CN103077371A - Method and apparatus for lane detection of vehicle - Google Patents

Abstract

The present invention applies a weighted value of multiple wave filters and each wave filter on an image which is acquired by a camera, thereby eliminating an error identification problem of lane and stably applying LKAS control. Namely the method for lane detection of vehicle comprises the following steps: applying more than two left partial wave filters on left lane information which is acquired from a vehicle image device; applying more than two right partial wave filters on right lane information which is acquired from the vehicle image device; calculating a penalty factor of the left lane information and the right lane information; applying the penalty factor on the result which is filtered by more than two left partial wave filters for calculating a left lane calculation value; and applying the penalty factor on the result which is filtered by more than two right partial wave filters for calculating a right lane calculation value.

Description

The lane recognition method of vehicle and device thereof

Technical field

The present invention relates to lane recognition method and the device thereof of vehicle, specifically in the lane keeping backup system (LKAS:Lane Keeping Assistance System), can correctly identify lane recognition method and the device thereof of the vehicle in track.

Background technology

Lane keeping backup system (LKAS:Lane Keeping Assistance System) is to utilize the technology of camera identification track and auto-steering, utilize the image of camera to process, measure the vehicle on lane width, the track the lateral attitude, with the track, both sides between distance and track form, road curvature radius, utilize institute's vehicle location that obtains and road information to control vehicle.

Described LKAS performance is the lane information confidence level of obtaining according to by camera, and its control performance is produced a very large impact.But the track on the Ordinary Rd is not solid line, is to be made of dotted line, because railing or central shunting district, railing shadow etc. cause phenomenon unidentified and mistake identification.

Be not only straight line on the actual highway, also have a lot of segment of curve, even straight-line segment, the pavement states such as state and rainy day of tinting according to the track on the road surface, the situation that can not stablize received image signal also happens occasionally.

No. 2010/0076684, Korea S's publication application 10-2009-53412 number and 10-2010-34409 number, United States Patent (USP) 7,532,981, U.S.'s publication application etc. are the conventional arts of relevant identification track method.

According to the filtering technique in the conventional art employing lane recognition method, bring into play more firmly performance during unidentified track, but miss when identifying then relatively relatively weak.

With regard to the conventional art that utilizes image processing techniques for lane identification, need to arrange through paying close attention to the district a plurality of stages of problem and image processing, its operand increases relatively thereupon.

Summary of the invention

Technical task

The present invention creates under described technical background, and its purpose is to provide a kind of and identifies strong vehicle lane recognition methods and the device thereof of still accurately identifying the track under track or the Unidentified situation in mistake.

Another object of the present invention is to provide a kind of under the situation of long-time unidentified and mistake identification, lane recognition method and the device thereof of vehicle that also can steady implementation LKAS control.

Solution

For solving described problem, this technical scheme that adopts is to the image that obtains from camera, to use the weighted value of multiple filter and each wave filter, the mistake identification problem in eliminating track, steady implementation LKAS control.

One aspect of the present invention relates to the lane recognition method of vehicle, and implementation step comprises: to the applicable plural left side of the left-hand lane information local filter that obtains from the vehicle image device; The right-hand lane information that image device from described vehicle is obtained is suitable for plural right side local filter; Calculate the penalty factor of described left side and right-hand lane information; Result by described plural left side local filter filtering is used described penalty factor obtain the left-hand lane estimated value; Result by described plural right side local filter filtering is used described penalty factor obtain the right-hand lane estimated value.

Described left side and right side local filter are to satisfy the Kalman filter of following [mathematical expression 1].

[mathematical expression 1]

$P_K^{-}=F_{K-1}{P}_{K-1}^{+}{F}_{K-1}^T+Q_{K-1}$

$K_K=P_K^{-}{H}_K^T{(H_K{P}_K^{-}{H}_K^T+R_K)}^{-1}$

$P_K^{+}=(1-K_K{H}_K)P_K^{-}{(1-K_K{H}_K)}^T+K_K{R}_K{K}_K^T$

(P is system's covariance, and Q and R are respectively the process noise covariance and measure noise covariance, and K is the kalman gain that utilizes covariance to calculate).

Described penalty factor can calculate according to following [mathematical expression 2].

[mathematical expression 2]

$W\frac{\mathrm{Vt}}{D}+(\mathrm{\α}\×{\mathrm{filter}}_1+(1-\mathrm{\α})\×{\mathrm{filter}}_2)(1-\frac{\mathrm{Vt}}{D})$

(W: minimum lane width

D: maximum unidentified command range

V: speed

T: unidentified/time-out time of mistaking

α: the filter weight value).

The lane identification device of the vehicle that the present invention relates on the other hand is, from the camera reception left side of vehicle and the image information of right-hand lane, obtain the estimated value of described left side and right-hand lane, its composition comprises: the left side local filter, have two at least, the described left-hand lane information of filtering; The right side local filter has two at least, the described right-hand lane information of filtering; The penalty factor piece calculates the penalty factor of described left side and right-hand lane information; The left side senior filter to utilizing the result of described plural left side local filter filtering, is used described penalty factor and is obtained the left-hand lane estimated value; The right side senior filter to utilizing the result of described plural right side local filter filtering, is used described penalty factor, obtains the right-hand lane estimated value.

Beneficial effect

According to the present invention, even because external disturbance only identifies a sidecar road, but adopt lane width information and with the multiple filter technology of interference level, also can improve the lane identification rate.

And the lane recognition method that utilizes multiple filter that relates to of the embodiment of the invention, can tackle forcefully mistake identification, utilize the function of time to reduce lane width, even unidentified/mistake is identified the track for a long time, still steady implementation LKAS control.

Description of drawings

Fig. 1 is the structural drawing of the multiple filter that uses in the lane recognition method of vehicle of the embodiment of the invention.

Fig. 2 is that expression utilizes the lane identification result of existing wave filter and the lane identification result schematic diagram of the embodiment of the invention.

Embodiment

For the purpose, technical scheme and the advantage that make the embodiment of the invention clearer, below in conjunction with the accompanying drawing in the embodiment of the invention, technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that obtains under the creative work prerequisite.The term that uses among the present invention in order to embodiment to be described, is not to limit the invention only.Singulative in this instructions does not have to comprise plural form under the prerequisite of special suggestion in sentence yet." the comprising (comprises) " of using in the instructions or " (comprising) that comprise " do not get rid of the existence of more than one other member, step, action and/or element beyond related member, step, action and/or the element or replenish.

Below in conjunction with accompanying drawing, lane keeping backup system and method thereof to the vehicle of the embodiment of the invention are described in detail.

Fig. 1 is the structural drawing of the multiple filter that uses in the lane recognition method of vehicle of the embodiment of the invention.

As shown in Figure 1, the multiple filter 100 that uses in the lane recognition method of the vehicle of the embodiment of the invention is that the right-side signal 120 of the left-side signal 110 of camera and camera is used respectively two local filter.In other words, camera left-side signal 110 is used local filter L1112 and local filter L2114, camera left-side signal 120 is used local filter R1122 and local filter R2124.Used wave filter only limits and uses two, can regulate according to the control level.The embodiment of the invention is for the facility in the explanation, is defined as two.

Each local filter is become by the group of Kalman filters such as following [mathematical expression 1] expression.

[mathematical expression 1]

$P_K^{-}=F_{K-1}{P}_{K-1}^{+}{F}_{K-1}^T+Q_{K-1}$

$K_K=P_K^{-}{H}_K^T{(H_K{P}_K^{-}{H}_K^T+R_K)}^{-1}$

$P_K^{+}=(1-K_K{H}_K)P_K^{-}{(1-K_K{H}_K)}^T+K_K{R}_K{K}_K^T$

(P is system's covariance, and Q and R represent respectively the process noise covariance and measure noise covariance that K is the kalman gain that utilizes covariance to calculate)

Signal is sent to left side senior filter 116 and right side senior filter 126 through each local filter.

After obtaining according to the signal difference variation that enters by camera and the weighted value of time, penalty factor piece 130 sends each senior filter 116,126 to.

Obtain weighted value according to following [mathematical expression 2] in the penalty factor piece.

[mathematical expression 2]

$W\frac{\mathrm{Vt}}{D}+(\mathrm{\α}\×{\mathrm{filter}}_1+(1-\mathrm{\α})\×{\mathrm{filter}}_2)(1-\frac{\mathrm{Vt}}{D})$

W: minimum lane width

D: maximum unidentified command range

V: speed

T: unidentified/mistake is identified time-out time

α: filtration combined weighted value

Then utilize the value obtained from local filter and from the weighted value with signal the change of divergence and time that the penalty factor piece obtains, obtain final track estimated value at senior filter 116,126.

Fig. 2 is that expression utilizes the lane identification result of existing wave filter and the lane identification result of the embodiment of the invention.The upper plot of Fig. 2 is existing wave filter, and the lower partial graph table of Fig. 2 is the lane identification result who represents respectively to utilize the embodiment of the invention, red expression right-hand lane, green expression left-hand lane.

No. 1 is the example of the own change lane of expression driver among Fig. 2, and No. 2 is the example that mistake is identified left-hand lane.

According to the lane identification example (upper No. 2) that adopts existing wave filter technology, directly infer left-hand lane, can't detect track mistake identification, but adopt the lane recognition method (lower No. 2) of the embodiment of the invention, then identify under the situation in track in mistake, the track estimated value can not followed mistake yet and be identified the track, but forcefully reply.

As mentioned above, according to the present invention, because of external disturbance, when only identifying a sidecar road, also can utilize lane width information and with the multiple filter technology of noise level, improve the lane identification rate.

The lane recognition method that utilizes various filters in the embodiment of the invention is the mistake identification situation that can tackle forcefully the track, and unidentified/mistake is identified the track for a long time even can utilize the function of time to reduce lane width, still steady implementation LKAS control.

Above embodiment and particular terms only in order to technical scheme of the present invention to be described, are not intended to limit; Although with reference to previous embodiment the present invention is had been described in detail, those of ordinary skill in the art is to be understood that: it still can be made amendment to the described technical scheme of aforementioned each embodiment, perhaps part technical characterictic wherein is equal to replacement; And these modifications or replacement do not make the essence of appropriate technical solution break away from the scope of the described technical scheme of various embodiments of the present invention.

Claims (6)

1. the lane recognition method of a vehicle, implementation step comprises:

To the applicable plural left side of the left-hand lane information local filter that obtains from the vehicle image device;

The right-hand lane information that image device from described vehicle is obtained is suitable for plural right side local filter;

Calculate the penalty factor of described left side and right-hand lane information;

Result by described plural left side local filter filtering is used described penalty factor obtain the left-hand lane estimated value;

Result by described plural right side local filter filtering is used described penalty factor obtain the right-hand lane estimated value.

2. the lane recognition method of vehicle according to claim 1 is characterized in that, described left side and right side local filter are to satisfy the Kalman filter of following [mathematical expression 1],

[mathematical expression 1]

$P_K^{-}=F_{K-1}{P}_{K-1}^{+}{F}_{K-1}^T+Q_{K-1}$

$K_K=P_K^{-}{H}_K^T{(H_K{P}_K^{-}{H}_K^T+R_K)}^{-1}$

$P_K^{+}=(1-K_K{H}_K)P_K^{-}{(1-K_K{H}_K)}^T+K_K{R}_K{K}_K^T$

P is system's covariance, and Q and R are respectively the process noise covariance and measure noise covariance, and K is the kalman gain that utilizes covariance to calculate.

3. the lane recognition method of vehicle according to claim 1 is characterized in that,

Described penalty factor is to calculate according to following [mathematical expression 2],

[mathematical expression 2]

$W\frac{\mathrm{Vt}}{D}+(\mathrm{\α}\×{\mathrm{filter}}_1+(1-\mathrm{\α})\×{\mathrm{filter}}_2)(1-\frac{\mathrm{Vt}}{D})$

W: minimum lane width

D: maximum unidentified command range

V: speed

T: unidentified/time-out time of mistaking

α: filter weight value.

4. the lane identification device of a vehicle is characterized in that, from the camera reception left side of vehicle and the image information of right-hand lane, obtains the estimated value of described left side and right-hand lane, and its composition comprises:

The left side local filter has two at least, the described left-hand lane information of filtering;

The right side local filter has two at least, the described right-hand lane information of filtering;

The penalty factor piece calculates the penalty factor of described left side and right-hand lane information;

The left side senior filter to utilizing the result of described plural left side local filter filtering, is used described penalty factor and is obtained the left-hand lane estimated value;

The right side senior filter to utilizing the result of described plural right side local filter filtering, is used described penalty factor, obtains the right-hand lane estimated value.

5. the lane identification device of vehicle according to claim 4 is characterized in that, described left side and right side local filter are the Kalman filter that meets following [mathematical expression 1],

[mathematical expression 1]

$P_K^{-}=F_{K-1}{P}_{K-1}^{+}{F}_{K-1}^T+Q_{K-1}$

$K_K=P_K^{-}{H}_K^T{(H_K{P}_K^{-}{H}_K^T+R_K)}^{-1}$

$P_K^{+}=(1-K_K{H}_K)P_K^{-}{(1-K_K{H}_K)}^T+K_K{R}_K{K}_K^T$

P is system's covariance, and Q and R are respectively the process noise covariance and measure noise covariance, and K is the kalman gain that calculates by covariance.

6. the lane identification device of vehicle according to claim 4 is characterized in that, described penalty factor is to calculate according to following [mathematical expression 2],

[mathematical expression 2]

$W\frac{\mathrm{Vt}}{D}+(\mathrm{\α}\×{\mathrm{filter}}_1+(1-\mathrm{\α})\×{\mathrm{filter}}_2)(1-\frac{\mathrm{Vt}}{D})$

W: minimum lane width

D: maximum unidentified command range

V: speed

T: unidentified/mistake is identified time-out time

α: filter weight value.

Images