`
`U.S. PATENT NO. 4,970,653, ISSUED TO KENUE ON NOV. 13, 1990
`
`("KENUE")
`
`TRW Automotive U.S. LLC: EXHIBIT 1105
`PETITION FOR INTER PARTES REVIEW
`OF U.S. PATENT NUMBER 8,599,001
`
`
`
`United States Patent [191
`Kenue
`
`[11] Patent Number:
`[45] Date of Patent:
`
`4,970,653
`Nov. 13, 1990
`
`[54] VISION METHOD OF DETECTING LANE
`BOUNDARIES AND OBSTACLES
`[75] Inventor: Surender K. Kenue, South?eld,M1ch.
`[73] Assignee: General Motors Corporation, Detroit,
`Mich-
`
`4,868,752 9/1989 Fujii et a1. .................... .. 364/424.()2
`OTHER PUBLICATIONS
`Dickmanns, E_ D and Zapp’ A’, “A Curvature_BaSed
`Scheme for Improving Road Vehicle Guidance by
`Computer Vision,” Proc. SPIE on Mobile Robots, vol.
`
`[22] Filed;
`[51] I t Cl 5
`
`Aim 6, 1989
`
`15/50
`
`Primary Examiner-(Sary Chin
`Attorney, Agent, or Fzrm-—Howard N. Conkey
`
`n .
`
`.
`
`............................................ .. G06F
`
`[56]
`
`[52] US. Cl. ............................. .. 364/461; 364/424.02;
`_
`358/103
`[58] Flew of Search
`364é/4246'o_234g1’5g;4'1(;;
`"""
`80/1 7’ 1 8’ 1 9’
`1 /
`’
`References Cited
`U s PATENT DOCUMENTS
`l
`'
`_
`ggifitilg 2' """""" " 3641/
`4:8l9:169 4/1989 Saitoh et a1. ......... ..
`. 364/424.02
`4,847,774- 7/1989 Tomikawa et al.
`. 364/424.02
`4,862,047 8/1989 Suzuki et a1.
`318/587
`
`.
`
`ABSTRACT
`[57]
`An image processing method operates on an image from
`a ccd camera viewing a roadway scene in front of a
`vehicle to detect lane markers and determine the rela
`tionship of the vehicle to the lane. Obstacles in the lane
`near the vehicle are detected and a warning is given to
`the driver. The method uses template matching tech
`niques or a Hough algorithm to detect the lane markers
`or road edges‘
`
`10 Claims, 7 Drawing Sheets
`
`COUNT eoce PIXELS
`w m LANE. BOUNDARlES
`AND > so r1:
`
`OBSTACLE
`WARNING
`
`DELETE VERT. AND HOR.
`roses AT msrmce
`> 50 FT.
`
`DEFINE SEARCH AREA
`
`PREVIOUS FRAME
`
`IN SEARCH AREA
`CORRELATE A 5+5
`WINDOW WITH IMAGE TO
`IDENTIFY MARKERS ~~5a
`
`CALCULATE CENTROID l
`OF EACH‘ MARKER
`
`CALCULATE MARKER
`POSITION BASED ON
`OPPOSITE MARKER
`AND LANE WlDTH
`___.__.__._l
`
`\
`
`68
`
`sumclgm-
`MATCH fom'rs
`
`NO
`
`REPEAT CORRELATTON
`WITH EDGE DATA
`
`6'2
`
`USE LEAST SQUARE
`FIT TO GENERATE
`LANE BOUNDARV
`
`USE LANE BOUNDARIES
`FROM PREVIOUS FRAME
`
`CHECK LANE EOUNDARIS
`FOR CONSISTENO‘ OF
`ANGLE AND WIDTH
`
`IS
`ENOLOVEER a
`0 LAN
`BOUNDARV NEAR
`MIDDLE OF
`IMAGE. z
`
`NO
`
`BTIMATE LOOK
`AHEAD PATH
`
`\
`\
`
`PREDICT
`NEW
`BOUNDARIES
`
`OVERLAY LANE
`BOUNDARIES
`
`ON (MAGE @
`
`1105-001
`
`
`
`US». Patent Nov. 13, 1990
`
`Sheet 1 of7
`
`4,970,653
`
`CESEDRA
`
`/
`
`/0/
`j‘
`
`'
`
`A D
`1
`
`/2
`
`4‘
`//
`
`'76
`/
`
`DlSPLAY
`
`COMPUTER
`
`ALARM
`
`2
`
`UTIL'ZATION
`CIRCUIT
`
`"8
`/
`29
`
`1105-002
`
`
`
`US, Patent Nov. 13, 1990
`
`Sheet 2 of 7
`
`4,970,653
`
`DIGI'I'IZATION
`
`I w
`#2
`
`CONTRAST EXPANSION
`
`EDGE IMAGE ./W
`SOBEI. FILTER
`
`4%
`
`COUNT EDGE PIXELS
`WIN LANE BOUNDARIES
`AND > 50 FT
`
`j'yféz
`
`OBSTACLE
`WARNING
`
`YES
`
`NO
`
`DELETE VERT. AND HOR.
`EDGES AT DISTANCE
`~42
`> so FT.
`I
`DEFINE SEARCH AREA
`FOR EACH MARKER
`BASED ON MARKER
`POSITION MP IN
`PREVIOUS FRAME
`
`IN SEARCH AREA
`CORRELATE A 5+5
`WINDOW WITH IMAGE TO
`IDENTIFY MARKERS '-56
`
`SUF FICIENT
`MATCH E’OINTS
`
`‘REPEAT CORRELATION‘
`WITH EDGE DATA
`
`1105-003
`
`
`
`U.SI Patent Nov. 13, 1990
`O
`
`Sheet 3 of 7
`
`4,970,653
`
`/
`I
`CALCULATE CENTROID 1
`OF EACH MARKER
`
`SIDE OF LANE
`I KNOWN
`
`'
`
`?
`
`CALCULATE MARKER
`POSITION BASED ON
`OPPOSITE MARKER
`AND LANE WIDTH
`____—____I
`USE MARKERS FROM
`PREVIOUS FRAME
`
`I
`(
`68
`
`ARE
`MARKERS
`FOUND WHERE
`
`EXPEC7TED
`
`YES
`
`_E_____I
`
`71"
`
`I
`USE I_ANE BOUNDARIES
`FROM PREVIOUS FRAME
`
`I
`79
`
`USE LEAST SQUARES
`FIT TO GENERATE
`LANE BOUNDARY
`
`'~’76
`
`CHECK LANE BOUNDARIES
`FOR CONSISTENCY OF
`ANGLES AND WIDTH
`
`"-175
`
`60
`
`YES‘
`
`5;
`
`INDICATE /
`LANE
`CHANGE
`
`84‘
`I
`
`PRED'CT
`NEW
`BOUNDARIES
`
`OVER LAY LANE
`BOUNDARIES
`ON IMAGE
`
`I START
`
`I
`
`\
`
`\
`
`IS
`LOWER
`END OF LANE
`BOUNDARY NEAR
`MIDDLE OF
`IMAGE _?
`
`NO
`
`LOOK“
`ESTIMATE
`AHEAD PATH
`
`5g, 4%
`
`1105-004
`
`
`
`S US“ Patent Nov‘, 13, 1990
`
`Sheet 4 of 7
`
`(ENTER -)
`
`4,970,653
`
`54%
`
`INCREMENT=
`
`R LANE COORDINATE-L LANE COORDINATE
`5
`
`WAS
`EITHER
`LANE MARKER
`NOT FOUND OR AN
`OLD MARKER USED IN
`PREVIOUS
`IMAGE
`
`5446
`
`INCREASE INCREMENT BY 20
`(LIMIT INCREMENT TO 128)
`
`DEFINE BOUNDS BR AND BL
`FOR EACH SEARCH AREA
`BL=MP- INCREMENT
`BR MP+INCREMENT
`
`CORRELATE
`IMAGE
`BL AND BR
`BETWEEN
`
`56
`
`lNDEX TO
`ANOTHER
`ROW
`
`{I
`55’
`
`ARE
`~
`ALL Rows
`
`57
`
`1105-005
`
`
`
`US. Patent N0v.13,1990
`
`Sheet 5 of7
`
`4,970,653
`
`( START -)
`
`DIGITIZATION /
`;
`CONTRAST EXPANSION
`
`//00
`M2
`
`EDGE DETECTION ///04
`
`
`
`THRESHOLDING Y
`
`NOISE CLEANING
`
`isxamm /--~/06’
`
`COUNT EDGE PIXELS
`W/IN LANE BOUNDARIES
`AND < 50 FT.
`
`‘ OBSTACLE '
`WARNING
`
`(TO START )
`
`LINE HOUGH TRANS F'O RM
`
`7
`
`DEFlNE SEARCH AREA
`IN
`HOUGH SPACE BASED ON
`PREVIOUS MARKER
`POSITION
`
`1105-006
`
`
`
`US” Patent Nov. 13,1990
`
`Sheet 6 of7
`
`4,970,653
`
`‘---@ w
`
`DELETE PIXELS < 20 "4
`OF MAXIMUM INTENSITY L
`
`I
`NoN- MAXIMUM
`SUPP RESS ION FILTER
`
`l/ep
`
`DELETE PIXELS _//Z2
`REPRESENTING vERT.
`AND HORIZ. LINES
`
`REMovE PIXELS
`
`’
`
`cLosER THAN
`
`sET AMoUNT
`
`SELECT PIXEL WITH
`MOST CONTRIBUTING
`EDGE ELEMENTS
`
`ARE
`LANE
`BOUNDARY
`PARAMETERS WITHIN
`10% OF PREVIOUS
`PARAMETERS
`
`YES
`
`‘
`
`g 65
`
`A
`BOUNDARY
`PARAMETERS
`NO
`
`YES
`
`/
`
`UsE AVERAGE OF /
`LAsT TWO FRAMES
`PARAMETERS
`I
`
`DETECT LANE {/g?
`CHANGE
`
`U6
`
`ESTIMATE LOOK
`AHEAD
`TH
`PA
`+
`INVERSE HOUGH
`TRA NsFoRM
`
`w
`OVERLAY LANE
`N
`mum/[1325255 ON
`
`w
`
`1105-007
`
`
`
`IU.SI Patent N0v..13, 1990
`
`Sheet 7 01'7
`
`4,970,653
`
`HA
`
`\
`
`R
`
`A
`
`STRAIGHT LINE IN
`IMAGE DOMAIN
`
`\ x
`
`AII
`
`180° -
`
`90° --
`
`/32
`
`RT L__J
`
`o
`
`STRAIGHT LINE IN
`HOUGH DOMAIN
`
`r
`R
`
`J7 1"
`
`1105-008
`
`
`
`VISION METHOD OF DETECTING LANE
`BOUNDARIES AND OBSTACLES
`
`FIELD OF THE INVENTION
`This invention relates to a vision method of detecting
`lane boundaries and obstacles close to a vehicle within
`the lane boundaries, and particularly to such a method
`employing image processing techniques and which is
`operative for moderately marked roads.
`
`25
`
`BACKGROUND OF THE INVENTION
`The use of an on-board video camera and image pro
`cessing of the roadway scenes allows useful information
`to be gathered for vehicle control. Detecting lane
`boundaries is a core capability for advanced automotive
`functions such as collision warning, collision avoidance
`and automatic vehicle guidance. If the lane boundaries
`and thus the road path can be detected several other
`functions can be implemented. Lane control uses the
`boundary information and vehicle dynamics knowledge
`to derive steering and braking commands for keeping
`the vehicle in the lane. Headway control uses a laser or
`radar system to track the vehicle ahead and keeps a safe
`driving distance. The lane boundary information can be
`used to prevent the detection of a vehicle in an adjacent
`lane on a curved road. Then the sensor beam can be
`directed to points within the lane. To monitor driving
`performance, the behavior of the driver is tracked and
`evaluated using the estimated position of the vehicle
`with respect to the lane boundaries.
`.
`Lane boundary detection for guiding vehicles along
`roadways has been reported in the paper by Dickmanns,
`E. D. and Zapp, A , “A Curvature-based Scheme for
`Improving Road Vehicle Guidance by Computer Vi
`sion,” Proc. SPIE on Mobile Robots, Vol. 727, October
`1986, which is incorporated herein by reference. Con
`tour correlation and high order world models are the
`basic elements of that method, realized on a special
`multi-processor computer system. Perspective projec
`tion and dynamical models (Kalman ?lter) are used in
`an integrated approach for the design of the visual feed
`back control system. That system requires good lane
`markings and thus is limited to only those roads having
`good lane markings. It is of course desirable to extend
`the bene?ts of the computer vision system to roads with
`less good markings and to incorporate other features
`such as obstacle detection.
`
`SUMMARY OF THE INVENTION
`It is therefore an object of the invention to provide a
`computer vision method of lane boundary detection
`operable on roads having moderate markings and even
`with some markers missing. It is a further object to
`provide such a method with obstacle detection capabil
`ity.
`The invention is carried out in an automotive vehicle
`having a computer vision system and an associated
`camera for viewing the scene ahead of the vehicle, by a
`method of detecting lane markers on a roadway com
`prising the steps of: obtaining an image of the scene and
`digitizing the image, normalizing the image, de?ning a
`search area in the image, searching for lane markers in
`65
`the search area of the image, estimating the position of
`any missing lane marker, and de?ning lane boundaries
`from the position of the lane markers.
`
`60
`
`1
`
`4,970,653
`
`2
`BRIEF DESCRIPTION OF THE DRAWINGS
`The above and other advantages of the invention will
`become more apparent from the following description
`taken in conjunction with the accompanying drawings
`wherein like references refer to like parts and wherein:
`FIG. 1 is a block diagram of a computer vision system
`for carrying out the method of the invention,
`FIG. 2 is an illustration of a roadway scene in a cam
`era image plane,
`FIG. 3 is an illustration of a roadway scene in a cam
`era image plane, the scene having an obstacle in the
`roadway,
`FIGS. 4a and 4b together comprise a ?ow chart of an
`embodiment of the method of the invention employing
`template matching,
`FIG. 5 is a ?ow chart of the dynamic search area
`de?nition method used in the FIG. 4 method.
`FIGS. 60 and 6b together comprise a ?ow chart of
`another embodiment of the method of the invention
`employing a Hough transform, and
`FIGS. 7a and 7b are an image plane view of a line and
`a Hough space representation of the same line, respec
`tively.
`DESCRIPTION OF THE PREFERRED
`EMBODIMENT
`As shown in FIG. 1, the hardware used to implement
`the method of the invention comprises a black and
`white CCD video camera 10 mounted in a vehicle, say
`at the upper center of the windshield to capture the
`driver’s view of the road ahead, an analog-to-digital
`converter 12 for coupling the camera output to a com
`puter 14, and output devices driven by the computer
`including a display 16, an obstacle warning alarm 18 and
`a utilization circuit 20 which may be any device using
`the lane boundary information for vehicle guidance,
`performance monitoring or headway control, for exam
`ple.
`The computer is programmed with algorithms for
`processing the images sensed by the camera. Two main
`algorithms for processing the image are disclosed
`herein. One uses a Hough transform and the other uses
`template matching. Both algorithms, however, dynami
`cally de?ne the search area for lane markers based on
`the lane boundaries of the previous frame, and provide
`estimates of the position of missing markers on the basis
`of current frame and previous frame information.Also
`in both cases preprocessing procedures detect obstacles
`in the lane within about 50 feet of the vehicle and give
`a warning via the alarm 18.
`FIG. 2 is an example of a typical image as projected
`on the camera image plane and includes the vehicle
`hood 22 and broken or solid stripes or lane markers 24
`painted on the road and terminating at the horizon 26.
`Since the image spaces above the horizon and below the
`hood line do not contain useful lane boundary informa
`tion those regions are ignored by the image processing
`algorithms. A special calibration procedure enables
`range estimation, based on known characteristics of the
`road and the camera. It assumes flat roads with parallel
`markers at known look-ahead distances. The calibration
`determines the relationships between the positions of
`the actual markers and the images of these markers.
`This process de?nes a set of “reference markers” which
`are used as an initial set of lane boundaries seen by the
`vehicle when it is in the middle of the lane.The refer
`ence markers are also used when consistency of width is
`
`1105-009
`
`
`
`4,970,653
`4
`3
`not updated since the presence of the obstacle obscures
`violated or when the two estimated lane boundaries
`them. The obstacle distance is determined by the
`cross each other.
`ground level obstacle image since the image plane cali
`The broken line boxes 28 around the markers 24 de
`bration does not take into account the vertical height of
`?ne the area to be searched for detecting markers and is
`determined by the position of the lane boundaries in the
`an object. As seen in FIG. 3, the top of the vehicle 30
`previous image frame. In reality many other objects are
`appears to be beyond the horizon as seen in the image
`plane although the vehicle 30 is actually close as is
`in the image such as trees, other vehicles, overpasses
`realistically shown near ground level. Thus for obstacle
`and signs. The de?ned search area helps eliminate much
`detection, only the bottom of the vehicle 30 image is
`of them from the processed image as well as to minimize
`scanned since it is in the near portion of the image plane.
`the processing time. FIG. 3 illustrates the presence of a
`Obstacles detected at distances greater than 50 feet
`vehicle 30 or other obstacle in the roadway of FIG. 2.
`(in the far ?eld portion of the image) are effectively
`If the other vehicle is close, say, within 50 feet of the
`deleted from both the image and the edge data <52>
`trailing vehicle, it tends to obscure the lane markers 24
`to facilitate the execution of the following steps. For
`to such an extent that there is insuf?cient information to
`this purpose it is assumed that the obstacle has a high
`determine the boundaries. In that case an obstacle warn
`ing is given and no further image processing is done on
`content of vertical and horizontal lines whereas the lane
`markers to be detected comprise diagonal lines. The
`that frame. When the obstacle 30 is more than 50 feet
`away the image is processed but the obstacle is effec
`obstacles are deleted from the image using the follow
`tively erased from the image by removing horizontal
`ing criteria: lines to be deleted must be either vertical or
`and vertical lines, thereby making the subsequent pro
`horizontal, their length must exceed 40 pixels, and their
`cessing steps simpler.
`intensity must exceed 150.
`The template matching algorithm is widely used in
`After the obstacle deletion step the search area is
`image recognition and computer vision applications. In
`de?ned dynamically <54>. Given the previous lane
`boundaries, a truncated triangular search region is de
`this algorithm, a template or window of desired inten
`sity and shape is correlated with the image to create a
`?ned for each marker location such that the search
`region for the two sides of the lane do not overlap.
`correlation matrix. The elements of this correlation
`matrix indicate the quality of the match between the
`Moreover, the search area changes with marker posi
`template and the image at all locations, according to
`tion MP, and increases in size if the marker was not
`found in the search of the previous frame. This limits
`some metric. Here, the absolute difference metric is
`the search space in which the markers can move from
`used as the correlation measure. Other types of mea
`one frame to another and shortens the processing time.
`sures such as normalized correlation can also be used.
`The dynamic search method is shown separately in
`The matching can also be done in the image-edge do
`main obtained by ?ltering the raw image with a Sobel
`FIG. 5. The search is conducted one row at a time and
`?lter. The advantage of edge matching is its insensitiv
`the search ?eld is selected for each row. It begins by
`ity to light variations; however, with this method, pot
`de?ning an increment for a given row as one ?fth of the
`distance between the right and left lane coordinates for
`holes, streaks, road texture and shadows may generate
`false objects. The template matching algorithm de
`that row <54a>. It is then decided <54b> whether
`scribed below uses both. image data and edge data.
`either a lane marker was not found or an old marker was
`used in the previous image. If so, the increment is in
`FIGS. 4a and 4c comprise a ?ow chart of an image
`processing method which uses a template matching
`creased by 20 pixels to enlarge the search area <54c>.
`Then the right bound BR and the left bound BL (for
`algorithm. The ?gures are joined at nodes A and B.
`each search area) is found <54d> by adding and sub
`Numerals in the text marked by angle brackets <>
`tracting the increment from the marker position MP for
`refer to reference numerals on the ?ow chart boxes.
`that line. Next a correlation procedure <56> is fol
`The raw image is digitized <40> into a 5l2><512><8
`image. The processing area is reduced in the vertical
`lowed for the de?ned search ?eld in the current line and
`and horizontal directions. In the vertical direction, it is
`if all the rows have not been processed <57> the
`limited by the horizon line 26 and by the hood 22. These
`search continues in another row <58> and the search
`?eld is rede?ned for that row.
`limits are obtained from the camera geometry and the
`calibration image. Then the gray level distribution of
`When the search area is de?ned a template matching
`operation begins <56>. For the de?ned search area, a
`the processing area is contrast expanded or normalized
`window of size 5X5 pixels with constant gray level of
`<42> to have standard statistics (e.g., a gray level
`240 is correlated with the image. The correlation mea
`distribution with mean= 100 and standard devia
`tion=20; the gray level range is 0 to 255 with 0 being
`sure is based on the absolute difference of the window's
`gray level and the image’s gray level. The window is
`full black and 255 being full white). Next, a 3X3 Sobel
`moved horizontally across the image pixel by pixel to
`?lter is used <44> to generate edge data 124x 512
`pixels in size.
`obtain a correlation measure at every point in the tra
`An obstacle in the lane ahead will give rise to strong
`verse and is then indexed vertically to make another
`edges in addition to those presented by the roadway.
`horizontal traverse. To reduce computations, the win
`The presence of obstacles is detected by counting the
`dow may be indexed vertically by more than one line so
`number of strong edge points within the area de?ned by
`that strips of the search area are sampled rather then the
`the lane boundaries of the previous frame <46> and
`complete search area. Using the absolute difference
`comparing to a threshold <48> for a given look-ahead
`metric, a perfect match will yield a zero element in the
`distance. For example, the number of pixels within 50
`correlation matrix. The correlation matrix is therefore
`searched for its minimum values. Elements with values
`feet of the vehicle having a gray level greater than 150
`are counted and compared to a threshold. If an obstacle
`under some threshold are selected to represent the
`marker positions.
`is detected within 50 feet, then a warning to the driver
`This correlation procedure will not yield any match
`is given <50> and the control returns back to the ?rst
`for yellow or faded white lane markers. If such markers
`step for a new frame. The location of lane boundaries is
`
`60
`
`55
`
`65
`
`20
`
`25
`
`35
`
`45
`
`1105-010
`
`
`
`20
`
`4,970,653
`5
`image area between the vehicle hood and the horizon is
`are present or markers are missing altogether, there will
`the only area operated upon. A thresholding operation
`be only a few match points. The match points are
`<106> generates a binary image by setting all gray
`counted <59> and if there are too few, say, 6 points
`out of a possible 100, the correlation is repeated with the
`levels above 150 to one and the rest to zero. Noise
`cleaning and boundary tracing <108> removes iso
`edge data <60>. Then if the marker is found <62> or
`lated, single pixels and traces the boundaries of the lane
`there were suf?cient match points <59>, the centroid
`and road markers. The boundary tracing algorithm
`of each marker is calculated < 64>. If the marker is not
`found <62> and the marker position on the opposite
`saves connected or adjacent pixels and follows the con
`side of the lane is known <66>, the missing marker
`tour of a lane marker or a road edge in all directions.
`This is done by dividing the image into three equal
`position is calculated based on the opposite marker
`segments representing the left, center and right portions
`position and the lane width <68> and its centroid is
`of the image. For each segment, the left and right edges
`calculated <64>. When the marker position on the
`opposite side of the lane is not known <66> the lane
`of the lane or road markers are searched for at least two
`adjacent edge pixels. After marking the edge pixels, the
`boundary from the previous frame is used <70>. After
`boundaries of the lane markers are traced until either
`the centroid of each marker is calculated <64>, if the
`the top of the image is reached or until no edge pixels
`markers are not found where expected <72> based on
`are found. The resulting boundary traced image has
`the previously detected lane geometry, the marker loca
`clean lane markers and road edges. Then obstacle detec
`tions from the previous frame are used <74>.
`tion is performed by counting edge pixels within the
`The determination of expected marker position
`<72> involves comparing the position of each marker
`lane boundaries and within 50 feet of the vehicle
`centroid with that of the previous frame. If the change
`<110> and comparing the count to a threshold
`<112>. If the count is greater than the threshold an
`is more than nine pixels a ?ag is set. Then a check is
`obstacle is detected and an obstacle warning is given to
`made for the presence of a second marker in each search
`the driver < 114> and the program returns to START.
`area as may occur, for example, when an exit lane is
`If no obstacle is detected a line Hough transform is
`sensed. The distance between two such markers is com
`performed <116>.
`puted and if it is above a threshold for some distance (15
`The Hough algorithm uses the boundary traced lane
`rows of the image) the outside marker (i.e., the right
`markers to estimate the location of the lane boundaries.
`most marker in the right search area) is rejected and a
`The Hough transform has been used extensively for
`?ag is set. Then for each row, if a marker is not found
`image processing applications. By knowing the desired
`the previous marker is used, provided no ?ag has been
`shape a priori, the transform maps complex shapes,
`set. If a flag is set, only new marker locations are used.
`whether lines or curves, into simple ones in the Hough
`The result of this procedure is a set of x,y coordinates
`domain. The simplest transform in terms of computa
`specifying lane markers. Then a least-squares line ?t is
`performed to generate the lane boundary <76>.
`tional burden is a line transform.Accordingly the lane
`boundaries are approximated by several straight lines.
`To validate the estimated lane boundaries they are
`FIGS. 7a and 7b show a line in the image plane and the
`checked for consistency of angles and width <78>. If
`the angle between the present lane boundary and the
`Hough transform of the same line. The normal parame
`previous lane boundary is greater than 35', then the
`terization of a straight line is given by R x cosA + y sinA
`present lane boundary is rejected. It is then estimated
`where R is the shortest distance from the origin to the
`line, and A is the angle between the normal to the line
`from the other lane boundary based on lane width.
`and the x axis. Each pixel in the Hough domain repre
`Similarly, the width of the lane is checked at two differ
`sents a potential line in the image space. After a line
`ent look-ahead distances for upper and lower bounds.
`Hough transform is performed on the boundary traced
`The algorithm also checks for lane changes by deter
`image <116> a search area is de?ned around the previ
`mining if the lower end of a lane boundary is near the
`ous boundary position <118> in Hough space as indi
`middle of the image <80>. If it is a lane change, warn
`ing is given <82> and depending on the lane change
`cated by the box 32 in FIG. 7b.
`The intensity of each pixel in the Hough domain
`direction (left or right), the markers are exchanged, i.e.,
`corresponds to the strength of a line in the image space
`for a left lane change, the left boundary is rede?ned as
`or the number of edge elements contributing to that
`the right boundary and a new left boundary is predicted
`line. To select the n strongest lines in the image, one
`based on the road/camera geometry.
`needs to identify the 11 pixels with the highest intensity
`The estimated lane boundaries are used to compute
`the vanishing point (where the left and right boundaries
`in the Hough domain. The following criteria are used to
`select the correct lines: (a) the intensity of the selected
`intersect) and the intersection of the right marker at a
`pixels should have local maxima, (b) the separation of
`?xed distance. These points and the vehicle dynamics
`the selected pixels should exceed a certain threshold,
`information are then used by a Kalman ?lter for estimat
`and (c) the intensity of the selected pixels should exceed
`ing the road curvature, the lateral shift from the center
`20% of the maximum intensity. The selection algorithm
`of the lane, and the heading error as disclosed in the
`is implemented by identifying the maximum pixel inten
`Dickmanns et al paper, supra. These estimates deter
`sity and deleting those that are less than 20% of that
`mine a coordinate transformation which determines the
`maximum intensity <119>. Then the Hough domain is
`look-ahead path required for autonomous vehicle con
`?ltered with an ll X 11 non-maximum suppression ?lter
`trol <86>. For display purposes the lane boundaries
`<120>, i.e., for a given pixel, if the other 120 pixels
`are overlaid on the image <88>.
`around it were higher in intensity, then the intensity of
`FIGS. 6a and 6b comprise a flow chart showing an
`image processing algorithm using a Hough transform.
`the subject pixel is set to zero. After ?ltering, only one
`local maximum pixel for each 11X 11 neighborhood
`The two ?gures are joined at node A. The process starts
`by digitization of the camera signals <100>, contrast
`could survive in the Hough domain but some points can
`still be close together. The number of Hough pixels is
`expansion <102> and edge detection by a Sobel ?lter
`further reduced by deleting pixels which would gener
`<104> as described for the previous method. The
`
`50
`
`55
`
`60
`
`65
`
`25
`
`35
`
`45
`
`1105-011
`
`
`
`4,970,653
`8
`7
`The description of a preferred embodiment of the
`ate near horizontal or vertical lines (A=0°, A=90°,
`invention for the purpose of illustrating the invention is
`A= 180') < 122>. Additional pixels are removed if the
`distance between two pixels is less than 34 <124>.
`not to be considered as limiting the invention since
`Finally the pixels are sorted based on intensity or the
`many modi?cations may be made by the exercise of skill
`number of contributing edge elements and the top four
`in the art without departing from the scope of the inven
`pixels are chosen <126>.
`tion.
`In a typical application, the roadway may include
`The embodiments of the invention in which an exclu
`discernible lane markers de?ning two lanes, and road
`sive property or privilege is claimed are de?ned as
`edges may also be visible even if not painted. In this case
`follows:
`four edges will be searched for to de?ne the road and
`1. In an automotive vehicle having a computer vision
`the vehicle relationship to the road. When the Hough
`system and an associated camera for viewing a scene
`transform is performed about 300 points may be present
`ahead of the vehicle, a method of detecting lane mark
`in Hough space which represent minor edges, major
`ers on a roadway comprising the steps of:
`edges including the ones of interest, and noise. Each
`obtaining an image of the scene and digitizing the
`operation reduces the number of remaining edges. For
`image,
`example, the deletion of those less than 20% of maxi
`normalizing the image,
`mum intensity may eliminate all but 50 points. The
`de?ning a search area in the image,
`non-maximum ?lter may leave 10 points, and the dele
`searching for lane markers in the search area of the
`tion of pairs that are close together and of those repre
`image to locate lane marker positions,
`senting horizontal and vertical lines may remove a few
`20
`estimating the position of any missing lane marker
`more points. Finally the strongest four pixels of the
`from said located lane marker positions, and
`remainder are selected as the lane markers and road
`de?ning lane boundaries from the said located lane
`edges.
`marker positions and the estimated position of the
`These pixels represent the lane boundaries. The pixels
`missing lane marker.
`are checked to determine if any lane boundary parame
`2. The invention as de?ned in claim 1 wherein the
`ters are missing <128> . If they are, the average of the
`step of estimating the position of missing lane markers
`last two frames parameters are used
`comprises averaging the corresponding marker posi
`If the parameters are not missing they are validated
`tions from previous images.
`by comparison with parameters of previous frames
`3. In an automotive vehicle having a computer vision
`<132> and if they are within 10% they are considered
`system and an associated camera for viewing a scene
`to be valid. If not, the average parameters of the last
`ahead of the vehicle and obtaining a series of images of
`two frames is used <130>. Then lane change is de
`the scene, a method of locating lane boundaries on a
`tected <134> and the look-ahead path is estimated
`roadway comprising for each image the steps of:
`<136> using the Kalman ?lter as described for the
`digitizing the image and normalizing the image,
`previous method, and an inverse Hough transform gen
`dynamically de?ning a search area in the image based
`erates lane boundaries <138> in the image space
`on the lane boundaries located in a previous image,
`which are overlaid on the image for display purposes
`searching for lane markers in the search area of the
`(140).
`image to locate lane marker positions,
`The Hough domain is quantized to a 256x256 space
`estimating the position of any missing lane marker
`for this implementation. Since pixels are selected based
`based on information selected from the current
`on a distance of 34, it is evident that the quantized space
`image and previous images, and
`can easily be reduced to 128><256 for decreasing the
`de?ning lane boundaries from the said located lane
`processing time. For real time applications, systolic
`marker positions and the estimated position of the
`array chips could be used for fast calculations of the
`missing lane marker.
`Hough transform. The image could be divided into
`4. The invention as defined in claim 3 wherein the
`several sub-images with several systolic chips working
`step of estimating the position of missing lane markers
`in parallel, mapping the results into several Hough do
`comprises the steps of:
`mains. Since the Hough transform operation is linear for
`determining the lane width from said previous image,
`the continuous case, the ?nal transform can be obtained
`by summing the partial Hough results. Another advan
`tage of the transform is that a priori knowledge about
`lane and road boundaries can easily be embedded in the
`algorithm.
`The method of the invention performed by either
`algorithm, template matching or Hough transform, has
`several advantages:
`(a) The method works well with missing and discon—
`tinuous markers, both yellow and white.
`(b) The search area in both algorithms is de?ned
`dynamically and adapts to the location of the lane mark
`ers, which saves processing time.
`(c) This method estimates road boundaries with low
`radius of curvature (e.g., entrance and exit highway
`ramps).
`(d) This method detects lane changes and warns the
`driver when lanes are crossed.
`(e) This method can detect certain obstacles in the
`lane.
`
`calculating the estimated position of a missing lane
`marker from the position of a marker on the oppo
`site side of the lane and the lane width.
`5. The invention as de?ned in claim 2 including the
`steps of:
`checking the