Image analysis – Applications – Motion or velocity measuring
Reexamination Certificate
2000-11-30
2004-06-22
Boudreau, Leo (Department: 2621)
Image analysis
Applications
Motion or velocity measuring
Reexamination Certificate
active
06754372
ABSTRACT:
The present invention relates to a method for detecting objection motion in a sequence of video images.
Such motion detecting methods already are known and illustratively are described in an article “Determining optical flow: a retrospective” by B K P Horn and B G Schunkin in ARTIFICIAL INTELLIGENCE (1993), pp 81-87 and in a book,“Video Coding” by L. Torres and Murant Kunt in KLUWER ACADEMIC PUBLISHERS. Illustratively the methods of the prior art are based on optical flux, matching blocks, contours etc.
These methods incur the drawback to demand much computational time and therefore are of little use when detecting moving objects in real time.
On the other hand a simpler and quicker method consists in calculating the differential image of each pixel between two consecutive images of the video sequence. This differential in general is calculated soled using the brightness information.
Accordingly, for each pixel's coordinates (x, y), the differential in the brightness signal of the images of order i and i+1 will be determined, namely
∀
i
&Dgr;L
(
x,y
)=
L
i−1
(
x,y
)−
L
i
(
x,y
)
where L
i−1
(x,y) and L
i
(x,y) resp. denote the brightness signal from the pixel of coordinates (x,y) of the images of orders i and i−1.
Due to the noise caused by the electronics and the camera pickups, in general a threshold will be introduced, whereby the differential signal is quantized as follows
∀
x, ∀y I
(
x,y
)=when &Dgr;
L
(
x,y
)<threshold
I
(
x,y
)=1 if
L
(
x,y
)≧threshold
where I(x,y) is quantized signal differential for the coordinates (x,y).
The value of a quantized signal associated with a pixel of coordinates (x,y) is a first value, here zero, if the brightness differential signal is less than a threshold value and will be a second value, here 1, if the differential signal is larger than said threshold value.
As regards a standard commercial camera, the applicable thresholds are about 5% of the possible excursion of the brightness values (typically, if the brightness varies between 0 and 255, the threshold will be about 12).
This procedure detects mainly the contours of the moving objects. It requires long calculation times to gain all the zones composing a moving object.
The patent document WO 90 01706 A also discloses an image processing method allowing acquiring objects with a stochastic background. In this method, each stage generates first and second image differentials which then are processed in a two-level threshold detector, said levels representing mean positive and negative noise and being fitted with coefficients. If the value of an image differential pixel exceeds the positive threshold value, said pixel is assigned a value of 1. If the value of a image differential pixel is less than the negative threshold value, a value of −1 is assigned to said pixel. When the pixel value is between the two threshold values, the assigned value is zero.
Using the above notation, one may then write
∀
x, ∀y I
(
x,y
)=1 if
I
(
x,y
)=1 if &Dgr;
L
(
x,y
)>threshold
I
(
x,y
)=0 if−threshold>&Dgr;
L
(
x,y
)>threshold
where I(x,y) is the quantized signal differential for the pixel of coordinates (x, y).
When using this method, and considering that the threshold levels depend on the noise, it is quite difficult, even impossible, to detect motion at values much below the camera noise. To resolve this problem, the said document uses lowpass filters for the image differential values.
Nevertheless, detection sensitivity is linked to the threshold being used to calculate the image differential and therefore it is at the level of the noise of the camera being used.
REFERENCES:
patent: 5764803 (1998-06-01), Jacquin et al.
patent: 6188776 (2001-02-01), Covell et al.
patent: 6188798 (2001-02-01), Lee
patent: 6229854 (2001-05-01), Kikuchi et al.
patent: 90/01706 (1990-02-01), None
W. Pratt, “Digital Image Processing”, 2ndEd. 1991, pp. 597, 608-609.
Boudreau Leo
France Telecom S.A.
Lowe Hauptman & Gilman & Berner LLP
Nakhjavan Shervin
LandOfFree
Method for determining movement of objects in a video 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 Method for determining movement of objects in a video image..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for determining movement of objects in a video image... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3337129