Television – Special applications – Observation of or from a specific location
Reexamination Certificate
1998-05-13
2001-08-28
Kelley, Chris (Department: 2613)
Television
Special applications
Observation of or from a specific location
C382S105000
Reexamination Certificate
active
06281928
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to a positional detector device which is provided to detect license plates of motor vehicles.
2. Description of Prior Art
A motor vehicle control device has been introduced to administer entrances and exits of motor vehicles by reading the number of the license plate to obtain the numerical information on such places as parking lots and highway toll gates. In order to automatically recognize the numerical information of the license plate, it is necessary to detect a position of the license plate, and isolate the number to make the number distinct.
Whether or not the positional detection of the license plate is accurate significantly affects on the precision of the subsequent treatments. This is because it is particularly important to precisely carry out the positional detection of the license plate.
The following methods have been introduced to detect the position of the license plate from a frontal image of the motor vehicle.
(1) It is detected from straight lines (horizontal, vertical) of the contour of the license plate with the use of Huff's transformation.
(2) It is detected by tracing horizontal and vertical edges as disclosed by e.g., Laid-open Japanese Patent Application No. 6-28596.
(3) It is recognized by detecting the consecutive four or less digits of the number drawn on the license plate so as to distinct the number plate region with the use of two-value treatment and labeling treatment as disclosed by e.g., Laid-open Japanese Patent Application Nos. 60-193084 and 4-101286.
However, especially under adverse photographing conditions, it is difficult to detect the position of the license plate with the methods (1)~(3) because an image is produced in which the contour of the license plate and its numerical region are illegible. The same holds true when the license plate has a similar coloration of the background components (e.g., bumper and chassis).
Therefore, it is a main object of the invention to provide a positional detector device for a license plate which is capable of detecting a positon of a license plate with high precision even when a photographed image is produced in which the contour of the license plate and its numerical region are illegible.
SUMMARY OF THE INVENTION
According to the present invention, there is provided a positional detector device for a license plate comprising: a camera member which photographs a front and rear portion of a motor vehicle so as to produce an image signal; an A/D converter which converts the image signal into a digital image; a positional detector member which detects a position of a license plate of a motor vehicle based on the digital image; a cut-off treatment member which cuts off a specified region from an original image photographed by the camera member while scanning the original image when the motor vehicle approaches the camera member within a predetermined distance; an edge refining treatment member which emphasizes a contour of a pattern represented by the specified region cut off from the original image; a contraction treatment member which contracts an image size of the specified region which was treated with the edge refining treatment member; a calculation treatment member which feeds the contracted pattern to a learned neural network so as to calculate an output value for a positional detection neural network; a coordinate transformation member which transforms the output value of the positional detection neural network to such a scale as to meet the original image; an addition treatment member which adds data at specified points which were multiplied by Gaussian window so as to obtain a projective addition value after carrying out the coordinate transformation; and a distinction treatment member which recognizes a position of the license plate based on a maximum value on which the projective addition value falls.
According to another aspect of the invention, there is provided a positional detector device for a license plate comprising: a camera member which photographs a rear portion of a motor vehicle so as to produce an image signal when the motor vehicle approaches the camera member within a predetermined distance; an A/D converter which converts the image signal into a first sized original image; a positional detector member which detects a position of a license plate of a motor vehicle based on the original image; a cut-off treatment member which cuts off a second sized region from the first sized original image photographed by the camera member while scanning the first sized original image; an edge refining treatment member which emphasizes a contour of a pattern represented by the second sized region cut off from the first sized original image; a contraction treatment member which contracts the second sized region into a third sized image after treated with the edge refining treatment member; a calculation treatment member which feeds the contracted pattern of the third sized image to a learned neural network so as to calculate an output value for a positional detection neural network as a fourth sized image value; a coordinate transformation member which projects the output value of the positional detection neural network to a positional distinction pattern, a size of which is the same as that of the first sized region, and transforming the fourth sized image value into the second sized region; an addition treatment member which adds data at specified points which were multiplied by Gaussian window to a positional distinction pattern of the license plate so as to obtain a projective addition value after carrying out the coordinate transformation; and a distinction treatment member which recognizes a position of the license plate based on a maximum value on which the projective addition value falls.
According to another aspect of the invention, learning of the neural network is based on a back propagation method which employs learning date and teaching data, and a unit threshold function is based on sigmoid function.
According to still another aspect of the invention, the detector device is used to recognize a number of the license plate when the motor vehicle passes through a highway toll gate without stopping there.
With photographed images of the frontal and rear portion of coming and going motor vehicle taken by the camera, the image signal is produced, The image signal is converted into the digital image through the A/D converter.
On the basis of the original image taken when the motor vehicle approaches within the predetermined distance, the specified small image is cut off from the original image while scanning the original image (cut-off treatment).
The edge refining treatment is carried out to extract common features such as the letters and the contour of the license plate among various types of the license plates in which black, white and black letters are in turn drawn with white, green and yellow backgrounds.
The photographing conditions generally vary depending on seasons of the year and hours of the day so as to deteriorate the contrast of the image. However, the edge refining treatment involves differential calculus procedures. This remains the image substantially immune to the direct current component.
After going through the edge refining treatment, the size of the imgage is contracted to eliminate the unsubstantial information, and meeting the input size for the neural network (contraction treatment).
The contracted pattern of the image is fed to the learned neural network so as to calculate the output value for the positional detective neural network (calculation treatment).
The output value is reduced to a scale proportionate to the original image by way of the coordinate transformation (coordinate transformation treatment).
In order to achieve the high precision, the data obtained after the coordinate transformation treatment are added at each point multiplied by the Gaussian window so as to attain the projective addition value (addition treatment).
The position of the license plate is specified
Kanade Takeo
Matsumoto Noriyuki
Umezaki Taizo
An Shawn S.
Chuo Hatsujo Kabushiki Kaisha
Kelley Chris
Sughrue Mion Zinn Macpeak & Seas, PLLC
LandOfFree
Positional detector device for a vehicular license plate does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Positional detector device for a vehicular license plate, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Positional detector device for a vehicular license plate will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2474403