KR101180676B1 - A method for controlling high beam automatically based on image recognition of a vehicle - Google Patents

Abstract

This document discloses a method for automatically controlling a vehicle's high beams. According to the automatic control method of the high beam disclosed in this document, the day and night are determined based on the image captured by the front camera based on image recognition, and the presence or absence of the front vehicle is controlled to control the high beam on and off according to the determination result. do. In particular, the determination of day and night is improved according to the area classification method, and the accuracy of the forward vehicle determination is improved by analyzing light source elements such as light source position and motion in consideration of vehicle characteristics.

High beam automatic control

Description

A METHOD FOR CONTROLLING HIGH BEAM AUTOMATICALLY BASED ON IMAGE RECOGNITION OF A VEHICLE}

The present invention relates to the automatic control of the high beam of the vehicle, and more particularly to the automatic control of the high beam based on the image recognition of the vehicle.

In the automatic control of the headlight uplight of a vehicle, a light installed at one side of the headlight of a vehicle so that a driver of a vehicle in the opposite lane during the night driving does not obstruct the view of the vehicle by a high beam so that the risk of an accident of a relative vehicle can be eliminated. The output signal of the comparator is input to the non-inverting terminal (+) on one side of the sensing sensor and the non-inverting terminal (+) on the other hand, and the reference value (Vref) is input to the inverting terminal (-) on the other side. A method using various sensors, such as an automatic headlight booster automatic control device that automatically flashes a high beam on the basis, has been applied.

SUMMARY OF THE INVENTION The present invention has been made in the above-mentioned background, and provides a method for automatically controlling a vehicle's image recognition-based upward lamp.

According to an aspect of the present invention, there is provided a method of automatically controlling a vehicle's image recognition-based upward light as a means for solving the above-mentioned problems. Day and night determination step of determining whether the detection, the vehicle candidate detected by the light source extracted from the image, one of the preceding vehicle and the opposite vehicle, the vehicle detection step, and if it is determined that at night and both the preceding vehicle and the opposite vehicle are not detected and controlling to turn on the high beam.

In the day and night determination step, the average RGB value for each area may be calculated and compared with a threshold set for each area, and may be determined to be night if it is less than or equal to the threshold.

In the vehicle detecting step, a distribution map may be generated for a change in a light source extracted from a plurality of consecutive images of the image, and the measured probability of the vehicle candidate may be calculated based on the generated distribution map to calculate the probability of the vehicle to detect the vehicle. Can be.

A Gaussian distribution may be generated based on an average and a standard deviation of the Y-axis position values according to the X-axis position values of the light sources in the image, while generating a distribution diagram for the positional change of the light source.

While generating the distribution of the shape change of the light source, a Gaussian distribution can be generated by the mean and standard deviation of the X, Y-axis size according to the Y-axis position value of the light source in the image.

While generating a distribution chart for the change in the movement of the light source, a Gaussian distribution chart may be generated based on the mean and standard deviation of the moving angle values according to the Y-axis position value of the light source in the image.

In the vehicle detecting step, when there are a large number of pixels having a red component in the vehicle candidate attention from the red detection image of the image, it may be determined that there is a preceding vehicle.

In the vehicle detecting step, when the two light sources at the same height on the Y axis in the video image appear constant in the X-axis value, it may be determined that the vehicle detection.

In the vehicle detecting step, the overtaking vehicle may be detected using the optical flow and the light source color of the light source extracted from the image.

In the controlling step, the operation may be performed only in consideration of the vehicle speed at a speed higher than a predetermined speed, and if the speed is maintained at a speed lower than the predetermined speed for a while, the operation may be maintained.

According to the present invention, a high beam may be automatically controlled based on an image. As a result, convenience is increased by not having to perform cumbersome high beam flashing operation regardless of whether the vehicle is in the opposite lane and at the same time, there is an effect of eliminating the accident risk of the relative vehicle without disturbing the vision of the opposite lane vehicle driver. In addition, various solutions based on image are presented, which improves control performance.

DETAILED DESCRIPTION OF THE INVENTION Hereinafter, embodiments of the present invention related to an automatic control of an overhead light for automatically turning on and off a high beam without a driver's operation according to a situation determined based on image recognition using a front camera installed in a vehicle will be described. Do it.

FIG. 1 is a structural diagram of an automatic high beam control system according to an embodiment of the present invention, and FIG. 2 is a structural diagram of a control unit in an automatic high beam control system according to an embodiment of the present invention.

According to the present embodiment, as shown in FIG. 1, the automatic control system for the high beam lamp includes an image processing universal controller 100, an image module 110, an image conversion board 120, a high beam 130, and a driver 140. In addition, the data storage unit 150 controls the turning on and off of the high beam 130 based on the image processing result.

In particular, referring to FIG. 2, the image processing universal control unit 100 receives a video captured by the front camera 200 and decodes the video decoder unit 10 and a power supply unit 11 that supplies power to the front camera 200. In addition, the video processing unit 10 may process an image necessary for controlling an image-based high beam 130 by using the signal processed by the video decoder 10 and may include a digital signal processor (DSP) 12. The digital signal processor 12 may process the downlight 15 input signal and output a high beam 18 control signal based on the image processing result. In addition, the memory unit 13 including a temporary storage (DDR RAM) / permanent storage (FLASH ROM) for the data storage for the operation of the digital signal processing unit 12, a transceiver for transmitting and receiving signals inside and outside the control unit 100 (HS CAN) TRANCEIVER, 14).

3 is a flowchart of a method for automatically controlling a high beam according to an embodiment of the present invention.

Referring to FIG. 3, first, an image is acquired in step S300 in order to automatically control a high beam according to the present embodiment. In operation S310, it is determined whether the vehicle external environment is day or night. In step S320, it is detected whether the host vehicle ahead or the counter vehicle exists. Then, in step S330 it is controlled to turn on or off the high beam according to the day and night determination result of step S310 and the presence of the preceding vehicle or the counter vehicle of step S320. That is, at night, a high beam is turned on, but when it is bright enough due to a tail light or a high beam of the preceding vehicle or the opposite vehicle, the high beam is turned off.

More specifically, the day and night determination method of step S310, ROI separation process for dividing the region in the video image of step S312 into, for example, the sky region and the ground region, color information for each region of step S314, for example RGB value calculation processing process The process may include comparing the area-specific RGB value of step S316 with a reference threshold, and day / night determination process of step S318.

The determination method of the preceding vehicle or the counter vehicle in step S320 includes the image preprocessing step in step S321, the vehicle candidate detection step in step S322, the vehicle and vehicle candidate tracking step in step S323, the vehicle candidate verification step in step S324, and the vehicle in step S325. A candidate conversion process, a vehicle and vehicle candidate removing process in step S326, vehicle information stored as a vehicle in the vehicle list in step S327, for example, a process of outputting a detected vehicle number.

The method of controlling the high beam of step S330 is positive if there is no vehicle according to the preceding vehicle or the counter vehicle presence information of step S320 in step S331, and if it is night determination according to the day and night judgment result of step S310 in step S332. If the flow goes to step S336 as affirmative and both are positive in step S336, the control is performed by turning on a high beam in step S337.

In this case, in step S330, if the predetermined speed, for example, 40 KPH or more is satisfied based on the vehicle speed measured in step S340 (S333), or if the downlight is turned on based on the state of the downlight determined in step S350 (S334). On the basis of the steering angle measured in step S360, it may be controlled to turn on or off the high beam by further considering the case (S335).

4 is a diagram illustrating an example of image segmentation during the day and night determination in the automatic high beam control according to an embodiment of the present invention.

In the image-based automatic high beam control, the image is controlled to turn off the high beam when it is determined to be daytime, and when it is determined to be night, it is controlled to turn on the high beam. For this purpose, it is important to accurately distinguish between day and night through images. According to the present exemplary embodiment, the sky area 410 and the ground area 420 are divided and processed among the entire image screen 400 of the acquired image. For example, the sky area may be defined as an area of 1 to 55 pixels, and the ground area may be defined as an area of 51 to 90 pixels.

FIG. 5 is a flowchart illustrating a method for determining day and night according to an image segmentation method in automatically controlling a high beam according to an exemplary embodiment of the present invention.

According to the present exemplary embodiment, after acquiring the front image in step S500, the sky area 410 is separated in step S510, an RGB value for the sky area 410 is extracted in step S520, and an RGB average value is calculated in step S530. In operation S570, when the RGB average value is smaller than or equal to the threshold value compared to the threshold value set in the sky area 410, the process proceeds to step S580.

In step S540, the ground area 420 is separated, and in step S550, an RGB value for the ground area 420 is extracted, and in step S560, an RGB average value is calculated. In operation S570, when the RGB average value is less than or equal to the threshold value compared to the threshold value set in the ground area 420, the process proceeds to step S580.

If the RGB average value for the sky area 410 is less than or equal to the threshold value and the RGB average value for the ground area 420 is also less than or equal to the threshold value in step S580, the process proceeds to step S590 to turn on the high beam. To control.

6 to 9 are graphs showing the performance verification of the day and night determination method according to the image segmentation method in the automatic automatic control of the upward light according to an embodiment of the present invention.

6 is a graph showing that the night is determined in the night condition, FIG. 7 is a graph showing the day is determined in the daytime condition, FIG. 8 is a graph indicating that the night is determined in the tunnel condition at night, and FIG. 9 is a tunnel at the daytime condition. It is a graph showing that night is determined even under the conditions.

In general, since the front camera for photographing the front image is mounted at a fixed position, as shown in FIGS. 6 to 9, it can be confirmed that the area classification method produces a relatively accurate determination result. In this case, the threshold for turning on the high beam and the threshold for turning off the high beam may be dualized and set separately for each case. 6 to 9, it can be seen that a threshold value (blue line) for turning off the high beam is set to be larger than a threshold value (red line) for turning on the high beam.

10 is a flowchart illustrating a method of detecting a vehicle in automatic control of a high beam according to an exemplary embodiment of the present invention.

In the automatic control of the high beam, even when the vehicle is at night, the preceding vehicle or the counter vehicle is sufficiently bright, so that the vehicle detection is applied in that the surrounding can be grasped even if the high beam is not turned on. At this time, since it is determined based on the front image, it is important to accurately track the reflection plate and the front vehicle on the road or the main surface.

According to the present embodiment, first, an image preprocessing process is performed in step S600, a vehicle candidate detection and tracking process in step S610, a vehicle candidate verification process in step S620, and a vehicle candidate determination process in step S630.

More specifically, according to the image preprocessing process of step S600, an original image is obtained in step S601 and binarized image processing in step S602. FIG. 11 is a diagram illustrating before and after binarization processing in vehicle detection for automatic control of an overhead light according to an embodiment of the present invention. FIG. According to this embodiment, only Y components are extracted from the YCbCr image and binarized. This is to reduce the image processing time and find the light source by the brightness value.

In step S603, the light source is extracted. 12 is a diagram illustrating before and after a light source extraction process in vehicle detection for automatic control of a high beam according to an embodiment of the present invention. A light source is extracted from the Y image extracted in FIG. 11. Since only the headlights and taillights of the vehicle are visible at night, the light source is an important factor in determining the existence of the vehicle. The farther the light source is, the weaker the brightness is. Therefore, the lower the vertical axis value is, the lower the light source determination threshold value is preferably set.

In step S604, the RGB image is processed using the original image acquired in step S601, and the red detection process is performed in step S605. 13 is a diagram illustrating before and after a red detection process in vehicle detection for automatic control of a high beam according to an embodiment of the present invention. It is because it is very effective to detect the red component pixel based on the rear light source color (red color) in detecting the preceding vehicle, especially the front vehicle.

According to the vehicle candidate detection and tracking process of step S610, a new vehicle candidate is detected in step S611. In the extracted light source image, the vehicle candidates are grouped to store information such as position and shape in a memory, and the positions of the vehicle candidates on the image are found by projecting onto the image X and Y axes. Then, it stores information such as location, variance and size of found candidates and adds them to candidate list. In step S612, the position of the vehicle candidate is tracked, and in step S613, the position of the vehicle is tracked. Set the search area in consideration of the candidates stored in the vehicle candidate list and the position and size of the previous frame of the vehicle stored in the vehicle list, and if it is found again in the search area, the new position, distribution, size, etc. Update the information of the vehicle and track the location of the candidates.

 According to the vehicle candidate verification process of step S620, it is determined whether the vehicle candidate is a vehicle in consideration of factors such as the light source position, the shape of the light source, the light source movement, the light source color, the light source clustering, and the overtaking vehicle.

First, referring to FIG. 14 for the light source position. 14 is a view for explaining a method of considering the light source position in the vehicle detection for automatic control of the high beams according to an embodiment of the present invention. According to the present embodiment, since the front camera is generally fixedly installed, it is conceived that the preceding vehicle and the opposite vehicle pass almost the same area in the image, and the Gaussian is represented by the mean and standard deviation of the Y-axis position value according to the X-axis position value. Create a distribution map and store it in memory. The probability of the vehicle is calculated by comparing the stored Gaussian distribution and the position of the vehicle candidate.

For the shape of the light source, see FIG. 15. 15 is a view for explaining a method of considering the shape of the light source in the vehicle detection for automatic control of the high beams according to an embodiment of the present invention. According to the present embodiment, the shape and size of the light source from the vehicle are different according to the detected distance of the vehicle. The Gaussian distribution is generated by the mean and standard deviation of the X and Y axis sizes according to the Y axis position value. And store it in memory. The probability of the vehicle is calculated by comparing the stored Gaussian distribution and the position of the vehicle candidate.

See FIG. 16 for light source movement. FIG. 16 is a view illustrating a method of considering light source movement in vehicle detection for automatically controlling a high beam according to an embodiment of the present invention. According to the present embodiment, it is conceived that the movement of the vehicle shown in the image has almost constant directionality. The Gaussian distribution is generated as the average and the standard deviation of the angle value of the light source moved according to the Y-axis position value and stored in the memory. do. The probability of the vehicle is calculated by comparing the stored Gaussian distribution and the angle of the candidate.

See FIG. 17 for the light source color. FIG. 17 is a view illustrating a method of considering a light source color in vehicle detection for automatically controlling a high beam according to an embodiment of the present invention. According to the present exemplary embodiment, since the tail light of the preceding vehicle has a red color, it is determined as a vehicle when a large number of pixels having a red component around the vehicle candidate exist from the red detection image.

For light source clustering, see FIG. 18. 18 is a diagram illustrating a method of considering light source clustering in vehicle detection for automatic control of an overhead light according to an embodiment of the present invention. According to this embodiment, the light source in the vehicle is generally conceived in that both appear at the same time and have similar size and movement. If it appears constant, and the shape and direction of the light source are the same, the likelihood of being a light source of a vehicle increases.

Refer to FIG. 19 for the overtaking vehicle. 19 is a view for explaining a method of verifying a passing vehicle in vehicle detection for automatic control of a high beam according to an embodiment of the present invention. According to the present embodiment, it is important to recognize quickly because the preceding vehicle to be overlooked in a relatively close position is conceived in that it moves quickly with a certain direction. For example, you can use Optical Flow and Light Source Color to quickly recognize objects moving on either side of the camera.

According to the vehicle candidate determination process in step S630, the vehicle candidate is converted in step S631, the vehicle candidate is removed in step S632, and the vehicle is removed in step S633.

20 is a flowchart illustrating a method of detecting a vehicle in automatic control of a high beam according to an embodiment of the present invention.

According to the present embodiment, it is determined 810 that the vehicle is based on the input information 800, and the result of the determination is output 820. As described above, the input information 800 includes measurement results for the light source position 801, the light source shape 802, the light source movement 803, the light source color 804, the light source clustering 805, and the overtaking vehicle 806. May be included.

If the measurement results satisfy the vehicle determination criteria, it is determined that the vehicle is a vehicle. For example, when the vehicle probability according to the light source position 801 exceeds 32% (S811), and when the vehicle probability according to the light source shape 802 exceeds 32% (S812), the light source movement 803 If the probability of the vehicle exceeds 32% (S813), if it is determined as a tail light according to the light source color 804 (S814), if the light source clustering 805 (S815), and counts as the number of vehicles in step S817 When the number of vehicles is 2 or more in step S818 or when the overtaking vehicle 806 is detected in step S819 and a tail light is detected, the control unit turns on the high beam in step S822. If the vehicle probability according to the light source position 801 does not exceed 32% (S811), and the number of vehicles is not two or more in step S818, it is controlled by turning off the high beam in step S821.

21 is a graph showing a result of automatic high beam control according to an embodiment of the present invention.

The automatic control of the high beam according to the present embodiment applies the operation time difference to determine the condition of turning on the high beam and turning off the high beam after a predetermined time (3000 ms). After a predetermined time (<= 100ms) can be controlled to turn off the actual high beam (high beam).

22 is a graph of control results when the vehicle speed is considered in the automatic control of the high beam according to the exemplary embodiment of the present invention.

According to the present embodiment, in controlling the high beam, the current vehicle speed information may also be taken into consideration. For example, if the vehicle speed is less than 40KPH, the automatic high beam control system may be set to not operate. In addition, when driving for more than 40KPH in a vehicle speed is continued for a predetermined time, for example, 5 seconds or more, even if the vehicle speed is less than 40KPH, for example, the vehicle speed is more than 36KPH, for example, the driving speed is recognized for a predetermined time, for example, 5 seconds. This will prevent the high beam on and off control from repeating beyond the reference.

Although described above with reference to the drawings and embodiments, those skilled in the art that the present invention can be variously modified and changed within the scope without departing from the spirit of the invention described in the claims below I can understand.

1 is a structural diagram of an automatic high beam control system according to an embodiment of the present invention.

Figure 2 is a structural diagram of a control unit in the automatic high beam control system according to an embodiment of the present invention.

3 is a flowchart of a method for automatically controlling a high beam according to an embodiment of the present invention.

4 is a diagram illustrating an example of image segmentation during day and night determination in automatic high beam control according to an embodiment of the present invention.

5 is a flowchart of a day / night determination method according to an image segmentation method in automatic high beam control according to an embodiment of the present invention.

6 to 9 are graphs showing the performance verification of the day and night determination method according to the image segmentation method in the automatic automatic control of the upper lamp according to an embodiment of the present invention.

10 is a first flowchart of a vehicle detection method in automatic high beam control according to an exemplary embodiment of the present invention.

FIG. 11 is a diagram illustrating before and after binarization processing in vehicle detection for automatic control of an overhead light according to an embodiment of the present invention; FIG.

12 is a view showing before and after a light source extraction process in vehicle detection for automatic high beam control according to an embodiment of the present invention.

13 is a view showing before and after a red detection process in vehicle detection for automatic control of a high beam according to an embodiment of the present invention.

14 is a view for explaining a method of considering the light source position in the vehicle detection for automatic control of the high beams according to an embodiment of the present invention.

15 is a view for explaining a method of considering the shape of the light source in the vehicle detection for automatic control of the high beams according to an embodiment of the present invention.

FIG. 16 is a view for explaining a method of considering light source movement in vehicle detection for automatic control of a high beam according to an embodiment of the present invention; FIG.

FIG. 17 is a view illustrating a method of considering a light source color in vehicle detection for automatic control of a high beam according to an embodiment of the present invention. FIG.

18 is a view for explaining a method for considering light source clustering in vehicle detection for automatic control of a high beam according to an embodiment of the present invention.

19 is a view for explaining a method for verifying a passing vehicle in vehicle detection for automatic control of a high beam according to an embodiment of the present invention.

20 is a second flowchart of a vehicle detection method in automatic high beam control according to an embodiment of the present invention.

21 is a graph showing a result of automatic high beam control according to an embodiment of the present invention.

22 is a control result graph in the case of considering the vehicle speed in the automatic control of the high beam according to an embodiment of the present invention.

Claims (10)

In the automatic recognition method of the vehicle based image recognition, Obtaining a forward image; Day and night determination step of determining whether or not it is night by comparing the color information for each area divided in the video image with the reference set for each area; Detecting a vehicle candidate detected by the light source extracted from the image as one of a preceding vehicle and an opposite vehicle; And Controlling to turn on a high beam when it is determined to be nighttime and fails to detect both the preceding vehicle and the opposite vehicle; Including, In the vehicle detecting step, a distribution map is generated for a change of a light source extracted from a plurality of consecutive images of the image, and a vehicle probability is detected by calculating a probability of the vehicle by comparing measured values of the vehicle candidates based on the generated distribution map. Characterized in that, the vehicle image recognition-based upward light automatic control method. The method according to claim 1, In the day and night determination step, the average RGB value for each area is calculated and compared with the threshold value set for each area, and if the same or less than the threshold value, characterized in that the night recognition based on the vehicle's image recognition based automatic automatic control method. delete The method according to claim 1, Generate a distribution map for the change in the position of the light source, generating a Gaussian distribution by the average and the standard deviation of the Y-axis position value according to the X-axis position value of the light source in the image, automatic recognition based on the vehicle's image recognition Way. The method according to claim 1, Generate a distribution of the shape change of the light source, but generates a Gaussian distribution with the mean and standard deviation of the X, Y-axis size according to the Y-axis position value of the light source in the image, the vehicle image recognition-based uplight automatically Control method. The method according to claim 1, Generating a distribution chart for the change in the movement of the light source, generating a Gaussian distribution map of the average and the standard deviation of the moving angle value according to the Y-axis position value of the light source in the image, automatic recognition based on the vehicle's image recognition Way. The method according to claim 1, In the vehicle detecting step, if there are a large number of pixels having a red component around the vehicle candidate from the red detection image of the image, characterized in that the preceding vehicle, characterized in that the automatic recognition of the vehicle based image recognition. In the automatic recognition method of the vehicle based image recognition, Obtaining a forward image; Day and night determination step of determining whether or not it is night by comparing the color information for each area divided in the video image with the reference set for each area; Detecting a vehicle candidate detected by the light source extracted from the image as one of a preceding vehicle and an opposite vehicle; And Controlling to turn on a high beam when it is determined to be nighttime and fails to detect both the preceding vehicle and the opposite vehicle; Including, In the vehicle detection step, when the two light sources at the same height on the Y axis in the image image is determined that the difference between the X-axis value is constant, it characterized in that the vehicle detection based on automatic recognition of the image recognition of the vehicle. In the automatic recognition method of the vehicle based image recognition, Obtaining a forward image; Day and night determination step of determining whether or not it is night by comparing the color information for each area divided in the video image with the reference set for each area; Detecting a vehicle candidate detected by the light source extracted from the image as one of a preceding vehicle and an opposite vehicle; And Controlling to turn on a high beam when it is determined to be nighttime and fails to detect both the preceding vehicle and the opposite vehicle; Including, In the vehicle detecting step, the overtaking vehicle is detected using the optical flow and the light source color of the light source extracted from the image. The method according to claim 1, In the control step, in consideration of the vehicle speed is operated only at a certain speed or more, and if maintained at a predetermined speed or more for a predetermined time, the operation is controlled so as to be maintained even if it is operated at a speed lower than the predetermined speed for a predetermined time. Automatic image recognition based on the vehicle automatic control method.

Images