KR101894731B1 - System and method for vehicle collision avoidance - Google Patents

Abstract

A collision avoidance system and method are disclosed.
According to an embodiment of the present invention, there is provided an anti-collision system comprising: a camera unit provided in a vehicle for capturing a stereo image; A ground map generation unit for generating a ground map indicating a disparity value of a ground in a stereo image; A weighted image generating unit for generating a weighted image by calculating a difference value between a pixel of the first image and a pixel of the second image matched according to the parallax value of the ground map; And a control unit for activating an alarm provided in the vehicle when an object existing on the ground approaches the vehicle by using the highlighted image.

Description

[0001] SYSTEM AND METHOD FOR VEHICLE COLLISION AVOIDANCE [0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a collision avoidance system, and more particularly, to a collision avoidance system and method of a vehicle using a stereo camera.

In order to acquire three-dimensional images, depth maps representing distance information of three-dimensional scenes should be acquired. Using depth information, images at the point in time that are not actually acquired can be synthesized through Depth Image Based Rendering

Depth information can be obtained with active depth sensors and passive depth sensors.

The active sensor method acquires the depth information directly using the physical sensor device, whereas the passive sensor method acquires and acquires the depth information from the image obtained with two or more cameras

Stereo matching, which is one of the passive sensor methods, is basically a technique for calculating the disparity existing in left / right images. It is a method of finding correspondence points of pixels on an epipolar line of two images .

Therefore, in general, parallax is calculated on a pixel-by-pixel basis, and an image composed of parallax values on a pixel-by-pixel basis can be defined as a disparity map or a depth map.

The parallax calculated in stereo matching is small for objects close to and large for objects far away.

Stereo cameras are available in a variety of digital cameras and smartphones.

The parallax map extracted through stereo matching is actively used in areas such as image refocusing and 3D image reconstruction. On the other hand, various attempts have been made to apply computer vision technology using image sensors in vehicles.

Especially, the technique of extracting scene depth using lidar sensor and constructing vehicle collision prevention system based on this is typical.

Much research has been done on systems that use both Raidasensor and stereo cameras.

Korean Patent Publication No. 10-2016-0116432 (published October 10, 2016)

The present invention provides a system and method for preventing collision of a vehicle using a stereo image.

It is also intended to provide a system and method for avoiding collision of a vehicle at low cost by replacing the expensive Lidar sensor with a stereo camera.

According to an aspect of the present invention, there is provided an image processing apparatus including a camera unit for photographing a stereo image provided in a vehicle; A ground map generating unit for generating a ground map representing a disparity value of the ground in the stereo image; A weighted image generating unit for generating a weighted image by calculating a difference value between a pixel of the first image and a pixel of the second image matched according to the parallax value of the ground map; And a control unit for activating an alarm provided in the vehicle when an object existing on the ground using the emphasis image approaches the vehicle.

Also, the ground map generator may generate the ground map using only the image included in the interest area, which is previously designated as the area where the ground is likely to exist in the moving direction of the vehicle in the stereo image.

Here, the region of interest may be the lower part of the center of the image of the stereo image.

Here, the ground map may have a representative parallax value for each scan line.

Also, the ground map generator calculates a representative parallax value of the upper scan line, calculates a representative parallax value of the upper scan line in the stereo image, and calculates a representative parallax value of the upper scan line, The time difference of the scan line can be searched.

In addition, the emphasized image generating unit may generate the emphasized image as a binary image by assigning 1 when the difference value exceeds the specified difference threshold value or 0 when the difference value is less than the specified difference threshold value.

Also, the controller may calculate the histogram by accumulating the number of pixels having a value of 1 in the Y-axis direction in the emphasized image, and may detect an object existing on the ground as a peak value of the histogram.

Also, the controller may estimate a position of the object in the X-axis direction and a moving direction of the object using the peak value of the histogram.

In addition, the controller may activate an alarm provided in the vehicle when moving in the center direction of the X-axis in the moving direction of the object.

Also, the controller may perform stereo matching on pixels in the emphasized image having a pixel value of 1 and an X-axis coordinate within a range specified by the peak value, wherein the stereo matching is performed within a predetermined dangerous parallax range If the number of pixels exceeds the danger threshold, the alarm of the vehicle can be activated.

Here, the danger parallax range may have a relatively large parallax range as the traveling speed of the vehicle is large.

According to another embodiment of the present invention, there is provided a method of preventing collision between an object on a ground and a vehicle, comprising the steps of: (a) obtaining a stereo image in front of the vehicle; (b) generating a ground map representing a disparity value of the ground in the stereo image; (c) calculating a difference value between pixels of the left and right images matching the parallax value of the ground map to generate a highlight image; And (d) activating an alarm on the vehicle when the object approaches the vehicle using the highlighted image.

Also, the step (b) may generate the ground map using only the image included in the region of interest, which is previously designated as the region where the ground is likely to exist in the moving direction of the vehicle in the stereo image.

In the step (b), a representative parallax value of the upper scan line may be calculated after calculating a representative parallax value of the lower scan line in the stereo image, and a representative parallax value of the upper scan line may be calculated And searching for the parallax of the upper scan line.

In addition, the step (c) may generate the binary image by assigning 1 when the difference value exceeds the specified difference threshold value or 0 when the difference value is less than the specified difference threshold value.

The step (d) includes the steps of: (d1) calculating a histogram by accumulating the number of pixels having a value of 1 in the weighted image in the Y-axis direction; And (d2) detecting an object existing on the ground as a peak value of the histogram.

The step (d) may further include estimating a position of the object in the x-axis direction and a moving direction of the object using the peak value of the histogram (d3).

The step (d) may further include activating an alarm provided in the vehicle when the moving direction of the object is the center of the X-axis.

The step (d) may further include the steps of: (d4) performing stereo matching on pixels in the emphasized image having a pixel value of 1 and an X-axis coordinate within a predetermined range of the peak value, And activating an alarm of the vehicle if the number of pixels matched within the threshold exceeds a critical threshold value.

According to another embodiment of the present invention, a computer-readable recording medium having recorded thereon a program for performing any one of the above-described methods for preventing collision can be provided.

The collision avoidance system and method according to the embodiment of the present invention has an advantage that a collision of a vehicle can be prevented by using a stereo image.

In addition, it is advantageous to replace the expensive Lidar sensor with a stereo camera to prevent collision of the vehicle at low cost.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram illustrating an internal configuration of a vehicle collision avoidance system according to an embodiment of the present invention; FIG.
2 is an example of a stereo image (KITTI data set).
3 is an example of a ground map according to an embodiment of the present invention.
4 is an example of a highlight image according to an embodiment of the present invention.
5 is an example of a histogram generated by the control unit according to an embodiment of the present invention.
6 is a sequential image taken in a situation in which the object is approaching the traveling direction of the vehicle.
7 is a highlight image obtained from the image of FIG.
FIG. 8 is a histogram generated by accumulating the number of pixels having a value of 1 in the Y-axis direction in the emphasized image acquired from FIG.
9 is an example of a continuous image captured when the object moves away from the traveling direction of the vehicle.
10 is a highlight image obtained from the image of FIG.
11 is a histogram generated by accumulating the number of pixels having a value of 1 in the Y-axis direction in the emphasized image acquired in FIG.
12 shows a stereo image in a case where a vehicle in a side lane passes a driver's vehicle while maintaining a lane.
13 is a highlight image obtained from the image of Fig.
FIG. 14 is a histogram generated by accumulating the number of pixels having a value of 1 in the Y-axis direction in the emphasized image acquired from the image of FIG.
15 is a flowchart illustrating a vehicle collision avoidance method according to an embodiment of the present invention;

BRIEF DESCRIPTION OF THE DRAWINGS The present invention is capable of various modifications and various embodiments, and specific embodiments are illustrated in the drawings and described in detail in the detailed description. It is to be understood, however, that the invention is not to be limited to the specific embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. In addition, numerals (e.g., first, second, etc.) used in the description of the present invention are merely an identifier for distinguishing one component from another.

Also, throughout the specification, when an element is referred to as being "connected" or "connected" with another element, the element may be directly connected or directly connected to the other element, It should be understood that, unless an opposite description is present, it may be connected or connected via another element in the middle.

Also, throughout the specification, when an element is referred to as "including" an element, it is understood that the element may include other elements as well, without departing from the other elements unless specifically stated otherwise. Also, the terms "a "," module ", and the like in the description mean a unit for processing at least one function or operation, which means that it can be implemented by one or more hardware or software or a combination of hardware and software .

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is a diagram illustrating an internal configuration of a vehicle collision avoidance system 100 according to an embodiment of the present invention. FIG. 2 is an example of a stereo image (KITTI data set) As shown in FIG.

1, a vehicle collision avoidance system 100 according to an embodiment of the present invention includes a camera unit 110, a ground map generation unit 120, a highlight image generation unit 130, a control unit 140, (150).

The components illustrated in FIG. 1 are shown separately for the sake of understanding and explanatory convenience of the present invention, but may be implemented as a part of the components provided in the existing vehicle.

The camera unit 110 can photograph a stereo image (i.e., left and right images) with a stereo camera.

Although the present embodiment is described as being provided in front of the vehicle, it may be provided on various sides of the vehicle according to the environment to which the present invention is applied.

In FIG. 2, a KITTI data set is illustrated as an example of a stereo image. In the following description, it is assumed that the camera 110 captures the image illustrated in FIG. 2.

The ground map generating unit 120 may generate a parallax image considering only the disparity value of the ground in the stereo image acquired from the camera unit 110. [ Hereinafter, an image representing a parallax value of a ground in an image is referred to as a 'ground map' in the present specification.

According to the embodiment of the present invention, the ground map generating unit 120 may set a region of interest according to a predetermined method in a stereo image acquired from the camera unit 110, and generate a ground map in a region of interest.

Here, the ROI is a predetermined area that is likely to be a ground in the traveling direction of the vehicle.

For example, in the case where the camera unit 110 is provided in front of the vehicle as described above, the ROI can be previously designated in the center lower part of the stereo image.

For example, when the camera unit 110 is provided on the right side of the vehicle, the region of interest may further include at least one of the left and right lower ends as well as the central lower end of the stereo image. And can be applied variously according to the requirements.

Therefore, according to the embodiment of the present invention, the ground map generation unit 120 can reduce the calculation load by processing only the image information included in the predetermined region of interest, without performing the arithmetic processing on all the regions of the stereo image .

The ground map generator 120 can generate the ground map using only the parallax values in the interest area of the stereo image.

The most basic local method used to calculate the parallax of a pixel is to find pixels of the right image corresponding to the pixels existing in the left image first and then to align the two images, a method of finding a pixel pair having the most similar color value within a specific search range on an epipolar line is used.

In the ground map according to the embodiment of the present invention, each pixel may have a representative scan line parallax value, which is a representative parallax value for each scan line, without having an independent parallax. Therefore, the ground map generation unit 120 according to the present embodiment obtains a representative parallax value for each scan line between the left and right images of the stereo image.

The ground map generator 120 according to the embodiment of the present invention may perform matching with the pixels in the search area of the right image with respect to each pixel of the left image in order to obtain the representative parallax of the scan line.

Here, the ground map generator 120 may compare color values of pixels or compare gradient values for matching, or may compare feature vectors extracted by applying a Cencus transform or the like.

The ground map generator 120 according to the embodiment of the present invention can use the difference between absolute values of pixels in the search area of the right image for each pixel of the left image. That is, the smaller the matching error (or the greater the similarity), the greater the likelihood that the corresponding pixel will be the corresponding pixel of the left image pixel with respect to the pixels in the search region of the right image.

For example, when the pixel corresponding to the left image x, y coordinate (10, 1) is searched in the right image, the ground map generator 120 calculates the absolute value of the pixel in the right image x, y coordinate (10-i, Calculate the absolute difference. In this case, i denotes a time difference, and a parallax value having a difference of the absolute value of the lowest among the candidate i values in the search area can be determined as a parallax value of the corresponding pixel.

Also, the ground map generator 120 may perform a matching operation on the pixels of the entire scan line, accumulate the parallax values when the absolute difference is the lowest, and histogramize the parallax values. The ground map generator 120 may use the parallax value corresponding to the highest value in the calculated histogram as the representative parallax value of the corresponding scan line.

In the present embodiment, it is assumed that the camera unit 110 is provided in front of the vehicle as described above. Therefore, since the image is acquired from the camera unit 110 attached to the vehicle, the stereo image has a feature that the bottom corresponds to the ground nearer to the vehicle and the ground moves away from the vehicle toward the upper side.

Therefore, it can be assumed that the lower portion in the region of interest of the stereo image has a larger parallax value than the upper portion (i.e., the depth value is smaller).

Therefore, the ground map generator 120 calculates the representative parallax value from the lowest scan line in the region of interest of the stereo image, and in the next upper scan line, The parallax search can be performed within a range that is smaller than or equal to the representative parallax value found in the parallax search range.

Through this method, the ground map generation unit 120 can obtain a ground map which shows a parallax with increasing distance from the bottom to the top, that is, a ground which is getting farther upward.

FIG. 3 illustrates an example of a ground map according to an embodiment of the present invention. In order to facilitate understanding and explanation of the present invention, a parallax value of a ground map is converted into a brightness value and illustrated. Referring to FIG. 3, it can be seen that the lower parallax value (that is, the higher brightness value in FIG. 3) and the lower parallax value (that is, the lower brightness value in FIG.

Referring back to FIG. 1, the emphasis image generator 130 may generate a highlight image by calculating a difference value between pixels of a stereo image matched according to a parallax value of a ground map.

More specifically, the emphasis image generation unit 130 calculates a difference value between the pixels of the right image corresponding to the representative parallax value obtained by the ground map for all the pixels of the left image, and generates a highlight image based thereon can do.

According to the embodiment of the present invention, when the left image pixels belong to the ground, the difference value can be calculated to be low because the pixels of the right image corresponding to the representative parallax values also have similar pixel values.

In contrast, when the left image pixel belongs to an object (automobile, pedestrian, etc.) lying on the ground, the pixel of the right image corresponding to the representative parallax value points to a position different from the object, and the difference of the corresponding pixel value may be large.

Therefore, if the difference between the absolute values of the two corresponding pixels is higher than a predetermined threshold value, the control unit 140 can determine that the pixel is a pixel corresponding to the object region existing on the ground.

The emphasis image generation unit 130 according to the embodiment of the present invention assigns 1 to a value of 1 if the difference value of the absolute value between the pixels between the corresponding right and left images corresponding to the representative parallax value of the ground map exceeds the specified difference threshold value, So that a highlight image can be generated as a binary image.

FIG. 4 shows an example of a highlight image according to an embodiment of the present invention. However, in order to understand the present invention, the emphasis image illustrated in FIG. 4 is exemplified by assigning an original color value to a pixel having a value of 1 in a highlight image, and visualizing it.

Referring again to FIG. 1, the control unit 140 can activate an alarm provided in the vehicle when an object existing on the ground approaches the vehicle using a highlight image.

To this end, the controller 140 estimates the X-axis direction position and the movement direction of the object in the highlighted image, and activates the alarm provided in the vehicle when the object movement direction is the x-axis center direction.

Hereinafter, the function of the control unit 140 will be described in detail.

The previously generated highlight image may be a result of only picking pixels that are likely to be on the ground. That is, in the emphasized image obtained every frame of the stereo image, the number of pixels located within the region of interest and having no parallax of the ground map (i.e., pixels having a pixel value of 1 in the emphasized image) Can be compared. If the number threshold is exceeded, it is likely that the object is near the front of the vehicle.

Accordingly, in this case, the controller 140 may generate a histogram in which the number of pixels having a pixel value of 1 is accumulated in the Y-axis direction in the emphasis image in the ROI in order to detect an object having a possibility of collision.

Here, if the density of the histogram has a high value at a specific position, it means that there is a high probability that the object exists at the corresponding position.

5 is an example of a histogram generated by the controller according to an embodiment of the present invention.

4 and 5, it can be seen that the density of the histogram is high at the position of the object located on the right side of the image.

The controller 140 may calculate an X coordinate having a peak value in the histogram to digitize the position of the object.

On the other hand, the histogram can be acquired somewhat unstably due to the noise included in the stereo image or errors included in the generation of the ground map. In this case, the controller 140 may first calculate the X-coordinate of the peak after smoothing the histogram.

According to another embodiment of the present invention, the controller 140 applies a Gaussian model or a Gaussian mixture model to the histogram to find a position corresponding to the average of the Gaussian model as a peak value, Can be calculated.

The control unit 140 can calculate the X-axis position of the peak when the number of pixels having the pixel value of 1 in the emphasized image within the ROI is greater than the number threshold value for every image frame.

Also, the controller 140 can estimate the moving direction of the object from the peak position change pattern.

FIG. 6 is a continuous image taken in a state in which the object approaches the traveling direction of the vehicle, FIG. 7 is a weighted image obtained in the image of FIG. 6, and FIG. The histogram is generated by accumulating the number in the Y-axis direction.

Referring to FIGS. 6 to 8, when the object approaches the traveling direction of the vehicle, it can be seen that the position of the peak value of the histogram moves in the X-axis center direction in FIG.

Accordingly, when the X-axis position of the peak value of the histogram moves in the X-axis center direction, the controller 140 detects that the object is approaching the traveling direction of the vehicle, and activates the alarm provided in the vehicle.

Here, the alarm activated by the control unit 140 may be provided to an existing vehicle, and may be sound, illumination, or the like designated to the driver of the vehicle, and may further include an alarm to be transmitted to an object approaching the outside through a horn or the like.

9 is an example of a continuous image taken when the object moves away from the traveling direction of the vehicle, FIG. 10 is a highlight image obtained from the image of FIG. 9, and FIG. The histogram is generated by accumulating the number of pixels in the Y-axis direction.

9 to 11, when the object moves away from the traveling direction of the vehicle, it can be seen that the peak position of the histogram moves away from the center of the X axis.

Accordingly, the controller 140 does not activate the alarm when the X-axis position of the peak value of the histogram moves away from the X-axis center, and can release the activated alarm.

All the objects for which the X axis position of the peak value of the emphasis image moves to the X axis center direction (i.e., the traveling direction of the vehicle) may not be a dangerous object. For example, assume that a vehicle in a side lane keeps a side lane and overtakes the driver's vehicle.

FIG. 12 is a stereo image in the case where the vehicle in the side lane keeps the lane and overtakes the driver's vehicle, FIG. 13 is a highlight image obtained from the image in FIG. 12, FIG. 14 is a highlight image obtained in the image in FIG. Is a histogram generated by accumulating the number of pixels having a value of 1 in the Y-axis direction.

According to the embodiment of the present invention, when the position of the peak of the histogram moves in the X-axis center direction in order to detect only a dangerous object, the controller 140 performs stereo matching ), The alarm of the vehicle can be activated when the number of pixels matched within the predetermined danger parallax range exceeds the danger threshold value.

According to the embodiment of the present invention, in consideration of the running speed of the vehicle, a parallax range corresponding to a dangerous position at which there is a risk of collision with the vehicle in an image photographed in the forward direction of the vehicle can be specified in advance.

For example, the higher the traveling speed of the vehicle, the wider the range of parallax can be specified, and vice versa, a narrower range of parallax can be specified.

Accordingly, when the stereo matching is performed within a predetermined dangerous parallax range (i.e., when the difference between the absolute values of the corresponding pixels of the left and right images is less than or equal to the specified matching threshold value) . If the number of pixels to be matched within the dangerous parallax range exceeds the designated danger threshold value, the control unit 140 can activate an alarm provided in the vehicle.

The internal construction of the anti-collision system 100 illustrated in FIG. 1 has been described in detail with reference to FIGS. 2 to 14. FIG.

Here, it is to be clarified that the division of the components illustrated in FIG. 1 is merely a division by main functions that each component is responsible for.

That is, two or more constituent parts may be combined into one constituent part, or one constituent part may be divided into two or more functions according to functions that are more subdivided.

In addition, each of the constituent units may additionally perform some or all of the functions of the other constituent units in addition to the main functions of the respective constituent units, and some of the main functions of the constituent units are dedicated to the other constituent units Of course.

15 is a flowchart illustrating a vehicle collision avoidance method according to an embodiment of the present invention.

Below. 15, the internal structure of the anti-collision system 100 of FIG. 1 can be variously modified and implemented as described above. Therefore, System 100 will be described.

In order to facilitate understanding and explanation of the collision avoidance method, a description overlapping with those described with reference to Figs. 1 to 14 will be simplified or omitted.

In step S1510, the vehicle collision avoidance system 100 may acquire a stereo image through a camera provided in the vehicle.

Subsequently, in step S1520, the anti-collision system 100 may generate a ground map reflecting only the parallax value of the ground in the stereo image.

At this stage, the collision avoidance system 100 does not operate on all the pixels of the stereo image, but performs only the predetermined region of interest, that is, only the region where the ground is likely to exist (for example, the lower center portion of the image) .

Also, the collision avoidance stem may generate a ground map by calculating a representative parallax value for each scan line in the ROI.

In addition, the collision avoidance system 100 may calculate the representative parallax value in the lowermost scalar line within the region of interest, and then search for the parallax within a range smaller than or equal to the representative parallax value of the lower edge in the scribe line.

Subsequently, in step S1530, the anti-collision system 100 may generate a highlight image by calculating a difference value between pixels of the left and right images matching the parallax value of the ground map.

In addition, the collision avoidance system 100 may generate a binary image by assigning 1 when the calculated difference value exceeds the specified difference threshold value, or 0 when the calculated difference value exceeds the specified difference threshold value.

In step S1540, the collision avoidance system 100 calculates a histogram by accumulating the number of pixels having a value of 1 in the emphasis image in the Y-axis direction, and detects an object existing on the ground as a peak value of the histogram can do.

More specifically, the object can be detected by calculating the X axis position of the peak value of the histogram and estimating the moving direction of the object according to the X axis position change.

Then, in step S1550, the collision avoidance system 100 can activate the alarm provided in the vehicle in step S1560 if the X-axis movement direction of the sensed object is the vehicle's travel direction.

Here, the collision avoidance system 100 may be configured to detect only the object

When the position of the peak of the histogram moves in the direction of the center of the X axis, stereo matching is performed on the pixels within the specified range from the peak value, and the number of pixels matched within the predetermined danger parallax range is If the danger threshold is exceeded, the vehicle alarm can be activated.

According to the embodiment of the present invention, in consideration of the running speed of the vehicle, a parallax range corresponding to a dangerous position at which there is a risk of collision with the vehicle in an image photographed in the forward direction of the vehicle can be specified in advance.

For example, the higher the traveling speed of the vehicle, the wider the range of parallax can be specified, and vice versa, a narrower range of parallax can be specified.

The anti-collision method according to the embodiment of the present invention has been described with reference to FIG.

The above-described collision avoidance method according to the embodiment of the present invention can be implemented as a computer-readable code on a computer-readable recording medium. The computer-readable recording medium includes all kinds of recording media storing data that can be decoded by a computer system. For example, it may be a ROM (Read Only Memory), a RAM (Random Access Memory), a magnetic tape, a magnetic disk, a flash memory, an optical data storage device, or the like. In addition, the computer-readable recording medium may be distributed and executed in a computer system connected to a computer network, and may be stored and executed as a code readable in a distributed manner.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the following claims And changes may be made without departing from the spirit and scope of the invention.

100: Collision avoidance system 110:
120: a ground map generating unit 130: a highlight image generating unit
140:

Claims (20)

A camera unit provided in the vehicle for photographing a stereo image;
A ground map generating unit for generating a ground map representing a disparity value of the ground in the stereo image;
A weighted image generating unit for generating a weighted image by calculating a difference value between a pixel of the first image and a pixel of the second image that match the parallax value of the ground map, One of the left / right images acquired by the camera unit and the other; And
And a controller for activating an alarm provided in the vehicle when an object existing on the ground approaches the vehicle using the emphasis image,
Wherein the ground map generator comprises:
The difference between the absolute values of the pixel values of the pixels in the first image of the corresponding scan line in the X-axis direction is the lowest A parallax value corresponding to a highest value is set as a representative parallax value of the corresponding scan line when a histogram storing the determined parallax values is generated after determining a parallax value as a parallax value for each pixel of the stereo image, And generates a ground map having a representative parallax value for every scan line in the axial direction.
The apparatus of claim 1, wherein the ground map generator
And generating a ground map using only an image included in a region of interest that is predetermined in an area in which the ground is likely to exist in the moving direction of the vehicle in the stereo image,
3. The method of claim 2,
Wherein the region of interest is the bottom of the image center of the stereo image.
delete The apparatus of claim 1, wherein the ground map generator
A representative parallax value of the upper scan line is calculated and a parallax of the upper scan line is searched within a range smaller than or equal to the representative parallax value of the lower scan line in the stereo image , Collision avoidance system.
The method according to claim 1,
Wherein the emphasized image generating unit comprises:
And generates 1 if the difference value exceeds a specified difference threshold value, or 0 if the difference value is less than a specified difference threshold value, to generate the emphasized image as a binary image.
The method according to claim 6,
Wherein the control unit calculates a histogram by accumulating the number of pixels having a value of 1 in the emphasis image in the Y axis direction and detects an object existing on the ground as a peak value of the histogram,
8. The method of claim 7,
Wherein the controller estimates a position of the object in the x-axis direction and a moving direction of the object using the peak value of the histogram,
9. The method of claim 8,
Wherein the controller activates an alarm provided in the vehicle when the moving direction of the object is the center of the X-axis.
9. The method of claim 8,
Wherein the control unit performs stereo matching on pixels having a pixel value of 1 and a X-axis coordinate within a range specified by the peak value in the emphasized image, And activates an alarm of the vehicle if the number exceeds a critical threshold.
11. The method of claim 10,
Wherein the dangerous parallax range has a relatively large parallax range as the running speed of the vehicle is greater.
A method for preventing collision between an object on the ground and a vehicle,
(a) obtaining a stereo image in front of the vehicle;
(b) generating a ground map representing a disparity value of the ground in the stereo image;
(c) calculating a difference value between pixels of a left image and a right image matched according to a parallax value of the ground map to generate a highlight image; And
(d) activating an alarm provided in the vehicle when the object approaches the vehicle using the emphasis image,
The step (b) may further include the step of, for each pixel of the left image of the corresponding scan line in the X-axis direction, matching an absolute value of pixel values between the pixels in the search area of the right image The parallax value corresponding to the highest value when the histogram in which the determined parallax values are accumulated is determined as the parallax value for each pixel of the stereo image, , Thereby generating a ground map having a representative parallax value for every scan line in the X-axis direction.
13. The method of claim 12,
Wherein the step (b) comprises: generating a ground map using only an image included in a region of interest that is predetermined as an area in which the ground is likely to exist in the moving direction of the vehicle in the stereo image.
13. The method of claim 12,
The step (b)
A representative parallax value of the upper scan line is calculated and a parallax of the upper scan line is searched within a range smaller than or equal to the representative parallax value of the lower scan line in the stereo image The method comprising the steps of:
13. The method of claim 12,
The step (c)
Generating a binary image by assigning 1 if the difference value exceeds a specified difference threshold value, and 0 if the difference value is less than or equal to a specified difference threshold value.
16. The method of claim 15,
The step (d)
(d1) calculating a histogram by accumulating the number of pixels having a value of 1 in the weighted image in the Y-axis direction; And
(d2) detecting an object existing on the ground as a peak value of the histogram.
17. The method of claim 16,
The step (d)
(d3) estimating a position of the object in the x-axis direction and a moving direction of the object using the peak value of the histogram.
18. The method of claim 17,
The step (d)
(d4) activating an alarm provided in the vehicle when the direction of movement of the object is the center of the X-axis.
18. The method of claim 17,
The step (d)
(d4) stereo matching is performed on pixels in the emphasized image in which the pixel value is 1 and the X-axis coordinate is within the range specified by the peak value, and the number of pixels matched within the predetermined danger parallax range Further comprising activating an alarm of the vehicle if the risk exceeds a critical threshold.
A computer-readable recording medium having recorded thereon a program for performing a collision avoidance method according to any one of claims 12 to 19.

Images