KR101185935B1 - Animal detection system and method thereof - Google Patents

Abstract

PURPOSE: An animal detecting system and a method thereof are provided to quickly predict a moving direction of an animal through a minimum algorithm if an object in front of a car is the animal. CONSTITUTION: An animal sensing system obtains a binarized image by binarizing a photographed image(S1315). The system extracts an area of a target object(S1320). The system detects at least one end point from the target object(S1330). The system partitions the area of the target object into four areas(S1340). The system determines whether the target object is an animal(S1350). [Reference numerals] (A1310) Taking photography; (A1315) Obtaining a binarized image; (A1320) Extracting an area of a target object; (A1330) Extracting an end point from the target object; (A1340) Partitioning the area of the target object into four areas; (A1350) Determining whether the target object is an animal; (A1360) Predicting the moving direction of the target object; (A1370) Generating a warning event; (AA) Start; (BB) End

Description

Animal Detection System and Method

The present invention relates to a system and method for utilizing the multipoint pointer and region to detect whether the animal is present and to predict the trend of the animal.

Accidents caused by so-called road kills that inadvertently enter the driveway not only kill the animals, but can also have a fatal effect on the person in the vehicle. These road killing accidents can be a serious problem not only in Korea but also in countries where travel by vehicles is common.

In order to prevent such road kills, various measures are being prepared. In particular, a night vision system may be installed in a vehicle to prevent road kills occurring at night. Night Vision is a device designed to enhance the safety of night driving by applying thermal-imaging technology to automobiles to secure the driver's vision at night. Generally, a special infrared surveillance camera mounted on the rear of the front grill is used. It is a device that can identify an object that is over three times the irradiation distance of the beam by the headlight by photographing the object in front of the vehicle and displaying it on the head-up display as an image image.

However, it is necessary to determine whether an object located in front of a moving vehicle is an object that cannot be moved or is currently moving or is likely to move later, and a warning necessary only in the latter case, which can be particularly dangerous to a moving vehicle. It will be important to make the driver do not disturb the driver's concentration.

Moreover, this warning system is very important if the object being moved or possibly moving is an animal. Unlike humans, animals with low risk perception of moving vehicles may be able to suddenly jump into the road even if they are stopped.

Therefore, a system that identifies whether the object is in particular an animal among the objects located in front of the driving vehicle, and further predicts the future direction of movement of the object, warns the driver only when necessary or warns by dividing the risk step by step. need.

To this end, a method and apparatus for predicting the future direction of movement of the animal need to be preceded if it is determined whether the object located in front of the driving vehicle is an animal and the animal. Considering the fact that the onboard CPU is not an advanced specification, and that the vehicle should be mounted on a high-speed vehicle, the prediction of animal status and direction of movement should be made as fast as possible. Need to be minimized.

It is an object of the present invention to provide an animal recognition system and method that enables recognition of whether an object located in front of a driving vehicle is an animal in a quick moment through a minimum algorithm.

Another problem to be solved by the present invention is an animal recognition system and method for predicting the future direction of movement of the animal when it is determined that the object located in front of the driving vehicle is an animal in a short time through a minimum algorithm. To provide.

Problems to be solved by the present invention are not limited to the above-mentioned problems, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

One aspect of the animal recognition method of the present invention for solving the above problems is the step of extracting the object region, which is the position where the photographed object exists in the photographed image, by dividing the object region to the left and right and up and down, a quadrant region Setting a step, detecting at least one end point of the photographed object, and determining whether the object is an animal according to a pattern in which the end point is located in the quadrant region.

One aspect of the animal recognition device of the present invention for solving the above problems is an object region extraction unit for extracting an object region, which is a position where a photographed object exists in the photographed image, by dividing the object region left and right and up and down, According to the pattern of 4 division region setting unit for setting the 4 division region, the end point extraction unit for detecting one or more end points of the photographed object, and the end point is located in the 4 division region, it is determined whether the object is an animal It includes a determination unit.

Other specific details of the invention are included in the detailed description and drawings.

Due to the present invention, if it is determined that the object located in front of the running vehicle is an animal and the object located in front of the running vehicle is determined to be an animal, the prediction about the future direction of movement of the animal is performed through a minimum algorithm. By being able to do it in a short time, it is possible to prevent vehicle accidents caused by road kills.

1 illustrates an example of an object region to be extracted in a captured image.
Figure 2 is an embodiment showing a process for detecting the three end points for the object in the extracted object region of the present inventors.
Figure 3 is an embodiment showing a process of detecting the three end points for the object in the extracted object region of the present inventors.
Figure 4 is an embodiment showing a process of estimating the left and right moving direction of the object based on the location of the four-divided object region and the end point of the present invention.
FIG. 5 is an embodiment illustrating a process of predicting left and right moving directions of an object based on the positions of the quadrant object region and the end point of the present invention.
Figure 6 is an embodiment showing a process of detecting the four end points for the object in the extracted object region for grasping the object in the forward or backward direction of the present invention.
7 is an embodiment showing a process of detecting the three end points for the object in the extracted object region for grasping the object in the forward or backward direction of the present invention.
8 is an embodiment showing a process of predicting the moving direction of the front and back of the object based on the location of the four-divided object region and the end point of the present invention.
9 is an embodiment showing a process of predicting the moving direction of the front and rear of the object based on the location of the quadrant object region and the end point of the present invention.
FIG. 10 is an embodiment illustrating a process of detecting an end point of an object in an extracted object area for grasping a still object of the present invention.
FIG. 11 is an embodiment illustrating a process of predicting a stationary state of an object based on the location of the quadrant object region and the end point of the present invention.
12 is an embodiment of the animal detection system of the present invention.
Figure 13 is a flow chart of one embodiment of the present invention animal detection method.
FIG. 14 is a flowchart specifically describing an operation S1360 of FIG. 13.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS The advantages and features of the present invention and the manner of achieving them will become apparent with reference to the embodiments described in detail below with reference to the accompanying drawings. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. Is provided to fully convey the scope of the invention to those skilled in the art, and the invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout.

Unless otherwise defined, all terms (including technical and scientific terms) used in the present specification may be used in a sense that can be commonly understood by those skilled in the art. Also, commonly used predefined terms are not ideally or excessively interpreted unless explicitly defined otherwise.

1 is an embodiment illustrating an object region to be extracted in a captured image, and FIGS. 2 and 3 are embodiments illustrating a process of detecting three end points of an object in an extracted object region. And Figure 13 is a flow diagram of an embodiment of the animal detection method of the present invention.

1 is a view showing an image obtained by photographing the front of a vehicle running during the day or night. In this shooting, after irradiating light of a predetermined frequency from a lamp mounted on a vehicle, the light is reflected by an object and photographed by a camera (step S1310 of FIG. 13), wherein the light of the predetermined frequency is infrared light. And especially near-infrared.

Next, the captured image is binarized to obtain a binarized image (step S1315 of FIG. 13). All pixels of the binarization image are displayed in black and white only. As a reference for this, a pixel having a low value may be expressed in black and a pixel having a high value may be expressed in white based on a predetermined threshold value. The predetermined threshold value may use various methods such as an average of brightness values of all pixels or an iterative selection and a tow peak method. Alternatively, the user may set and use a threshold value.

In addition, erosion and dilation operations are performed on the binarized image to remove noise of the binarized image. That is, the background region is extended with respect to the subject area through erosion, and the size of the subject is reduced to remove or reduce noise components around the subject area. In order to restore the size of the smaller subject image, the subject is expanded and the background is reduced through the expansion operation.

Through this process, FIG. 1 shows an example in which two objects 102 and 103 are photographed. The object photographed in this way is represented by regions 107 and 108 having a higher luminance than the surroundings as shown in FIG. 1. In this way, the high brightness region distinguished from the surroundings is extracted (step S1320 of FIG. 13). The region thus extracted is hereinafter referred to as an object region.

As an embodiment of the present invention, the object regions 107 and 108 may be extracted in the form of rectangles as shown in FIG. 1. The point where the high luminance pixel and the low luminance pixel are close to each other will be identified as a boundary line representing the shape of the object. Here, a rectangle including four vertices of the vertical and horizontal lines in contact with the boundary lines as the vertices of the vertices is extracted into the object area, or a rectangle further includes some free space up, down, left and right than the rectangular area. May be extracted to the object area. 2 and 3 show the object regions 206 and 306 thus extracted.

Next, the end point of the object in the object area is detected (S1330 in Fig. 13). In order to detect the end point, line shapes 202 and 302 having the shape characteristics of the objects 201 and 301 in the object area as shown in the right picture of Figs. If only one other point of the linear shape is included in the eight pixels surrounding any pixel existing on the acquired linear shapes 202 and 302, this point may be regarded as the end point of the object.

In one embodiment of the invention, these end points can in principle be three. The CPU installed in the vehicle may use a non-high-end specification in consideration of the selling price. Furthermore, in order to prevent an accident through quick judgment because it is mounted in a vehicle that runs at high speed, a minimum algorithm may minimize the computation time of the CPU. There is a need. Therefore, reducing the number of endpoints as much as possible can maximize the effect of accident prediction.

The three end points made in principle in the present invention, as shown in Figs. 2 and 3, respectively, are the first end points 205 and 305 representing the head of the animal and the second points representing the leg in the lateral view. And third end points 203 and 204 and 303 and 304.

In order to identify four-legged animals, the reason why the end points representing the legs are set to two instead of four in the present invention is that, as mentioned above, the algorithm can be minimized to increase the effectiveness of accident prevention measures through quick judgment. Because there is.

Next, the photographed object area is divided into left and right and up and down to set a quadrant area (S1340 in FIG. 13). The reason for setting the quadrant is to predict whether the object is an animal and the direction of movement of the object according to the pattern in which the end points are located in the quadrant.

In FIG. 13, the end point of the object is detected (S1330), and the object area is divided into four to set up a four-division area (S1340). However, the order of these two steps may be reversed. That is, the end point of the object may be detected after setting the quadrant region of the object area.

4 and 5 are exemplary embodiments illustrating a process of predicting left and right moving directions of an object based on the positions of the quadrant object region and the end point of the present invention.

First, as shown in FIG. 2, the extracted end points 203 to 205 are located in a predetermined region among the four divided regions, and FIG. 5 shows the extracted end points (as shown in FIG. 3). 303 to 305 are located in a predetermined region among the four divided regions.

For explanation, as a name for each of the four divided regions, the upper left and right two regions of the four divided regions are called the first region and the second region, respectively, and the lower left and right regions of the four divided regions are respectively referred to. This is referred to as a third region and a fourth region.

When the 2nd and 3rd end points exist one by one in the 3rd area | region and the 4th area | region, and the 1st end point exists in either one of a 1st area | region or a 2nd area | region, it is judged that this object is an animal standing sideways. (S1350 of FIG. 13).

Furthermore, when the first end point is located in the first region as shown in FIG. 5, the animal may suddenly jump to the left at any moment in the future, even if the object is currently moving to the left or is stationary. It can be judged that there is a possibility. That is, when the first end point is located in the first region, the actual or tentative movement direction of the object is anticipated to the left (S1360 in FIG. 13).

In addition, when the first end point is located in the second region as shown in FIG. 4, even if the object is currently moving to the right or stopped, the animal may suddenly jump to the right at any moment in the future. You can judge that there is. That is, when the first end point is located in the second region, the actual or potential moving direction of the object is predicted in the right direction (S1360 of FIG. 13). The driver may generate a warning event of various stages based on the prediction about the position and the moving direction in which the object exists (S1370 of FIG. 13).

6 and 7 are diagrams illustrating a process of detecting four or three end points of an object in the extracted object region to identify an object facing forward or backward, respectively. 8 and 9 are diagrams showing a process of predicting the moving direction of the front and rear of the object on the basis of the location of the quadrant object region and the end point of the present invention.

The present invention provides an embodiment for predicting forward and backward directions in addition to predicting left and right directions of an object. Here, the descriptions of the photographing and the division into the four divided regions are the same as those described with reference to FIGS. 2 to 5, and thus descriptions thereof will be omitted. However, the endpoint may be represented as three as shown in Figure 7 or may be shown as four as shown in FIG. In a general case, three end points may appear as shown in FIG. 7, but, for example, when there are horns such as some deer and animals, four end points may appear as shown in FIG. 6. However, the case of four may be considered as an exceptional case, because the object in the object area to be extracted is usually located at a distance from the driving vehicle, so that the two horns are not clearly distinguished. This can be taken, in which case only one endpoint representing the head will be detected.

If two of the three or four end points are located in the third area and the fourth area, respectively, and one or two end points are predetermined from the boundary or boundary line between the first area and the second area. When located within a distance, the object can be determined to be a four-legged animal facing forward or backward.

8 and 9, as an embodiment of the present disclosure, a new region may be added and divided in addition to the first region and the second region. That is, when a boundary line of the original first and second regions and a predetermined distance are set in the left and right directions from the boundary line, the remaining one or two end points are located within this predetermined distance as shown in FIGS. 8 and 9. The object is judged as a four-legged animal facing forward or backward.

Compared to the case in which the animal moves in the left or right direction in FIG. 2 to FIG. 2 or the movement is expected, the driver may move to the front or rear term as in FIGS. It may be considered as a relatively low risk, but considering that the change of direction of movement of animals is very quick compared to humans, it is also necessary to inform the driver that there is a risk. Still exists.

FIG. 10 is a diagram illustrating a process of detecting an end point of an object in an object region extracted for grasp of the stationary object of the present invention, and FIG. 11 is based on the location of the quadrant object region and the end point of the present invention. It is a figure to explain the process of predicting the stationary state of an object.

When the animal is sitting as shown in the right figure of FIG. 10, the end point representing the leg is not detected. Accordingly, as shown in FIG. 11, there are no end points in the third and fourth regions of the four-divided region, and two end points exist in the first region or the second region as illustrated in FIG. 11, or when the animal has no horns. Only one endpoint may be present (not shown). That is, although the luminance is extracted to the object region because the brightness is high, if the detected end point does not exist in the third or fourth region, it is determined as a sitting animal. In this case, although the end point indicating the leg is not displayed, an area (1001 of FIG. 10) representing the body of the animal is displayed on the display unit 1230 of FIG. 12 as needed so that the driver can reconfirm. You may.

12 is an embodiment of the animal detection system of the present invention. The animal detection system of the present invention includes a binarization image acquisition unit 1105, an object region extraction unit 1110, a four-division region setting unit 1120, an end point extraction unit 1130, the determination unit 1140, the control unit ( 1150, and a warning unit 1150, and may further include a photographing unit 1200 and a display unit 1230 including a lamp 1210 and a camera 1220 to photograph an object. have.

The lamp 1210 of the photographing unit 1200 irradiates light of a predetermined frequency toward the moving direction of the vehicle for photographing, and the light of the predetermined frequency may be infrared rays, particularly near infrared rays. The camera 1220 photographs a state in which the light is reflected by the object and transmits the photographed result to the binarization image acquisition unit 1105.

The binarization image acquisition unit 1105 obtains the binarization image by binarizing the photographed image. All pixels of the binarization image are displayed in black and white only. As a reference for this, a pixel having a low value may be expressed in black and a pixel having a high value may be expressed in white based on a predetermined threshold value. The predetermined threshold value may use various methods such as an average of brightness values of all pixels or an iterative selection and a tow peak method. Alternatively, the user may set and use a threshold value.

In addition, the binarization image acquisition unit 1105 removes noise of the binarization image by performing erosion and dilation operations on the binarization image. That is, the background region is extended with respect to the subject area through erosion, and the size of the subject is reduced to remove or reduce noise components around the subject area. In order to restore the size of the smaller subject image, the subject is expanded and the background is reduced through the expansion operation.

The object region extractor 1110 may extract the object region in the form of a rectangle, as described with reference to FIG. 1, based on the binary image obtained by the binarization image acquirer 1105. The point where the pixels of high brightness and pixels of low brightness are close to each other is identified as the boundary line representing the shape of the object, and the rectangle whose four vertices are the points where the outermost straight lines among the vertical and horizontal lines that contact the boundary line are the four vertices as the object area. Alternatively, a rectangle may further be extracted as the object region, which further includes some free space on the upper, lower, left and right sides of the rectangular region. The extracted object region is transmitted to the four division region setting unit 1120 and the endpoint extraction unit 1130.

The four division region setting unit 1120 divides the object region into four vertical sections. Through the four divided areas, the determination unit 1140 later performs identification on the object and the prediction of the moving direction.

The endpoint extractor 1130 detects an endpoint of the object in the object region. In order to detect the end point, the linear shapes 202 and 302 having the shape characteristics of the objects 201 and 301 in the object area as shown in the right figure of Figs. If the eight pixels surrounding any pixel present on 302 contain only one other point in the line shape, this point may be regarded as the end point of the object. And, in one embodiment of the present invention, these end points will be three in principle, but may be one, two or four, depending on the shape of the object.

The determination unit 1140 determines whether the object is an animal, whether the object moves in the left, right, or forward and backward directions, the moving direction, and the expected movement direction through the positions where the end points exist in the divided region. Transfer to control unit 1140. Since a detailed determination method has been described above with reference to FIGS. 2 through 11, a description thereof will be omitted.

The controller 1140 determines the level of warning to be issued to the driver based on the determination result received through the determination unit 1140 and transmits the warning to the warning unit 1160. Meanwhile, in FIG. 12 and the description thereof, the determination unit and the control unit are separately displayed and described, but the determination unit and the control unit may be configured as one component to perform the above functions.

The warning unit 1160 generates a warning event to the driver according to the degree of danger determined by the control unit, and the warning event includes a visual warning of different intensities, a voice warning of different volumes, and automatic braking according to the degree of danger. Performance and the like. Among these, the voice alert may be a sound for notifying the driver, or a sound file in a frequency range in which only an animal can respond.

The display unit 1230 allows the driver to see the captured image as it is. The display unit displays a cathode ray tube (CRT), a liquid crystal display (LCD), a light-emitting diode (LED), and an organic light emitting diode (OLED) capable of displaying an input image signal. -A module having an image display means such as an emitting diode (PDP) or a plasma display panel (PDP), and serves to display the received image information.

Until now, each component of FIG. 12 may refer to software or hardware such as a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC). However, the components are not limited to software or hardware, and may be configured to be in an addressable storage medium and configured to execute one or more processors. The functions provided in the above components may be implemented by more detailed components, or may be implemented as one component that performs a specific function by combining a plurality of components.

FIG. 14 is a flowchart illustrating an embodiment of the present invention specifically describing operation 1360 of FIG. 13. When the number of the identified endpoints is three (S1402), when it is determined whether two of the three identified endpoints exist in each of the third and fourth regions of the divided object region (S1403) (Y1403) ), Where the other end point is located. If the object is present in area 1 (S1406), the object is currently moving in the right direction or is expected to be moved later (S1407). If the object is present in area 2 (S1408), the object is currently moving in the right direction; Or it is expected to move later (S1409). If the object is included in a predetermined distance from the boundary line between the first region and the second region or within a predetermined boundary line, the object may be currently moving forward or backward or may be expected to move later (S1405).

If, in step S1403, two of the three end points are not present in each of the third and fourth regions of the object region divided into four (N), it is determined that the object is currently stopped (S1410). ).

In the step S1402, the number of the identified end points is not three (N), even if it is determined that the four end points (Y in S1411), it is determined whether the two end points are present in the 3, 4 areas, respectively, the remaining two end points are the first It is the same to expect the movement in the right direction, the left direction, or the front and rear directions depending on whether it is within a certain range of the region, the second region or the 1,2 region boundary line.

If the number of identified endpoints is not three or four (N in S1411), it is determined whether the endpoint exists in the third region and the fourth region, and if it is not present (N), the object is called a stationary state. To judge.

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, You will understand. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive.

102, 103, 201, 301: object
107, 108, 206, 306: object area
202, 302: line shape
203-205, 303-305: Endpoint
1105: binarization image acquisition unit
1110: object region extraction unit
1120: 4-part division setting unit
1130: endpoint extraction
1140: determination unit 1150: control unit
1160: warning unit 1210: lamp
1220: camera 1230: display unit

Claims (16)

Binarizing the photographed image to obtain a binarized image;
Extracting an area of an object included in the image based on the obtained binarized image;
Detecting one or more endpoints from the object;
Dividing the area of the object into left and right and up and down to set a quadrant area; And
And determining whether the object is an animal according to a pattern in which the endpoint is located in the quadrant region.
The end point means a linear end point having the shape characteristics of the object , animal detection method. The method of claim 1,
And predicting a moving direction of the object according to the pattern. The method of claim 1,
The end points are three, each of the first to third end points, and the upper left and right areas of the four divided areas are called the first area and the second area, respectively, and the lower left, When the two regions are called the third region and the fourth region, respectively,
When the first end point is located in any one of the first or second regions, and the second end point and the third end point are located in the third region and the fourth region, respectively, the object is determined as an animal. How to detect animals. The method of claim 3,
When the first end point is located in the first region, the moving direction of the object is predicted to the left,
And detecting the movement direction of the object in the right direction when the first end point is located in the second area. The method of claim 1,
The photographing of the object, after irradiating light of a predetermined frequency, the animal detection method for photographing the state in which the light is reflected by the object. The method of claim 5,
The light of the predetermined frequency is near-infrared animal detection method. The method of claim 1,
When the upper left and right areas of the four divided areas are called the first area and the second area, respectively, and the lower left and right areas of the four divided areas are called the third area and the fourth area, respectively,
And the end point is not present in the third and fourth areas, and determines the object as a sitting animal. The method of claim 1,
The end points are three or four, and are the first to third or the first to fourth end points, respectively, and the upper left and right areas of the four divided areas are called the first area and the second area, respectively. When the lower left and right areas of the four divided areas are referred to as the third area and the fourth area, respectively,
When two of the endpoints are each located in the third region and the fourth region, respectively, and one or two endpoints are located within a predetermined distance from the boundary or boundary line between the first region and the second region, Animal detection method for determining the object as a four-footed animal facing forward or backward. A binarization converter for binarizing the photographed image to obtain a binarized image;
An object region extracting unit extracting an area of an object included in the image based on the obtained binarized image;
A 4-division region setting unit which divides the extracted region of the object into left and right and up and down to set a 4-division region;
An endpoint extraction unit for detecting one or more endpoints from the object; And
According to the pattern of the end point is located in the four-division region, including a determination unit for determining whether the object is an animal,
The end point means a linear end point having the shape characteristics of the object . 10. The method of claim 9,
The determination unit is an animal detection system for predicting the moving direction of the object according to the pattern. 10. The method of claim 9,
The end points are three, each of the first to third end points, and the upper left and right areas of the four divided areas are called the first area and the second area, respectively, and the lower left, When the two regions are called the third region and the fourth region, respectively,
The determination unit, when the first end point is located in any one of the first or second region, the second end point and the third end point is located in the third region and the fourth region, respectively, the object Animal detection system to judge the animal as. The method of claim 11,
The determination unit predicts a moving direction of the object to the left when the first end point is located in the first area, and moves the moving direction of the object to a right direction when the first end point is located in the second area. Predictive animal detection system. 10. The method of claim 9,
The photographing of the object includes a lamp for irradiating light of a predetermined frequency;
And a camera for photographing the light reflected by the object. The method of claim 13,
And the light of the predetermined frequency is near infrared. 10. The method of claim 9,
When the upper left and right areas of the four divided areas are called the first area and the second area, respectively, and the lower left and right areas of the four divided areas are called the third area and the fourth area, respectively,
The determination unit, if the end point is not present in the third region and the fourth region, the animal detection system for determining the object as a sitting animal. 10. The method of claim 9,
The end points are three or four, and are the first to third or the first to fourth end points, respectively, and the upper left and right areas of the four divided areas are called the first area and the second area, respectively. When the lower left and right areas of the four divided areas are referred to as the third area and the fourth area, respectively,
The determination unit may include two of the endpoints, one each in the third region and the fourth region, respectively, and the other one or two endpoints are within a predetermined distance from the boundary or boundary line between the first region and the second region. When positioned, the animal detection system for determining the object as a four-footed animal facing forward or backward.

Images