KR20150092505A - A method for tracking a vehicle, a method for warning a distance between vehicles and a device for warning a distance between vehicles - Google Patents

Abstract

The present invention relates to a method for warning a distance between vehicles and a device for detecting a forward collision. The method for warning a distance between vehicles comprises the steps of: setting one or more areas of interest in one or more images among a plurality of images; detecting a first feature point corresponding to a vehicle from the areas of interest, and setting vehicle candidates based on the detected first feature point; detecting one or more vehicles from tracking candidates by using reference data obtained based on the geometry of the vehicle; and tracking one or more vehicles detected by the above step.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to an inter-vehicle distance warning method and a forward collision detection apparatus,

An inter-vehicle distance warning method, and a front collision detection apparatus.

A vehicle is a kind of transportation means capable of moving a vehicle, such as a person, an object, or an animal, from a departure point to a destination while traveling on a road or a track. A vehicle can generally be moved in various directions by rotation of two or more wheels. There are various types of vehicles, such as three-wheeled or four-wheeled vehicles, motorcycles such as two-wheeled cars, construction machines, prime mover bicycles, bicycles, and trains running on the track.

A frontal collision detection device is a device that prevents the impulse of the vehicle and helps the driver to operate the vehicle safely by warning the driver of the vehicle about the proximity of the vehicle when the vehicle and the vehicle are in close proximity and there is a risk of collision to be. The distance between the vehicle and the vehicle may include both the distance between the preceding vehicle and the following vehicle or the distance between the vehicle and the vehicle traveling on the side of the specific vehicle.

An object of the present invention is to provide an inter-vehicle distance warning method and a front collision sensing device capable of warning a vehicle distance by detecting a vehicle using a plurality of models obtained by classifying the vehicle model having geometrically identical characteristics.

In order to solve the above problems, an inter-vehicle distance warning method and a forward collision sensing apparatus are provided.

The inter-vehicle distance warning method includes the steps of: setting at least one region of interest in at least one of a plurality of images; detecting a feature point that may correspond to the vehicle from the at least one region of interest; and based on the detected first feature point Detecting the at least one vehicle from the tracking candidate group using reference data obtained based on the geometrical structure of the vehicle, and tracking the detected at least one vehicle .

The reference data may include at least one of first reference data including a sedan and a sports utility vehicle, second reference data including a bus or a truck, and third reference data including another specific vehicle.

Wherein the step of detecting at least one vehicle from the tracking candidate group comprises the steps of applying at least one of a horizontal edge mask and a vertical edge mask to a vehicle image and a non-vehicle image to calculate an edge size and an angle, Calculating a second feature point by calculating a histogram for at least one of the segmented zones, and calculating at least one vector from the tracking candidate group using at least one vector obtained based on the second feature point and the support vector machine, May be detected.

The inter-vehicle distance warning method includes calculating a distance between the at least one vehicle to be tracked and the reference vehicle based on the position of the lower end of the vehicle and the vehicle width information, and calculating a distance between the detected vehicle and the camera, And warning the driver if the distance is smaller than the predetermined distance.

A frontal collision sensing apparatus includes a photographing section capable of continuously photographing a plurality of images and at least one region of interest in at least one of the plurality of images and a feature point capable of being corresponded to the vehicle from the at least one region of interest Detecting at least one vehicle to be tracked from the tracking candidate group using reference data obtained based on the geometrical structure of the vehicle, detecting the at least one detected vehicle based on the detected first feature point, And a tracking unit for tracking the tracked vehicle.

The reference data may include at least one of first reference data including a sedan and a sports utility vehicle, second reference data including a bus or a truck, and third reference data including another specific vehicle.

Wherein the tracking unit calculates a distance between the at least one vehicle to be tracked and the reference vehicle based on the position of the lower end of the vehicle and the width information and sets a distance between the at least one tracked vehicle and the reference vehicle to a predefined value It is possible to determine whether to warn the inter-vehicle distance.

According to the headway distance warning method and the frontal collision detection apparatus as described above, it is possible to detect a vehicle using a plurality of models obtained by classifying the vehicle with geometrically identical characteristics.

In addition, since the vehicle is classified into a vehicle having geometrically identical characteristics, it can be advantageous in detecting various types of vehicles without missing.

Therefore, it is possible to acquire the distance between various kinds of vehicles while driving with the vehicle, thereby improving the efficiency of the inter-vehicle distance alarm and further assuring the safety of the driver and the passenger.

1 is a view for explaining an inter-vehicle distance and a front collision sensing apparatus.
2 is a perspective view showing an appearance of an embodiment of a vehicle.
3 is a configuration diagram of an embodiment of a frontal collision sensing apparatus.
4 is a diagram showing an embodiment of an image in which a region of interest is set.
5 is a view showing one embodiment of the minutiae of the vehicle rear surface.
6 is a diagram showing an embodiment of a method of detecting a vehicle using a scan window.
7 is a diagram showing an embodiment of an image in which a vehicle candidate group is detected.
8 is a diagram for explaining a result of determination as to whether or not the vehicle is an actual vehicle.
9 is a diagram for explaining a method for determining whether or not to keep track of a vehicle.
10 is a flowchart of an embodiment of a headway distance warning method.
11 is a flowchart showing an embodiment of a method for determining whether or not vehicle tracking is maintained.

An embodiment of the front collision sensing apparatus will be described below with reference to FIGS. 1 to 9. FIG.

1 is a view for explaining an inter-vehicle distance and a front collision sensing apparatus. Referring to FIG. 1, the inter-vehicle distance d means a distance between the first vehicle 1 and the second vehicle 10. Here, the inter-vehicle distance d may be the distance between the preceding first vehicle 1 and the following second vehicle 10, and the distance between the first vehicle 1 and the second vehicle 2 parallel to each other Lt; / RTI > When the inter-vehicle distance d is too short, when the driver of the second vehicle 10 performs an unexpected operation such as stopping the first vehicle 1 or changing lanes, It may not react quickly to the operation of the vehicle 1, and an accident or the like may be caused. The frontal collision sensing device judges the distance d between the first vehicle 1 and the second vehicle 2 and uses at least one of a sound such as voice and a warning light to the driver in accordance with the determined inter- To warn that the inter-vehicle distance d is short.

2 is a perspective view showing an appearance of an embodiment of a vehicle. 2, the vehicle 10 includes a vehicle body 12 forming an outer appearance of the vehicle 10, wheels 13 and 14 for moving the vehicle body 12 in a predetermined direction while rotating, gasoline or diesel An engine room 15 in which an engine for burning and obtaining power is installed, and a wind shield 16 for providing a driver in the vehicle body 12 with a front view of the vehicle. The vehicle 10 may also include an electronic device 17 for electronic control of the interior of the vehicle.

The vehicle 10 may further include a photographing section 11 capable of photographing the front in one embodiment. The photographing unit 11 may be implemented as a camera device capable of photographing at least one of a still image and a moving image. The camera device may include a digital camera device, a camcorder device, and the like. The camera device may include a lens and a light-receiving portion that receives light incident on the lens and converts the received light into an electrical signal. The light receiving unit may be implemented using a charge coupled device (CCD) or a complementary metal-oxide semiconductor (CMOS).

The photographing unit 11 may be installed inside the vehicle 10 or outside the vehicle 10. [ For example, the photographing section 11 may be installed inside the windshield 16 as shown in FIG. However, the mounting position of the photographing unit 11 is not limited to this, and may be installed on the upper part of the dashboard inside the vehicle or a part of the grill of the vehicle.

3 is a configuration diagram of an embodiment of a frontal collision sensing apparatus.

In one embodiment, the front collision sensing apparatus 20 includes a photographing section 11 capable of continuously photographing a plurality of images as shown in Fig. 3, and a control section 20 for detecting at least one vehicle from at least one of the plurality of images, And a tracking unit (21) for tracking at least one vehicle.

According to one embodiment, the tracking unit 21 may include a first image processing unit 22, a second image processing unit 23, and a control unit 26.

The first image processing unit 22 may set a region of interest that is a partial region of the image, detect a first feature point from the ROI, and then detect a candidate region based on the first feature point. The first image processing unit 22 may set a plurality of regions of interest.

4 is a diagram showing an embodiment of an image in which a region of interest is set.

The first image processing unit 22 may set the region of interest according to at least one of the predefined distance and the predefined width. According to an embodiment, the first image processor 22 projects a predetermined distance in a three-dimensional coordinate system of an actual space onto an image using a hierarchical matrix (H-matrix) . Here, the predefined distance may include the maximum distance at which the vehicle can be recognized. The maximum distance at which the vehicle can be recognized may be, for example, 120 m. On the other hand, the specific region of the image that is set as the region of interest may include at least one of a region that is far from the real space and a region that is near the real space. The first image processing unit 22 may limit the size of the region of interest in the horizontal direction in order to detect only the vehicle 1 traveling in the lane in which the second vehicle 10 is currently traveling and the left and right lanes.

As a result, the first image processing unit 22 can set the ROIs ROI1 and ROI2 as shown in FIG.

On the other hand, the first feature point used for setting the vehicle candidate group may be one defined to identify the vehicle. The first feature point can be defined based on a specific element of the vehicle rear surface that can be a feature of the vehicle. 5 is a view showing one embodiment of the minutiae of the vehicle rear surface. As shown in Fig. 5, the first characteristic point is a portion that can be used for vehicle identification when the vehicle 30 is viewed from the rear, for example, the brake lights 31 and 32, the rear fog lights 34 and 35, A bumper 3, left and right interface surfaces 37a and 37b of the vehicle, a license plate 38 and left and right side mirrors 39a and 39b. The first feature point may be stored in the feature point database 23 as shown in FIG. The first image processing unit 22 can receive the first feature point by browsing the feature point database 23.

FIG. 6 is a view showing an embodiment of a method of detecting a vehicle using a scan window, and FIG. 7 is a diagram showing an embodiment of an image in which a vehicle candidate group is detected.

According to an embodiment, the first image processing unit 22 may step-scan an image based on the first feature point to determine whether or not a vehicle exists in the image.

The first image processing unit 22 may apply at least one scan window w1 and w2 to the image in a cascade manner to determine whether a vehicle exists in the image and detect a candidate vehicle group. For example, the first image processing section 22 can detect a vehicle candidate group by detecting the same or similar to the first feature point from the images in the at least one scan window (w1, w2). In this case, at least one scan window (w1, w2) can be located in at least one position in the image as shown in Fig. 6 and can detect a vehicle candidate group in the image while moving in a predetermined direction w11, w21 . If necessary, the first image processing unit 22 can browse the minutiae point database 23, receive the first minutiae points, scan the images on the basis of the received minutiae points, and detect the vehicle candidate group. As a result, a plurality of vehicle candidates c1 to c4 can be detected as shown in Fig. The plurality of vehicle candidates c1 to c4 may include the vehicle but may not include the vehicle as the fourth candidate c4.

The set of vehicle candidates set in the first image processing unit 22 may be transmitted to the second image processing unit 23.

The second image processing unit 24 can detect at least one vehicle from the tracking candidate group using the reference data obtained based on the geometrical structure of the vehicle.

The reference data used in the second image processing unit 24 may be determined according to the classification of the vehicle. Vehicles can be classified according to various criteria. For example, a vehicle can be classified according to the geometric structure of the vehicle, for example, the similarity of the width or width of the vehicle or the similarity of the ratio between the width and the width. In addition, vehicles can be classified according to various criteria. The second image processing unit 24 can use a plurality of reference data according to the classification of the vehicle according to the geometrical structure. For example, the reference data may include first reference data for a sedan and a sports utility vehicle, second reference data for a bus or truck, and third reference data for other singular vehicles, May use them to detect at least one vehicle from the tracking candidate group.

According to an embodiment, the reference data database 25 may include first to third reference data. The second image processing unit 24 can browse the reference data database 25 and use at least one of the first reference data and the third reference data from the reference data database 25 for use in image processing.

The reference data may include a support vector. The support vector may be obtained based on the support vector machine.

The second image processing section 24 applies at least one of a horizontal edge mask and a vertical edge mask to the vehicle image and the non-vehicle image to calculate the size and angle of the edge, and divides the image into at least one zone The second feature point can be calculated by calculating a histogram of at least one zone partitioned after the second feature point.

The second image processing unit 24 may perform an inner product of the calculated second feature point and the support vector of the reference data and determine whether at least one of the candidate vehicle groups c1 to c4 is a vehicle according to the inner product result. As described above, when the reference data includes the first reference data to the third reference data, the second image processing section 23 calculates the difference between the second feature point and the support vector of the first reference data to the support vector of the third reference data It is possible to judge and detect the candidate as a vehicle when any one of the candidate candidates c1 to c4 corresponds to any of the first reference data to the third reference data.

Fig. 8 is a diagram for explaining a result of determination as to whether or not the vehicle is a vehicle. As shown in Fig. 8, the inner product between the second feature point and the support vector of the first reference data to the support vector of the third reference data, the inner product between the first candidate data and the third reference data, The third candidate (c1 to c3) is determined as a vehicle, but the fourth candidate (c4) that does not correspond to any one of them may not be judged as a vehicle. The data for the fourth candidate c4 that is not judged as a vehicle can be deleted.

The determination result of the second image processing unit 24 may be transmitted to the control unit 26.

The control unit 26 calculates the distance between the vehicle that tracks and tracks at least one vehicle detected by the second image processing unit 23 and the vehicle in which the forward collision sensing apparatus 20 is installed and generates a control command according to the calculation result To the notification unit (27).

9 is a diagram for explaining an embodiment of a method for determining whether or not to keep track of a vehicle.

According to one embodiment, the control unit 26 may judge candidates that are repeatedly detected in a plurality of images among the vehicle candidate groups (c1 to c4) as actual vehicles. The second image processing unit 23 tracks only the vehicles continuously appearing among a plurality of images using a Kalman filter so as to remove the noise n that may temporarily exist on the image as shown in FIG. can do.

According to one embodiment, the control unit 26 may stop the tracking of the vehicle if the detected at least one vehicle is not detected in a plurality of images continuously photographed for a predetermined period of time. In other words, the control unit 26 may temporarily stop the tracking of the vehicle even if the vehicle c2 is lost on the image.

In another embodiment, the control unit 26 calculates the distance between the preceding first vehicle 1 and the second vehicle 10 equipped with the front collision sensing apparatus, and calculates the distance between the first vehicle 1 and the second vehicle 10 10 may be compared with a predefined value to determine whether to warn the inter-vehicle distance. In this case, the control unit 26 can calculate the distance between the first vehicle 1 and the second vehicle 10 based on the position of the lower end of the vehicle in the image and the vehicle width information.

The tracking unit 21 described above can be implemented by a semiconductor chip and a printed circuit board. The tracking unit 21 may be implemented by a semiconductor chip and a printed circuit board separately installed in a dashboard or a gear box in a vehicle or by a semiconductor chip and a printed circuit board inside the camera 11. [ The tracking unit 21 may also be implemented by an electronic device 17 provided in the vehicle 10. [

The notification unit 27 may output a warning sound or light to inform the driver of the warning relating to the inter-vehicle distance d. The notification unit 27 may include an audio output means such as a speaker and a light emitting means such as an LED light as necessary. In addition, the notification unit 27 may include a display device for displaying the inter-vehicle distance.

10 is a flowchart of an embodiment of a headway distance warning method.

10, the camera 11 installed in the vehicle 10 can take an image of the front of the vehicle 10 (s100).

The tracking unit 21 provided in the vehicle 10 may set at least one region of interest in the photographed image (S110). The tracking unit 21 can project a predetermined distance in a three-dimensional coordinate system of the actual space onto an image using a hierarchical matrix to set a specific region of the image as a region of interest. The area selected as the area of interest in the image may be limited depending on the distance between the characteristic object in the image and the camera 11. [ The width of the area selected as the region of interest in the image can also be limited.

The tracking unit 21 may detect the vehicle candidate group using the first feature point that can identify the vehicle in at least one region of interest (s120). In this case, the tracking unit 21 may apply at least one of the scan windows w1 and w2 in a cascade manner to determine whether a vehicle exists in the image,

The tracking unit 21 may detect at least one candidate from the vehicle candidate group to detect the vehicle (s130). In this case, the tracking unit 21 can detect at least one vehicle from the tracking candidate group using the reference data obtained based on the geometrical structure of the vehicle. More specifically, the tracking unit 21 applies at least one of a horizontal edge mask and a vertical edge mask to the vehicle image and the non-vehicle image to calculate the size and angle of the edge, divides the image into at least one zone, Performing a dot product of a second feature point and a support vector of reference data after computing a histogram for at least one region to obtain a second feature point and determining at least one candidate of the candidate vehicle groups c1 to c4 It can be determined whether the vehicle is a vehicle.

If it is determined that the vehicle is a vehicle, the tracking unit 21 starts tracking the determined vehicle (s140).

Meanwhile, according to the embodiment, the tracking unit 21 may further perform a process of determining candidates c1 to c4 that are repeatedly detected on a plurality of images successively photographed by the camera 11 as actual vehicles (s150) . In this case, the tracking unit 21 may use a Kalman filter. The tracking unit 21 can track the candidates judged as actual vehicles (s160).

The above-described steps s100 to s160 may be repeated several times or more (s170).

11 is a flowchart showing an embodiment of a method for determining whether to keep track of a vehicle.

As shown in FIG. 11, if the candidate tracked by the tracking unit 21, that is, the vehicle is not lost on the image, the tracking unit 21 can maintain the tracking of the candidate (s200, s210). In some cases, the candidate, i.e., the vehicle, may be lost on the image (s200). In this case, the tracking unit 21 keeps track (S230) if the candidate, that is, the vehicle again appears on the image (S220) before a predetermined time elapses after the vehicle is lost (S230) If the vehicle does not appear again on the image even after a predetermined time has elapsed after the disappearance of the vehicle, the tracking can be terminated (S240).

The steps s200 to s240 described above may be repeated several times or more (s250).

1: vehicle 10: vehicle equipped with front collision detection device
11: Camera 20: Front collision detection device
21: tracking unit 22: first image processing unit
23: feature point database 24: second image processing section
25: Reference data database
26: control unit 27:

Claims (7)

Setting at least one region of interest in at least one of the plurality of images;
Detecting a first feature point that may correspond to a vehicle from the at least one region of interest and setting a vehicle candidate group based on the detected first feature point;
Detecting at least one vehicle from the tracking candidate group using reference data obtained based on the geometry of the vehicle; And
And tracking the detected at least one vehicle. The method according to claim 1,
Wherein the reference data includes at least one of first reference data including a sedan and a sports utility vehicle, second reference data including a bus or a truck, and third reference data including another specific vehicle. The method according to claim 1,
Wherein the step of detecting at least one vehicle from the tracking candidate group comprises the steps of applying at least one of a horizontal edge mask and a vertical edge mask to a vehicle image and a non-vehicle image to calculate an edge size and an angle, Calculating a second feature point by calculating a histogram for at least one of the segmented zones, and calculating at least one vector from the tracking candidate group using at least one vector obtained based on the second feature point and the support vector machine, And detecting an inter-vehicle distance warning. The method according to claim 1,
Calculating a distance between the at least one vehicle to be traced and the reference vehicle based on the position of the lower end of the vehicle and the vehicle width information; And warning the driver if the distance between the detected vehicle and the camera is less than a predefined value. A photographing unit capable of continuously photographing a plurality of images; And
Setting at least one region of interest in at least one of the plurality of images, detecting a feature point that can correspond to the vehicle from the at least one region of interest, setting a vehicle candidate group based on the detected first feature point And a tracking unit for detecting at least one tracked vehicle from the track candidate group using reference data obtained based on the geometrical structure of the vehicle and for tracking the detected at least one tracked vehicle, . 6. The method of claim 5,
Wherein the reference data includes at least one of first reference data including a sedan and a sports utility vehicle, second reference data including a bus or a truck, and third reference data including another specific vehicle. 6. The method of claim 5,
Wherein the tracking unit calculates a distance between the at least one vehicle to be tracked and the reference vehicle based on the position of the lower end of the vehicle and the width information and calculates a distance between the at least one tracked vehicle and the reference vehicle to a predefined value And determines whether or not to warn the headway distance.

Images