Image analysis – Applications – Vehicle or traffic control
Reexamination Certificate
1999-12-17
2001-02-13
Johns, Andrew W. (Department: 2721)
Image analysis
Applications
Vehicle or traffic control
C382S274000, C340S934000, C348S149000
Reexamination Certificate
active
06188778
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to method and apparatus for measuring a state of traffic congestion on a road on the basis of brightness levels of sample points which are set in a motion picture representing the road and vehicles coming and going on the road. The present invention also relates to image processing method and apparatus for setting sample points in an image and processing the image.
BACKGROUND OF THE INVENTION
There are a wide variety of systems for measuring a state of traffic congestion for the purpose of smoothing traffic on roads. Such traffic congestion measuring systems include two well known measuring systems, one of the two measuring systems being of a supersonic type and the other of the two measuring systems being of a loop-coil type. The supersonic type of traffic congestion measuring system adopts a supersonic sensor as means for sensing the presence and movement of vehicles, while the loop-coil type of traffic congestion measuring system adopts a loop-coil as means for sensing the presence and the movement of the vehicles.
The supersonic type of traffic congestion measuring system uses the supersonic sensor, which is positioned above the road at a predetermined height, to detect the presence and velocity of a passing vehicle. Based on results of the detection by the supersonic sensor, the state of traffic congestion is measured. On the other hand, the loop-coil type of traffic congestion measuring system uses the loop-coil, which is buried under the road, to detect the presence and the velocity of the vehicle passing above the loop-coil on the basis of the variation in magnetism above the loop-coil caused by the passing vehicle. Based on results of the detection by the loop-coil, the state of traffic congestion is measured.
In the supersonic sensor type and loop-coil type of traffic congestion measuring systems thus designed, the supersonic sensor and the loop-coil are merely operated to obtain information of only the vehicles lying directly under the supersonic sensor and directly above the loop-coil, respectively. In point of fact, the supersonic sensor type and the loop-coil type of traffic congestion measuring systems are merely operated to indirectly measure the state of traffic congestion on the basis of the number of the passing vehicles during a certain period or the velocities of sampled vehicles corresponding to extremely small part of the vehicles passing over the supersonic sensor and the loop-coil, respectively. For this reason, the supersonic sensor and the loop-coil of traffic congestion measuring systems have difficulty in automatically measuring, in real time, traffic congestion ranges with high accuracy. For instance, if the supersonic sensor type and loop-coil type of traffic congestion measuring systems are adopted for the purpose of controlling traffic signals in accordance with the length of a traffic queue extending from an intersection, a drawback is encountered in that the supersonic sensor type and loop-coil type of traffic congestion measuring systems have difficulty in controlling the traffic signals so as to quickly relieve the traffic congestion.
There are also two traffic congestion measuring systems adopting video cameras, which are superior to the supersonic sensor type and loop-coil type of traffic congestion measuring systems in that the state of traffic congestion is measured in real time with high accuracy. One of the traffic congestion measuring systems is disclosed in “INPACTS: A New TV Image Processing System for Monitoring Traffic Conditions,” by Wootton Jeffreys CONSULTANTS or European Patent publication No. 0403193. In the disclosed measuring system, an image of a road taken by a video camera is divided into a plurality of blocks for each traffic lane. The size of each block is roughly equal in length to each of the vehicles represented in the image. By processing the image, the blocks are classified into three different block groups. The first block group includes blocks in which no vehicle is detected. The second block group includes blocks in which a moving vehicle is detected. The third block group includes blocks in which a stationary vehicle is detected. According to the arrangement of the three different type blocks, the state of traffic congestion is measured indicating, for instance, conditions substantially stationary, slowly moving, and smoothly moving.
Other traffic congestion measuring systems are disclosed in Kitamura et al., “Traffic Congestion Measuring System Using Image Processing”, Annual Conference sponsored by The Institute of Electrical Engineers of Japan, Industry Application Society, 1991. The traffic congestion measuring system operates in a manner which comprising the steps of abstracting three different feature values representative of the density of vehicles, the movement quantities of vehicles and the brightness of a road surface, respectively, from images taken by video camera, inputting the feature values into an input layer partially forming a neural network, calculating an output value stepwise varied between “0” and “1” by 0.2, and detecting, on the basis of the output value, the state of traffic congestion. The state of traffic congestion is any of five different conditions; a first condition that there is no traffic, a second condition that the vehicles are smoothly moving, a third condition that the vehicles are moving but crowded, a fourth condition that there is produced a slight traffic congestion, and fifth condition that there is produced a serious traffic congestion.
The conventional traffic congestion measuring systems using the video camera, however, also encounter drawbacks. In the former traffic congestion measuring system adopting the video camera, a correlation of the arrangement pattern of the blocks classified into three types and the state of traffic congestion in each of measurement points is required to be previously learned. In addition, the former traffic congestion measuring system must perform a process of dividing each of the traffic lanes into the blocks. For this reason, the traffic congestion occurring over two or more traffic lanes cannot be detected and, as a result, the state of traffic congestion cannot be measured with accuracy. Furthermore, the size of the vehicle represented in the image becomes smaller as a distance from the video camera to the vehicle becomes larger. The block size is, therefore, adjusted with difficulty to the size of vehicle spaced apart from the video camera at a long distance. Consequently, the former traffic congestion measuring system adopting the video camera cannot measure the state of traffic congestion over a long distance.
In the traffic congestion measuring system adopting the video camera, the traffic congestion measuring system is limited in forming the neural network based on a large amount of data for the sake of learning. The process of forming the neural network takes much time and labor and is a difficult task. In addition, the neural network must be re-formed whenever road circumstances are changed. Furthermore, the whole image is processed in the lump in order to measure the state of traffic congestion, so that detailed traffic information for each part of the image cannot be detected. The detailed traffic information includes, for instance, information pertain to traffic queue length expressed as “there is a traffic congestion of y meter length extending from point X
1
to point X
2
on the road”.
The above traffic congestion measuring systems which use the video cameras encounter the foregoing drawbacks and, for this reason, are put to practical use with difficulty. In the view of the foregoing surroundings, the present invention is made and has an objective of providing a traffic congestion measuring method and apparatus capable of reducing the previous learning, detecting the traffic congestion traversing two or more traffic lanes, stably measuring the state of traffic congestion independently of the changes in circumstances such as changes of measurement points and periods,
Higashikubo Masakatsu
Ito Yoshiyuki
Johns Andrew W.
Pillsbury Madison & Sutro LLP
Sumitomo Electric Industries Ltd.
LandOfFree
Traffic congestion measuring method and apparatus and 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 Traffic congestion measuring method and apparatus and image..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Traffic congestion measuring method and apparatus and image... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2563205