Image analysis – Applications – 3-d or stereo imaging analysis
Reexamination Certificate
1998-09-24
2001-03-20
Tran, Phuoc (Department: 2721)
Image analysis
Applications
3-d or stereo imaging analysis
C382S103000, C382S190000, C382S218000, C382S276000, C382S283000, C345S182000, C356S012000
Reexamination Certificate
active
06205242
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to an image monitor apparatus and a method to extract the existence of a monitored object or the motion of the monitored object from an image of a plane area such as a marked crosswalk.
BACKGROUND OF THE INVENTION
An ultrasonic sensor is used to measure the traffic flow or detect cars waiting for a traffic signal. The ultrasonic sensor can detect only solid objects, which adequately reflect the ultrasonic wave. Therefore, the ultrasonic sensor is not suitable for detection of a person. Furthermore, the measurement area of the ultrasonic sensor is narrow. Therefore, it is difficult to detect the existence or the number of persons waiting for a red signal adjacent to the marked crosswalk, or the flow quantity of pedestrians going across by the blue signal on the marked crosswalk.
Accordingly, a method for monitoring pedestrians using an input image through TV camera is developed. As a monitoring method using a video, the following three methods are mainly considered. As a frame differential method, the difference between two images inputted at two different points in time is calculated to detect the moving area as the pedestrian. As a background image subtraction method, the difference between the input image and the background image previously prepared is calculated to detect the area whose brightness is different from the background as the pedestrian. As an optical flow method, motion of each pixel in the image is calculated to detect the area of the pedestrian, in which each pixel is moving straight at constant speed along a predetermined direction.
However, the above-mentioned three methods respectively include the following defects. The frame differential method is not applied to a person standing still such as a pedestrian waiting for the traffic signal. In the background image subtraction method, in the case of an outdoor monitor area, an update of the background image according to changes in sunlight is difficult. Furthermore, a shadow area of the pedestrian and a brightness change area on the road caused by a car headlight are included in the detection area. Therefore, these areas must be excluded. In the optical flow method, expensive calculation cost is required for extraction of the optical flow from the image. Furthermore, the person standing still is not detected by the optical flow method.
SUMMARY OF THE INVENTION
It is one object of the present invention to provide an image monitor apparatus and a method to correctly extract the monitored object from the image without erroneous detection caused by the shadow cast by the monitored object and the brightness change area on the road plane.
According to the present invention, there is provided an image monitor apparatus, comprising: a plurality of image input means for respectively inputting an image based on each particular view position; an image projection means for projecting each of the plurality of images onto a predetermined plane in three-dimensional space; a comparison means for comparing projected image data of two images by unit of predetermined area on the plane and for extracting the predetermined areas, the projected image data of which are different, from the plane; an extraction means for extracting the existence position of a monitored object from the predetermined areas extracted by said comparison means; and a integration means for integrating or normalizing the existence positions of the monitored object extracted for the projected image data of all of the two images in the plurality of images as all existence positions of the monitored objects on the plane.
Further in accordance with the present invention, there is also provided an image monitor apparatus, comprising: at least two image input means for respectively inputting an image based on each particular view position; a transformation means for transforming the image inputted through one image input means according to a view position of the other image input means; a comparison means for comparing image data of the transformed image with image data of the image inputted through the other image input means by unit of same position on the image, and for generating a mask image representing the position of different image data according to the comparison result; and an extraction means for extracting an area of a monitored object from the mask image.
Further in accordance with the present invention, there is also provided a method for monitoring an object in an image, comprising the steps of: inputting a plurality of images from each particular view position; projecting each of the plurality of images onto a predetermined plane in three-dimensional space; comparing projected image data of two images by unit of predetermined area on the plane; extracting the predetermined areas, the projected image data of which are different, from the plane; extracting an existence position of a monitored object from the predetermined areas extracted; and integrating the existence positions of the monitored object extracted for the projected image data of all of the two images in the plurality of images as all existence positions of the monitored objects on the plane.
Further in accordance with the present invention, there is also provided a method for monitoring an object in an image, comprising the steps of: inputting two images from each particular view position; transforming one image inputted from one view position according to the other view position; comparing image data of the transformed image with image data of the other image inputted from the other view position by unit of the same position on the image; generating a mask image representing the position of different image data according to the comparison result; and extracting an area of a monitored object from the mask image.
Further in accordance with the present invention, there is also provided a computer readable memory containing computer readable instructions, comprising: an instruction means for causing a computer to input a plurality of images from each particular view position; an instruction means for causing a computer to project each of the plurality of images onto a predetermined plane in three-dimensional space; an instruction means for causing a computer to compare projected image data of two images by unit of predetermined area on the plane; an instruction means for causing a computer to extract the predetermined areas, the projected image data of which are different, from the plane; an instruction means for causing a computer to extract the existence position of a monitored object from the predetermined areas extracted; and an instruction means for causing a computer to integrate the existence positions of the monitored object extracted for the projected image data of all of the two images in the plurality of images as all existence positions of the monitored objects on the plane.
Further in accordance with the present invention, there is also provided a computer readable memory containing computer readable instructions, comprising: an instruction means for causing a computer to input two images from each particular view position; an instruction means for causing a computer to transform one image inputted from one view position according to the other view position; an instruction means for causing a computer to compare image data of the transformed image with image data of the other image inputted from the other view position by unit of same position on the image; an instruction means for causing a computer to generate a mask image representing the position of different image data according to the comparison result; and an instruction means for causing a computer to extract an area of a monitored object from the mask image.
REFERENCES:
patent: 5159648 (1992-10-01), Ogura et al.
patent: 5202928 (1993-04-01), Tomita et al.
patent: 5530441 (1996-06-01), Takatou et al.
patent: 5694483 (1997-12-01), Onoguchi
patent: 5706355 (1998-01-01), Raboisson et al.
patent: 5748778 (1998-05-01), Onoguchi
patent: 5859922 (1999-01-01), Hoffmann
p
Kabushiki Kaisha Toshiba
Mariam Daniel G.
Oblon & Spivak, McClelland, Maier & Neustadt P.C.
Tran Phuoc
LandOfFree
Image monitor apparatus and a method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image monitor apparatus and a method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image monitor apparatus and a method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2477055