0 оценок0% нашли этот документ полезным (0 голосов)
11 просмотров12 страниц
A method is provided for detecting road lane markers in a vehicle road using an imaging device. Road input data is captured using the imaging device. Lighting normalization is applied to the road input data. Candidate lane markers are selected in response to the edgebased filter response and line-based filter response in each main orientation.
A method is provided for detecting road lane markers in a vehicle road using an imaging device. Road input data is captured using the imaging device. Lighting normalization is applied to the road input data. Candidate lane markers are selected in response to the edgebased filter response and line-based filter response in each main orientation.
Авторское право:
Attribution Non-Commercial (BY-NC)
Доступные форматы
Скачайте в формате PDF, TXT или читайте онлайн в Scribd
A method is provided for detecting road lane markers in a vehicle road using an imaging device. Road input data is captured using the imaging device. Lighting normalization is applied to the road input data. Candidate lane markers are selected in response to the edgebased filter response and line-based filter response in each main orientation.
Авторское право:
Attribution Non-Commercial (BY-NC)
Доступные форматы
Скачайте в формате PDF, TXT или читайте онлайн в Scribd
1
cu») United States
c2) Patent Application Publica
Zhang et al.
(a) CAMERA-BAS
DETECTION
Wende
(73) Inventors
MI(US); Varsha Sadekar, String
Heights, MI (
CComespondenee Address
43604 (US)
GM GLOBAL TECHNOLOGY
OPERATIONS, INC., Detroit, MI
ws)
(03) Assignee:
(21) Appl.Nos — 120175,631
(22) Filed Jul 18,2008,
Publication Classification
(1) neck
G6K 9/78 (2006.01)
G06K 9700 (2006.01),
US 20100014714.
(10) Pub. No.: US 2010/0014714 Al
(43) Pub, Date Jan. 21, 2010
Gosp 102
GOOF 1700
GOK 946
200501)
(2006.01),
(200501),
(2) US.CL ‘382/104; 382/190; 701/28,
on ABSTRACT
A method is provided for devcting road lane markers in a
vehicle road using an imaging device. Road input data is
captured using the imaging device. Lighting normalization is
‘applied to the road inpnt data, The method detec the road
Jane markers in a few main orientations in the normalized
input data In each main orientation, the normalized input
data is convolved with an oriented edge detection filter for
‘genersting an oriented edge-based filter response. The nor
‘malized input data is convolved with an oriented line detee-
tion filter forgeneratingan oriented line-ased filter response
Candidate lane markers are selected in response tothe edge-
based filter response and line-based filter response in each
‘main orientation. transformation techaigue is applied tothe
candidate lane markers for identifying the lane markings in
cach main orientation,Patent Application Publication Jan, 21, 2010 Sheet 1 of 7 US 2010/0014714 AI
OUTPUT DEVICE a
12 t
\ Cf roao sense”, —Faocesson 18
DEVICE t
MEMORY Wy
HG. 1
a=0
FALSE
LINE HOUGH
LI bo} ALARM
20 DETECTION ["] TRANSFORM |" ntiGaTiON
CAPTURE eae FALSE
IMAGE UNE J.J HOUGH [| Alaa
=< DETECTION [| TRANSFORM || satiGaTiOn DETECTED
LANE
° . .
=" : MARKERS
NORMALIZATION :
.
v e : 25
a , : *
FALSE
LINE HOUGH
LI be} ALARM
DETECTION [| TRANSFORM | sairiGation
a= 150 ‘
2 23 24
FIG. 2Patent Application Publication Jan. 21,2010 Sheet 2 of 7 ‘US 2010/0014714 AL
LINE DETECTION FILTERS:
a=0 30
60 90 120 150
ORIENTED FILTERS
FIG. 4Patent Application Publication
Jan. 21, 2010 Sheet 3 of 7
US 2010/0014714 AI
jvewarner
Z| L
oo pLOlIvity
FILTER RESPONSE
FIG. 5A FILTER (UNE)
INTENSITY INTENSITY \
+ ; * RESPONSE ;,
i 5
; 5
i Jt
> > 4 >
CENTER CENTER
FIG. 5B
LANE-MARKER |
<———' &)
EDGE veTEcTION ==
FIG. 6A ites (EDGE)
'
|NTENSITY INTENSITY 7
' & '
; . CENTER
CENTER
FIG. 6BPatent Application Publication Jan, 21,2010 Sheet 4of7 US 2010/0014714 AL
\
INTENSITY ee i
a ———— CENTER
INTENSITY f
t RESPONSE ;
on 1
JS
> + >
CENTER
FIG. 7A
| LANE-MARKER &}
LARGE NEGATIVE
RESPONSE
[Ze
A |=
EDGE DETECTION SO LTER RESPONSE
FILTER (EDGE)
FIG. 7B
g
LINE DETECTION
FILTER
FILTER RESPONSE
(LINE)Patent Application Publication Jan, 21, 2010 Sheet 5 of 7 US 2010/0014714 AI
LINE DETECTION
FILTER
FILTER RESPONSE
(LINE)
FIG. 8
FIG. 9Patent Application Publication Jan, 21, 2010 Sheet 6 of 7 US 2010/0014714 AI
CAPTURING ROAD INPUT DATA }-~ 90
1
APPLYING LIGHTING NORMALIZATION TO THE INPUT DATA }~31
M
CONVOLVING THE NORMALIZED INPUT DATA WITH THE 32
‘ORIENTED EDGE DETECTION FILTERS
M
CONVOLVING THE NORMALIZED INPUT DATA WITHTHE |_—33
ORIENTED LINE DETECTION FILTERS
1
SELECTING CANDIDATE LINE MARKERS FROMTHE |_-~34
FILTERED RESPONSES
1
APPLYING THE HOUGH TRANSFORMATION TECHNIQUE _}-~25
APPLYING A FALSE ALARM MITIGATION ANALYSIS. |-36
IS THE
FALSE ALARM MITIGATION
38 ANALYSIS SATISFIED?
\
IDENTIFYING THE LOCATION
OF THE MARKERS IN AN 39
OUTPUT DEVICE
FIG. 10Patent Application Publication Jan, 21, 2010 Sheet 7 of 7 US 2010/0014714 AI
DETERMINING A LENGTH (1) OF THE IDENTIFIED |_—50
LANE MARKERS
Y
COMPARING THE LENGTH (2) TO A PREDETERMINED.
LENGTH
LENGTH (2) > PREDETERMINED
HIGHLIGHTING THE LANE
MARKERS IN THE IMAGE 39
DISPLAY DEVICE
FIG. 11
DETERMINING A DISTANCE (d) BETWEEN A PAIR OF |_—60
PARALLEL LANE MARKERS
Y
61
COMPARING THE DISTANCE (c) TO A }~
PREDETERMINED WIDTH
HIGHLIGHTING THE LANE
MARKERS IN THE IMAGE 39
DISPLAY DEVICE
FIG. 12US 2010/0014714 AI
CAMERA-BASED LANE MARKER
DETECTION
BACKGROUND OF INVENTION
[0001] An embodiment relates generally t0 lane marker
detection of a road using a image-ased capture device.
[0002] Camers-based lane marker detection systems are
used to detect roads or road sepments ofa vehicle oad. Most
systems work reasonably wel in highway seenatos and can
not handle the complex environmen ia local scenarios. Suck
‘camers-based systems are ssceptible to incoretly distin=
[Bishing rad objects scl as carbs and lane markers fom
shadows of trees, buildings, and other environmental condi
tions. Furthermore, camera-based systems are typically chal-
Jenged by certain conditions such as sharp curves inthe road
‘or the weatherilow sun angle. Since curves in roads are more
«dificult to detect in comparison to straight Fines, such sys-
tems have the disadvantage of running slower processing
times wo accurately detet the lane markers.
SUMMARY OF INVENTION,
10003} An advantage ofthe invention provides for a tw
‘ost and reliable detection of lane markers in a oad using an
Sage caparedevie. The invention provides forthe se of
Jow-costimage capture device soch a camera for detecting
Jane markers relialy along curved reads or straight roads
“The lane markers ofthe sharp curves are distinguished ra
strcting factors sich os variooe shadows from buildings
and eosin ition to poorly painted lane markers
0004} An embodiment contemplates a method of dtet-
dng oad lane markers ia a vehicle rad sng a imaging
device. Roa input image daa is captured using the imaging
‘device. Lighting sonmaliation is applied to the road it
image data. The metho detects the road lane markers ina few
‘ain oriotations in the input image. hn each min orienta
tion, the normalized input data is convolved with an oriented
‘edge detection filter for generating sn oriented edge based
filer response, The normalized input deta isconveved with
‘an ovcnted Tine detection filer for generating an oriented
lineshsed filter response. Candidate lane markers ate
selected in response to the edge-based filter response and
lineshase filter response inthe neighboring angles of each
‘ain orientation A tsformation technique i applied tthe
‘eanddat lane markers for identifying the lane markings in
the neighboring angles ofeach main orientation
{0005} As cmbovinent contemplates lane marker detec
Sionsystem. An imaging devices provided foreaptring road
nut data. A processor receives the captred rod ip data
rovcived by the imaging devi. The processor apples High
ing normalization t the oad input data. The processor pr
‘esses the normalized input data na few main orientations. 1a
‘ech main orientation, the processor comvlves the normal
ized input data with an oriented edge deteton filter for
penerting an oriented ege-basa filer response. The poe
‘cessor convoles the noma input data with an oriented
Tine detection iter fr generating an oriented line based fier
response. The processor selects candidate lane markers fom
the edge-bosed filter response and the Tine-based ter
response. In cach man orenation, the processor applies
transformation eelnigue tothe selected candidate lane mark-
Jan. 21, 2010
ers for identifying the line segments of lane markings. An
‘uit device is provided for identifying location of each of
the fane marker,
BRIEP DESCRIPTION OF DRAWINGS
[0006] FIG. 1 isa schematic view of an image-based lane
‘marker detection system,
[007] FIG. 2 isa block diagram of a method for detecting
Jane markers ina oad,
0008] FIG. 3 is display of captured image,
[0009] FIG. 4 is an image representation of orientation
Setecton fiers,
[010] FIG. Sa is an image representation of an oriented
line filtering process
[0011] FIG. 8bisa graphical representation ofthe oriented
line filtering process.
[012] FIG. 6a is an image representation of an oriented
cdg filtering process
[013] FIG. 66s a graphical representation ofthe oriented
cede fering process
[0014] FIG. 7a isa praphical representation of a combined
‘oriented edge and lie filtering process.
[0015] FIG. 7b isan image represen
‘oriented edge and lie filtering process.
[0016] FIG. 8 is an imaye representation ofa false alarm
mitigation process
[0017] FIG. 9isa display highlighted lane markers in the
‘captured image
[0018] FIG. 10isa flowchart of a method for detecting ane
‘markers in the road
[0019] FIG. 11 isa flowehat ofa method fora false alarm
‘itgation analysis according ta frst embodiment
[0020] FIG. 12 isa fowehurt ofa method fora false alarm
‘mitigation analysis aeeording toa second embodiment.
12 of @ combines
DETAILED DESCRIPTION
[0021] There is shown in FIG. 1 a lane marker detection
system 10 for detecting lane markers in a road of the driven
vehicle. The lane marker detection system 10 may be used in
‘an autonomous steering system for rod detection or the lane
‘marker detection system 10 may be used for driver awareness
{oravehicle driven by a driverto enhance visibility during the
daytime or nighttime when road or lane marker visibility is
oor orduring other instances when visibility enhancement is
‘wareante.
[0022] The lane marker detection system 10 includes an
mage cupture device 12 including, bu not limited, acam-
cera. The image capture device 12 captures an image of the
oa typically thearea drectedin the frontof the vehicle. The
captured image is processed for identifying both edges of a
Jane marker as well as @ line (i.e, main body) of the lane
marker
[0023] The lane marker dewetion system 10. further
includes a processor 16 for receiving and processing the
captured image data by the image capture device 12. A
‘memory storage device 18 may also be provided for storing
And retieving the captured data
0024} ‘The processor 16 exceutes a program that filters the
‘captured image data in realtime for determining the presence
fant location ofone or more lane markers inthe vehicle roa.
‘The detected lane markersare provided to an output device 14
stich as an aslonomots steering module or an image display
device. The autonomous steering module may use the pro-US 2010/0014714 AI
‘cess information for autonomously maintaining vebicle
Postion within the road betwoon the detected lane markers
‘The image display device which may inlode, but is 01
limite, nontorsype displys, a projevton-ype imaging
holograp