US2010033603A1PendingUtilityA1
Method for extracting raw data from an image resulting from a camera shot
Est. expiryMar 23, 2024(expired)· nominal 20-yr term from priority
H04N 23/84H04N 23/81G06T 3/00
48
PatentIndex Score
0
Cited by
0
References
0
Claims
Abstract
The method according the invention allows the extracting raw data from an image resulting from a camera shot. It comprises determining, for each point of the image of a combination V 0 [C,L] of colour components of the image, calculating, for each point of the image, of a value V N+1 [C,L], iterating said calculating a predetermined number of times then taking into account the values of the final image V Nfinal [C, L] in each point of the image, calculating for each point of the image of the difference D [C, L]=V Nfinal [C, L]−V 0 [C, L], calculating of a noise contextual datum V S , correcting the extracted raw data D[C, L], with the contextual datum V S , calculating of a corrected value I*[C, L] taking into D*[C, L] and presenting the extracted data under a desired angle.
Claims
exact text as granted — not AI-modified1 . Method for extracting raw data from a digital image taken by a photographic or cinematographic for correcting the extracted raw data and for generating an improved digital image containing the corrected raw data in the case when extracted data from an image or of an image containing them, are shown according to a desired view angle, from an image taken by a camera under any incidence, which comprises the steps:
searching for at least four identifiable characteristic points of a pattern present in the image taken by the camera defining contextual data, these characteristic points may consist of corners of the image, optionally extracting data according to predetermined criteria, calculating geometrical deformations to be made to the raw image or to the extracted data or to the image containing them, from the relative position of the four points, with respect to relative reference positions, determining corrections to be made to the raw image or to the extracted data or to the image containing them, depending on the geometrical deformations, and generating a corrected image taking into account the thereby determined corrections.
2 . The method according to claim 1 , for obtaining a corrected image having the same proportions as the object, said method comprises the determination of the real height/width ratio of the quadrilateral defined by the said points and the taking into account of this ratio r in the generation of the corrected image.
3 . The method according to claim 2 , wherein said quadrilateral is the projection of a rectangle, and in that the determination of the proportions of the rectangle is performed in accordance with a process comprising the following steps:
determining vanishing points from contours of the pattern and determining a horizon line connecting the vanishing points, determining the coordinates of the projection point F of the optical centre O of the camera on the horizon line, calculating the camera base point as orthogonal projection of the optical centre of the camera on the plane of the pattern from distances between the vanishing points and the projection point F and from the distance between this projection point F and the optical centre O, calculating the focal length from the distances between the optical centre, the projection point F and the camera base point, calculating the coordinates of the intersection points between the vanishing lines and the lines connecting the camera base point and the vanishing points as well as points O 1 , O 2 , P 1 , P 2 , located on the vanishing lines, at a conventional distance from the camera base point, and calculating the ratio of the sides of the initial pattern from the coordinates calculated beforehand by considering that the rectangle O 1 , O 2 , P 1 , P 2 , is the projection of a square extending in the plane of the pattern.
4 . The method according to claim 3 ,
which comprises the following steps in the case when both vanishing points exist,
calculating the coordinates of the point F by projecting the centre of the image O on the horizon line (F 1 , F 2 )
calculating the position of the base of the camera by its distance to point F, dist (X,F), by means of the relationship:
dist
(
X
,
F
)
=
dist
(
F
,
F
1
)
·
dist
(
F
,
F
2
)
dist
(
O
,
F
)
calculating the focal length f with the relationship:
f =√{square root over ( dist ( O,X )· dist ( O,F ))}{square root over ( dist ( O,X )· dist ( O,F ))}
determining the coordinates of points M 1 , N 1 , O 1 and P 1 from the values calculated beforehand
determining the coordinates of points M 2 , N 2 , O 2 and P 2 from the values calculated earlier, and
calculating the physical aspect ratio r by using the crossed ratios and the fact that the rectangle O 1 , O 2 , P 1 , P 2 is the projection of a square extending in the plane of the pattern according to the relationship
r
=
dist
(
M
1
,
N
1
)
dist
(
M
2
,
N
2
)
·
dist
(
F
1
,
O
1
)
dist
(
F
1
,
M
1
)
·
dist
(
F
1
,
P
1
)
dist
(
F
1
,
N
1
)
·
dist
(
F
2
,
M
2
)
dist
(
F
2
,
O
2
)
·
dist
(
F
2
,
N
2
)
dist
(
F
2
,
P
2
)
5 . The method according to claim 2 wherein, in the case when only one pair of vanishing lines intersect at a vanishing point whereas the other two vanishing lines are parallel, wherein the vanishing point projected to infinity, the calculation of the ratio r will be performed from a pre-established focal length f of the camera.
6 . The method according to claim 5 wherein said ratio r is obtained according to the following relationship wherein f is
r
=
dist
(
A
,
B
)
·
dist
(
C
,
D
)
(
dist
(
C
,
D
)
-
dist
(
A
,
B
)
)
·
f
2
+
dist
(
O
,
F
2
)
2
the focal length of the camera wherein the focal length f is calculated beforehand.
7 . The method according to claim 2 wherein, if there is no vanishing point, the ratio r is equal to the ratio
r
=
dist
(
A
,
B
)
dist
(
A
,
D
)
8 . The method according to claim 1 wherein, the generation of a corrected image comprises an operating sequence including the following phases:
creating an initial, deformed binary mask of the area to be corrected by isolating the useful portion of the initial image containing the extracted data and by assigning the same binary value to the pixels of this useful portion, calculating an ideal binary mask by a direct homographic transformation of the initial mask based on the transformation of any polygonal shape into a reference polygonal shape, and for each pixel (u, v) of the ideal binary mask, calculating by inverse homography, the position (x, y) in the initial image, determining the value of the final image at pixel (u, v) by an (x, y) interpolated value in the initial image.
9 . The method according to claim 8 wherein said step for creating a binary mask comprises assigning a zero value to all the pixels which are outside a quadrilateral surrounding the useful portion of the image as well as to the pixels which do not correspond to optionally extracted data.
10 . The method according to claim 9 wherein the pixels are considered to be within the quadrilateral if they are always on the same side as a point G inside the quadrilateral relatively to the limits of the quadrilateral, point G may consist of the centre of gravity or the point of intersection of the diagonals.
11 . The method according to the claim 1 comprising a precalculation of the images of intermediate lines and intermediate columns in order to obtain the images of subpixels by intersection of precalculated images of intermediate lines and intermediate columns.
12 . The method according to the claim 8 , wherein, in the calculation of the final image, to each pixel of the ideal binary mask is assigned an intensity value which is calculated by finding the position of this pixel in the initial image or the image of the extracted data.
13 . The method according to the claim 8 , wherein, in the phase for calculating the final image, the images are precalculated by inverse homography of the lines and columns of the ideal mask, and the position of a given pixel in the initial image is then inferred by calculating the intersection of two lines.
14 . The method according to claim 1 , wherein the step for creating the final image comprises the calculation of the position in the deformed image of a pixel u,v of the ideal mask, by the intersection of the precalculated inverse images of line v and column u, an intersection which defines an x,y point of the initial image, and in that an intensity value is assigned to the pixel (u,v) which is then interpolated at the x, y point of the initial image or image of data extracted from the luminance image or from each colour channel.
15 . The method according to claim 14 ,
wherein the interpolation is bilinear.
16 . The method according to the claim 1 , comprising the simulation of an image, projected from data contained in a rectangular object with apices (A, B, C, D) having a prescribed physical aspect ratio r=CD/AB, a prescribed projected point D in the image and a known projected distance (CD), with a camera having a prescribed focal length (f), a tilt angle ((π/2)−i) where i is the angle of incidence, an angle of rotation α around the axis of the camera and if i≠0, a prescribed skew angle (β) relatively to one of the existing vanishing points F 1 ,
said method comprising the following steps:
a first step for calculating the position of the three unknown points (A,B,C) consistently with the physical aspect ratio r,
a second step for calculating homographic relationships in order to project the information contained in the original rectangular object onto the simulated image, and
a third step for determining luminance and chrominance of the simulated image with the homographic relationships determined earlier.Join the waitlist — get patent alerts
Track US2010033603A1 — get alerts on status changes and closely related new filings.
We store only your email — no account needed. See our privacy policy.