Image analysis – Applications – Vehicle or traffic control
Reexamination Certificate
1999-04-20
2003-07-08
Mehta, Bhavesh M. (Department: 2625)
Image analysis
Applications
Vehicle or traffic control
C701S010000, C701S028000, C340S907000
Reexamination Certificate
active
06591000
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a lane mark recognizing system for recognizing a lane mark on a traveling road, and to a picked-up image preprocessing apparatus incorporated in this lane mark recognizing system. Furthermore, the present invention relates to a vehicle traveling control system for controlling the vehicle according to the lane mark recognized by the lane mark recognizing system, and to a recording medium storing a program for executing the preprocessing of the picked-up image in a computer system.
To realize an automatic driving of a vehicle, it is necessary to accurately recognize a traveling road ahead of the vehicle. According to a conventionally developed recognizing method, a lane mark is recognized from a picked-up image of the traveling road ahead of the vehicle. The lane mark color is generally white or yellow which is discriminable from gray or black of the road color. For example, the unexamined Japanese patent publication No. Kokai 5-289743 discloses a method for recognizing the lane mark based on binary-coded data of an original picture. The unexamined Japanese patent publication No. Kokai 7-239996 discloses a method for recognizing the lane mark based on binary-coded data resultant from the edging processing applied to an original picture.
However, the recognizing method disclosed in the unexamined Japanese patent publication No. Kokai 5-289743 is based on a discrimination of a bright portion (i.e., lane mark) from a dark portion (i.e., road). All of brightness change is thus detected as indicating the lane mark. In other words, this conventional recognizing method erroneously recognizes the bright portions other than the lane mark. To eliminate this problem, an appropriate postprocessing is required after obtaining the binary-coded data for removing the noise data corresponding to the bright portions other than the lane mark.
Similarly, the recognizing method disclosed in the unexamined Japanese patent publication No. Kokai 7-239996 requires a complicated postprocessing. For example, the picked-up image may include another vehicles traveling ahead of the subjective vehicle. In such a case, the edging processed data comprise the edge data corresponding to the preceding vehicles as well as the lane mark. Thus, the postprocessing is required to discriminate the edge of the lane mark from the edges of the preceding vehicles.
On the other hand, the unexamined Japanese patent publication No. Kokai 3-194669 discloses a lane mark detecting filter capable of emphasizing a bright region surrounded by a dark region. The likelihood of the lane mark is judged based on the output of the lane mark detecting filter. The noise data is then removed according to the likelihood judgement.
According to the unexamined Japanese patent publication No. Kokai 3-194669, the brightness distribution is monitored along a lateral direction crossing the lane mark on the picked-up image. The lane mark is generally brighter than the road. Thus, when the above-described lane mark detecting filter is applied to the picked-up image, a positive filtering output is produced in response to the bright portion. Generation of the filtering output according to this conventional system is explained in more detail with reference to
FIGS. 20A
to
20
D.
It is now assumed that the left half of an input image is a white region (brightness=200) and the right half is a gray region (brightness=100), as shown in FIG.
20
A.
The filtering output is obtained by applying the lane mark detecting filter with respect to an objective pixel (X, Y). The lane mark detecting filter has a plurality of filtering parameters arranged in a matrix pattern including a total of four matrixes each consisting of 3×3 pixels, with two matrixes (referred hereinafter to as “near-side matrixes”) positioned next to both sides of the objective pixel and another two matrixes (referred hereinafter to as “far-side matrixes”) positioned far from the both sides of the objective pixel. Each pixel has a positive pixel value (+1) in the two near-side matrixes. On the contrary, each pixel has a negative pixel value (−1) in the far-side matrixes.
FIG. 20B
shows a case where an objective pixel (Xa, Ya) resides in the white region so that all of the two near-side matrixes and the left far-side matrix are involved in the while region. Only the right far-side matrix is involved in the gray region. In this case, each of the near-side matrix produces 200*1*9=1,800. The left far-side matrix produces 200*(−1)*9=−1,800. The right far-side matrix produces 100*(−1)*9=−900. Therefore, a summed-up filtering output becomes 1,800×2−1,800−900=900.
FIG. 20C
shows another case where an objective pixel (Xb, Yb) resides in the gray region so that all of the two near-side matrixes and the right far-side matrix are involved in the gray region. Only the left far-side matrix is involved in the white region. In this case, each of the near-side matrix produces 100*1*9 =900. The right far-side matrix produces 100*(−1)*9=−900. The left far-side matrix produces 200*(−1)*9=−1,800. Therefore, a summed-up filtering output becomes 900×2−900−1,800=−900.
FIG. 20D
shows a filtered picture processed by the above-described lane mark detecting filter. As apparent from
FIG. 20D
, the lane mark detecting filter produces a positive filtering output in any transition from the bright region to the gray region. Thus, the positive filtering output may be produced in response to an erroneous bright portion other than the lane mark. For example, if a specific portion is as bright as the lane mark, a filtering output produced in a transition from this erroneous brighter portion to the road is quite similar to and not discriminable from the filtering output produced in the transition from the true lane mark to the road. Furthermore, if the brightness of this erroneous bright portion is fairly larger than that of the lane mark, the lane mark may be undesirably neglected due to the brightness difference between them.
In other words, the lane mark recognizing method of the unexamined Japanese patent publication No. Kokai 3-194669 is only reliable in an ideal condition where the lane mark has a clear contrast against the road with no erroneous bright portions. If there is any erroneous bright portion in the input image, the erroneous data must be removed in the detection of the lane mark.
SUMMARY OF THE INVENTION
An object of the present invention is to provide a picked-up image preprocessing apparatus capable of discriminating a bright portion corresponding to the true lane mark from other erroneous bright portions, performing preprocessing for obtaining only the data of the true lane mark, and improving the processing efficiency in the lane mark recognition.
Another object of the present invention is to provide a method for preprocessing the picked-up image used in the lane mark recognition.
Another object of the present invention is to provide a lane mark recognizing system incorporating the picked-up image preprocessing apparatus.
Another object of the present invention is to provide a vehicle traveling control system for controlling a vehicle based on the recognition result of the lane mark recognizing system.
Another object of the present invention is to provide a recording medium storing a program of a computer system for executing the above-described preprocessing.
The picked-up image preprocessing apparatus of the present invention is preferably applicable to a lane mark recognizing system for recognizing a lane mark on a traveling road ahead of the vehicle based on a picked-up image of the traveling road. For example, a CCD camera is installed on a vehicle body. The preprocessing apparatus of the present invention is applied to the image picked up by the CCD camera. The lane mark recognition is performed based on the preprocessed image.
The picked-up image preprocessing is
Hibino Katsuhiko
Imanishi Masayuki
Oike Tatsuya
Denso Corporation
Harness Dickey & Pierce PLC
Mehta Bhavesh M.
Tabatabai Abolfazl
LandOfFree
Apparatus and method for preprocessing a picked-up image,... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for preprocessing a picked-up image,..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for preprocessing a picked-up image,... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3102045