embodiment.

Specific structural or functional descriptions of the embodiments are disclosed for purposes of illustration only, and may be changed in various forms and implemented. Accordingly, the embodiments are not limited to a specific disclosure form, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical idea.

Terms such as first or second may be used to describe various components, but these terms should be interpreted only for the purpose of distinguishing one component from other components. For example, a first component may be referred to as a second component, and similarly, a second component may be referred to as a first component.

When a component is referred to as being "connected" to another component, it is to be understood that it may be directly connected or connected to the other component, but other components may exist in the middle.

Singular expressions include plural expressions unless the context clearly indicates otherwise. In the present specification, terms such as "comprise" or "have" are intended to designate that the described feature, number, step, action, component, part, or combination thereof is present, but one or more other features or numbers, It is to be understood that the possibility of addition or presence of steps, actions, components, parts, or combinations thereof is not preliminarily excluded.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the relevant technical field. Terms as defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and should not be interpreted as an ideal or excessively formal meaning unless explicitly defined in the present specification. Does not. Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. The same reference numerals in each drawing indicate the same members.

1 is a flowchart illustrating a method of extracting an LED area including a method of extracting an LED boundary according to an exemplary embodiment.

In general, the boundary can form the outline of the object. The boundary of the image may be defined as a point (pixel) that shows a discontinuity and the brightness or intensity of the image changes rapidly. That is, it can represent the boundary between the object and the background. The process of the edge detection algorithm applies a Gaussian filter to reduce noise, determines the intensity gradient of the image, reduces the risk of detecting false edges without suppressing the maximum, and uses a double threshold to detect potential edge points. It can consist of the last five steps to identify. By tracking the boundary with hysteresis, all weak or isolated candidates can be removed. This step is illustrated by FIG. 1. The adaptive threshold technology contributes to the double threshold step to handle the LED intensity function.

The edge detection apparatus may apply a Gaussian filter that convolves with the input image to reduce image noise and prevent false detection. A Gaussian filter having a kernel size of 5 may be used. If the input image is represented by A and the smoothed image is represented by B, A and B may have a relationship of Equation 1.

는 수평 방향을 나타내는 x 축에 대해 도함수이고

는 수직 방향을 나타내는 y 축에 대해 도함수라고 가정한다. 평활화 된 이미지 B의

및

를 계산하기위한 방정식은 수학식 2와 같다.In computer vision, an image gradient is defined as a change in the intensity or color direction of an image and is one of the essential elements for finding the edges. At each point in the image, the magnitude and direction of the intensity gradient can be calculated. The direction determines the boundary direction and the magnitude value can determine whether this point is on the boundary. If this value is high, the intensity changes rapidly, which means that there is an edge at the point, whereas if there is no significant change in the intensity, it may mean that it is not at the boundary.

Is the derivative of the x-axis representing the horizontal direction

Is assumed to be a derivative with respect to the y-axis indicating the vertical direction. Of the smoothed image B

And

Equation 2 is the equation for calculating.

및

가 획득 된 후, 그레디언트 크기 g 및 방향

은 각각 수학식 3과 같이 식별된다.

And

After is obtained, the gradient size g and direction

Are each identified as in Equation 3.

Consider the case where the gradient value generated in the previous step creates a blurred boundary. The end result should ideally have a thin border. That is, only the point where the gradient value represents the strongest intensity value change should be preserved. This reduces computational complexity and supports the next step, resulting in significantly better results. For this reason, suppression can be performed to reduce the risk of false responses to boundary detection by suppressing unwanted points that may not constitute an actual boundary that is not the maximum. Non-maximum suppression can compare the gradient magnitude of each candidate edge point with eight connected adjacent regions in positive and negative gradient directions. If the value is the largest, candidate edge points can be preserved. Otherwise this point is not displayed. For example, when the gradient direction is horizontal, a point whose gradient size value is larger than points in the left and right directions may be regarded as possible boundary points.

After non-maximum suppression is applied, some points due to noise or color change are still present and may produce false boundary detection results. In order to solve the above problem, a method to be described later may use two values of a first threshold value and a second threshold value, respectively, to classify all image points into three classes of boundary points, non-boundary points, and boundary uncertain points. . All points having a gradient value smaller than the first threshold value are displayed as non-boundary points, and may be filtered as a background for example. Points having a gradient value higher than the second threshold value may be identified as boundary points. The remaining points having a gradient value between the first threshold value and the second threshold value are boundary uncertain points, and may or may not be boundary points. A method of classifying boundary points by classifying them into three classes will be described in detail with reference to FIGS. 2 to 6.

2 is a flowchart illustrating a method of extracting an LED boundary according to an embodiment.

The point inside the LED can be the brightest point. Therefore, the intensity of the points inside the LED is higher than that of other points, which may mean that the magnitude of the gradient at the boundary is relatively high. The conventional technique generates an image threshold value based on clustering, but has a disadvantage in that it is suitable only for images having a bi-modal gradient amplitude histogram. In this case, the brightest object is to distinguish the background from other objects. Therefore, the boundary uncertainty points may belong to the boundary of the LED, belong to the boundary of another object, or simply be part of the surrounding noise.

Before step 210, the processor performing the LED boundary extraction method may generate an LED image frame including a plurality of all points based on reception of an external LED visible light signal. The external LED visible light signal may be received in an image format through an LED photographing device. A plurality of points may be formed for an image including the received visible light signal. The plurality of points may be unit pixels, and the size of the unit pixel may be determined according to an image size including a visible light signal.

The processor may extract gradient values of all points and map the gradient values to all points. The gradient value may be extracted through a method of extracting an image gradient value by a person skilled in the art. The processor may extract a gradient value of each point among all points, and the memory may map the gradient value and points of each point one by one.

In step 210, the processor may set a median value in the gradient intensity range of all points as the first threshold value. The median value may be a value that becomes the middle of the total number, and when the total number is odd, the number of points with a value less than the median value and the number of points with a value greater than the median value may be the same based on the median value . The processor may classify all points into a boundary candidate point and a non-boundary point based on the set first threshold value. The processor may extract points having a gradient value greater than the first threshold among all points as boundary candidate points.

)은 m의 절반으로 설정될 수 있다.According to an embodiment, the gradient intensity range of all points is [0, m], m is the maximum value, and the first threshold value (

) Can be set to half of m.

According to another embodiment, the processor may set an average value in the gradient intensity range of all points as the first threshold value.

In step 220, the processor may classify points whose gradient value is less than or equal to the first threshold among all points as non-boundary points. The processor may determine that the non-boundary points do not correspond to the boundary of the LED in the image frame and are portions excluding the boundary. For example, the non-boundary point may be determined as a background outside the LED in the image frame.

In step 230, the processor may classify boundary candidate points into boundary uncertain points and boundary points based on the set second threshold value. The step of setting the second threshold will be described in detail with reference to FIG. 3.

In operation 240, the processor may classify points having a gradient value equal to or less than the second threshold among the boundary candidate points as boundary uncertain points. The processor may reclassify the boundary uncertainty points based on surrounding points of the boundary uncertainty points in the image.

According to an embodiment, the processor may reclassify the boundary uncertain point into a boundary point in response to a case where a boundary point exists at a position adjacent to the boundary uncertain point. On the other hand, the processor may reclassify the boundary uncertain point as a non-boundary point in response to a case where only non-boundary points exist in a position adjacent to the boundary uncertain point. The position adjacent to the boundary uncertain point may be a position at which points surrounding the boundary uncertain point in the image are located. For example, the point may be a square pixel, and the points at a position adjacent to the boundary uncertain point may be eight square pixels surrounding the boundary uncertain point. When at least one of the eight square pixels is a boundary point, the processor may reclassify the boundary uncertainty point into a boundary point. On the other hand, when the eight square pixels are non-boundary points, the processor may reclassify the boundary uncertainty point into a non-boundary point.

In operation 250, the processor may classify points whose gradient value exceeds the second threshold among the boundary candidate points as boundary points. The processor may determine that the boundary points correspond to the boundary of the LED in the image frame.

3 is a flowchart illustrating a method of setting a second threshold value hi according to an exemplary embodiment.

Before step 310, the processor may classify points having a gradient value exceeding the first threshold among all points as boundary candidate points. The processor may calculate a distribution of the gradient values of the boundary candidate points. According to an embodiment, the distribution of the gradient value may follow the distribution of the graph shown in FIG. 4. The processor may calculate a distribution of the gradient values of the boundary candidate points, and assign the boundary candidate points to the boundary candidate points in ascending order from the point having the smallest gradient value.

+1 일 수 있다. i는

+1 부터 m까지 순차적으로 선택될 수 있고, i가 m이 될 때까지 제2 문턱 값(hi)을 계속 계산할 수 있다.In step 310, among the gradient values of the boundary candidate points, the target gradient value i may be sequentially selected from the smallest value to the largest value. Since the boundary candidate points are points having a gradient value exceeding the first threshold value, the smallest gradient value of the boundary candidate points is

Can be +1 i is

From +1 to m may be sequentially selected, and the second threshold value (hi) may be continuously calculated until i becomes m.

In step 320, the processor may calculate an inter-class variance value for the target gradient value i. The processor may classify the boundary candidate points into temporary boundary points and temporary boundary uncertain points based on the target gradient value i. The processor may calculate the inter-class variance value Var based on at least one of an average value of temporary boundary points, a weight, an average value of temporary boundary uncertain points, and a weight. A method of calculating the inter-class variance value Var will be described in detail later in FIG. 4.

In step 330, the processor may compare the inter-class variance value Var of the target gradient value i calculated in step 320 with the maximum inter-class variance value Max_Var. The maximum inter-class variance value Max_Var may be the largest variance value among the sequentially calculated inter-class variance values Var. When the inter-class variance value Var of the target gradient value i is greater than the maximum inter-class variance value Max_Var, the processor performs step 340, and the inter-class variance value Var of the target gradient value i ) Is less than or equal to the maximum inter-class variance value (Max_Var), the processor may perform step 350.

In step 340, the processor may reassign the inter-class variance value (Var) of the target gradient value (i) as the maximum inter-class variance value (Max_Var), and the temporary second threshold value (hi) as the target gradient value ( i) can be redesignated.

In step 350, the target gradient value i may be reassigned to the next target gradient value i+1. Thereafter, the process returns to step 310 and the process of steps 310 to 350 may be repeatedly performed until the target gradient value i becomes m.

When the target gradient value i becomes m, the processor may set the second threshold value that has been continuously redesignated as the final second threshold value in step 360.

According to the previous method, since the threshold value was manually selected through experience, the algorithm may be complex and may lack flexibility in processing a variety of different images. Also, the selected threshold value may have to approximately reflect image characteristics.

4 is a graph showing a distribution of boundary candidate points for setting a second threshold value according to an embodiment.

)의 평균값, 가중치, 임시 경계 포인트들에 대한 그라디언트 크기 범위(

)의 평균값, 및 가중치 중 적어도 하나에 기초하여 상기 클래스 간 분산값(Var)을 계산할 수 있다.The processor has a range of gradient sizes for temporary boundary uncertainty points (

), the weight, and the gradient size range for temporary boundary points (

The inter-class variance value Var may be calculated based on at least one of an average value of) and a weight value.

The graph shown in FIG. 4 is a distribution diagram of a gradient value of points, an x-axis may be a gradient value, and a y-axis may be the number of points or a ratio of the number of points. Accordingly, the distribution diagram may indicate a distribution probability or distribution ratio for a gradient value.

) 및 임시 경계 포인트들에 대한 그라디언트 크기 범위(

)의 가중치(weights)는 수학식 5로 계산될 수 있다.First, the gradient size range for temporary boundary uncertainty points (

) And the gradient size range for temporary boundary points (

The weights of) can be calculated by Equation 5.

는 대상 그라디언트 값(i)에 해당하는 확률일 수 있다. 프로세서는 제2 문턱 값(hi, 410)을 기준으로,

및

의 그레디언트 크기 범위를 [lo + 1, hi] 및 [hi + 1, m]로 분리시킬 수 있다.

May be a probability corresponding to the target gradient value i. The processor is based on the second threshold value (hi, 410),

And

The gradient size range of can be separated into [lo + 1, hi] and [hi + 1, m].

) 및 임시 경계 포인트들에 대한 그라디언트 세기 범위(

) 의 평균값(420, 421)을 수학식 6으로 계산할 수 있다.The processor determines the gradient intensity range for temporary boundary uncertainty points (

) And the gradient intensity range for temporary boundary points (

The average values 420 and 421 of) can be calculated by Equation 6.

) 및 임시 경계 포인트들에 대한 그라디언트 세기 범위(

)의 분산값(variance)을 수학식 7로 계산할 수 있다.The processor determines the gradient intensity range for temporary boundary uncertainty points (

) And the gradient intensity range for temporary boundary points (

The variance of) can be calculated by Equation 7.

) 및 임시 경계 포인트들에 대한 그라디언트 세기 범위(

)의 혼합 평균값을 수학식 8로 계산할 수 있다.The processor determines the gradient intensity range for temporary boundary uncertainty points (

) And the gradient intensity range for temporary boundary points (

The mixed average value of) can be calculated by Equation 8.

)은 관련 가중치 (

의 클래스 내 분산과

의 클래스 내 분산의 합)로 곱한 두 분산의 합으로 정의될 수 있다.Within the variance of these two classes, the intraclass variance (

) Is the associated weight (

With variance within the class of

It can be defined as the sum of two variances multiplied by the sum of the variances within the class of.

) 및 임시 경계 포인트들에 대한 그라디언트 세기 범위(

)의 클래스 간 분산값(

)을 수학식 10을 통해 계산할 수 있다.The processor determines the gradient intensity range for temporary boundary uncertainty points (

) And the gradient intensity range for temporary boundary points (

Variance between classes of (

) Can be calculated through Equation 10.

)이 최대로 만들 수 있다. 클래스 간 분산값은 임시 경계 불확실 포인트들에 대한 그라디언트 세기 범위(

) 및 임시 경계 포인트들에 대한 그라디언트 세기 범위(

)의 가중치 및 평균값(421, 420)을 이용하여 간단하게 계산할 수 있다. 이에 따라, 클래스 내 분산값(

) 대신, 클래스 간 분산값(

)을 이용하여 프로세서는 제2 문턱 값(410)을 간단히 계산할 수 있다.The processor may determine a gradient value minimizing the intra-class variance value as hi from among the candidate gradient values of the second threshold value. According to Equation 10, the optimal hi value is the variance between classes (

) Can be made to the maximum. The variance between classes is the gradient intensity range for temporary boundary uncertain points (

) And the gradient intensity range for temporary boundary points (

) Can be calculated simply by using the weight and average values 421 and 420. Accordingly, the variance value within the class (

), instead of the variance between classes (

Using ), the processor may simply calculate the second threshold value 410.

5 is a diagram illustrating extraction of an LED boundary based on a first threshold value and a second threshold value according to an exemplary embodiment.

FIG. 5A is an original image frame before LED border extraction, and FIG. 5B is an image frame extracted from LED borders according to the embodiment described above with reference to FIGS. 1 to 4. 5C is an image frame from which an LED boundary is extracted according to an exemplary embodiment having a fixed first threshold value and a second threshold value. According to the third image from above, in (a) body skin other than the LED reflects light, so that the boundary between the body skin and the background may have a high gradient value. In (b), the boundary between the body skin and the background was not extracted as the LED boundary, whereas in (c), the boundary between the body skin and the background was extracted as the LED boundary. Therefore, rather than a method of extracting an LED boundary using a fixed first threshold value and a second threshold value, extracting the LED boundary by applying the first threshold value and the second threshold value fluidly and adaptively You can see that it is more accurate.

In order to evaluate the operation of the method of extracting the LED boundary by flexibly applying the first threshold value and the second threshold value, a signal-noise ratio (SNR) may be calculated through Equation 11.

Equation 11

는 LED 경계 추출 후 이미지 프레임 결과에 관한 데이터일 수 있다.I is the original image in MXN size,

May be data related to the image frame result after the LED boundary is extracted.

6 is a block diagram showing the general configuration of the LED boundary extraction device 600 according to an embodiment.

The LED boundary extraction device 600 may include at least one of an image generator 610, a memory 620, and a processor 630. The image generator 610 may generate an LED image frame including a plurality of all points based on reception of an external LED visible light signal.

The memory 620 may map and store gradient values for all points. Also, the memory 620 may temporarily store the first threshold value and the second threshold value. In the process of setting the second threshold value described with reference to FIG. 3, the memory 620 may at least temporarily store the temporary second threshold value, and may continuously replace and store the temporary second threshold value.

The processor 630 may load data stored in the memory 620 and perform the process described in FIGS. 1 to 5. The processor 630 may extract gradient values of a plurality of all points. Also, the processor 630 may extract points having a gradient value greater than the first threshold value among all points as boundary candidate points. The processor 630 may set a gradient value of a point having a maximum between-class variance value among the extracted boundary candidate points as the second threshold value. Also, the processor 630 may classify points having a gradient value greater than the second threshold among the boundary candidate points as boundary points. Since the method of performing each process by the processor 630 has been described above with reference to FIGS. 1 to 5, detailed descriptions will be omitted.

The embodiments described above may be implemented as a hardware component, a software component, and/or a combination of a hardware component and a software component. For example, the devices, methods, and components described in the embodiments are, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA). array), programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions, such as one or more general purpose computers or special purpose computers. The processing device may execute an operating system (OS) and one or more software applications executed on the operating system. In addition, the processing device may access, store, manipulate, process, and generate data in response to the execution of software. For the convenience of understanding, although it is sometimes described that one processing device is used, one of ordinary skill in the art, the processing device is a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that it may include. For example, the processing device may include a plurality of processors or one processor and one controller. In addition, other processing configurations are possible, such as a parallel processor.

The software may include a computer program, code, instructions, or a combination of one or more of these, configuring the processing unit to behave as desired or processed independently or collectively. You can command the device. Software and/or data may be interpreted by a processing device or to provide instructions or data to a processing device, of any type of machine, component, physical device, virtual equipment, computer storage medium or device. , Or may be permanently or temporarily embodyed in a transmitted signal wave. The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -A hardware device specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of the program instructions include not only machine language codes such as those produced by a compiler but also high-level language codes that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

As described above, although the embodiments have been described by the limited drawings, a person of ordinary skill in the art can apply various technical modifications and variations based on the above. For example, the described techniques are performed in an order different from the described method, and/or components such as a system, structure, device, circuit, etc. described are combined or combined in a form different from the described method, or other components Alternatively, even if substituted or substituted by an equivalent, an appropriate result can be achieved.

Claims (10)

Generating an LED image frame including a plurality of all points based on reception of an external LED visible light signal;
Extracting gradient values of the plurality of all points and mapping the gradient values for each of the all points;
Extracting points having a gradient value greater than a first threshold value among the total points as boundary candidate points;
Setting a gradient value of a point having a maximum between-class variance value among the extracted boundary candidate points as a second threshold value;
Classifying points having a gradient value greater than the second threshold among the boundary candidate points as boundary points
LED boundary extraction method comprising a. The method of claim 1,
Classifying points of the boundary candidate points having a gradient value greater than the first threshold value and less than the second threshold value as boundary uncertainty points
LED boundary extraction method further comprising a. The method of claim 2,
Reclassifying the boundary uncertain point into the boundary point in response to the boundary point being located at a position adjacent to the boundary uncertain point; And
Reclassifying the boundary uncertain point into the non-boundary point in response to a case where only non-boundary points exist in a position adjacent to the boundary uncertain point.
LED boundary extraction method further comprising a. The method of claim 1,
Setting the second threshold value,
Sequentially calculating the inter-class variance values for the boundary candidate points
LED boundary extraction method comprising a. The method of claim 4,
The step of calculating the variance value between the classes,
Sequentially selecting a target gradient value among gradient values of the boundary candidate points;
Calculating an inter-class variance value for the selected target gradient value
LED boundary extraction method comprising a. The method of claim 5,
Computing an inter-class variance value for the selected target gradient value,
Classifying a temporary boundary point and a temporary boundary uncertainty point from the boundary candidate points based on the target gradient value;
Calculating the inter-class variance value based on at least one of an average value of the temporary boundary point, a weight, an average value of the temporary boundary uncertainty point, and a weight
LED boundary extraction method comprising a. The method of claim 4,
The step of sequentially calculating the inter-class variance value for the boundary candidate points,
Calculating a distribution of the gradient values of the boundary candidate points
LED boundary extraction method comprising a. The method of claim 1,
Setting a median value in the gradient intensity range of all points as the first threshold value
LED boundary extraction method further comprising a. A computer-readable recording medium storing one or more computer programs including instructions for performing the method of any one of claims 1 to 8. An image generator configured to generate an LED image frame including a plurality of all points based on reception of an external LED visible light signal;
A processor that extracts gradient values of the plurality of all points; And
A memory for mapping and storing the gradient values for each of the points
Including,
The processor,
Extracting points having a gradient value greater than a first threshold among the total points as boundary candidate points,
A gradient value of a point having a maximum between-class variance value among the extracted boundary candidate points is set as a second threshold value,
Classifying points having a gradient value greater than the second threshold among the boundary candidate points as boundary points
LED boundary extraction device.

Images