WO 2012/101355 PCT/FR2012/050055
CLAIMS
1. A method for processing successive digital image data, acquired by a
camera immersed in a liquid comprising turbulence phenomena causing an
effect of apparent displacement of pixels of the images,
characterized in that it comprises:
- a modeling of the effect of the turbulence on the pixels of the images, and
- a de-convolution by said modeling of a time-averaged image.

2. The method as claimed in claim 1, characterized in that the modeling comprises the estimation of a model of probability density of a displacement vector u relating to the effect of turbulence.
3. The method as claimed in claim 2, characterized in that said model is of the exponentially decreasing type.
4. The method as claimed in claim 3, characterized in that said model hσ(u) is expressed as a function of the vector u, by an equation of the type:
A \ B\\uih u =—2expl—— r
where:
- σ represents the standard deviation of the norm of the vector u,
- A and B are positive real constants.

5. The method as claimed in claim 4, characterized in that A=3/π and B=V6.
6. The method as claimed in one of the preceding claims, characterized in that the time-averaged image is estimated by an average weighted by powers of a forgetting factor, respectively decreasing as a function of time.

WO 2012/101355

PCT/FR2012/050055

7. The method as claimed in claim 6, characterized in that the average image Iα(x,n), estimated at the time n for a pixel with vector coordinates x, is given by:

the term a being a real number corresponding to the forgetting factor.
8. The method as claimed in one of the preceding claims, characterized in that the time-averaged image is estimated by a recursive temporal filtering.
9. The method as claimed in claim 8 taken in combination with claim 7, characterized in that an nth output of the temporal filter, for the time n, denoted If(x,n), and the average image Ia(x,n) are linked by the equations:

1_ 1-er
Iax ,ny-1—Ifx ,n~l

10. The method as claimed in one of claims 6, 7 and 9, characterized in that the forgetting factor α is chosen equal to 0.99.
11. The method as claimed in one of the preceding claims, characterized in that said de-convolution comprises the application of a linear spatial filtering, with an expression corresponding to said modeling.
12. The method as claimed in claim 11, characterized in that the filtering is of the Wiener type.
13. The method as claimed in claim 12, taken in combination with claim 2, characterized in that the spatial filter is given in the spectral domain by:

WO 2012/101355

PCT/FR2012/050055


where:
- f is a two-dimensional vector of two frequencies normalized after two-dimensional Fourier transform,
- fc is a cutoff frequency of the filter for eliminating any potential spectral aliasing,
- Sb(f) and Ssharp(f) are respectively spectral power densities of a noise and of a sharp image to be determined, and
- Hσ(f) is the two-dimensional Fourier transform of the probability density of the effect of the turbulence hσ(u).
14. The method as claimed in claim 13, characterized in that the spatial filter is expressed by:

where the term RBS is expressed as a function of parameters chosen such that the term RBS is in the range between 10-2 and 10-4.
15. The method as claimed in one of claims 13 and 14, taken in combination with claim 4, characterized in that the two-dimensional Fourier transform H0(f) of the probability density of the effect of the turbulence h0(u), is given by:
Hg f = 1 + C o.|f| 2 2
where C is a positive real constant and o is a positive real number counted in number of pixels.

WO 2012/101355

PCT/FR2012/050055

16. The method as claimed in claim 15, characterized in that the two-dimensional Fourier transform Hσ(f) is given by:

Ha f =

\ 2 11*11 2
1 + - 7T(T.\\f\3 nil

3
2

with σ in the range between 0 and 30 pixels.
17. The method as claimed in one of the preceding claims, in which said camera moves in translation, characterized in that a repositioning of a preceding image is applied to a current image, for the estimation of the average image.
18. The method as claimed in claim 17, taken in combination with claim 8, characterized in that it comprises a temporal filtering with repositioning and normalization.
19. The method as claimed in claim 18, taken in combination with claim 9, characterized in that it comprises the steps:
- for resetting, with:
lf(x,l)=0 N(x,l) = 0,
- for recursive temporal filtering, with:
where
[6^-i]=([^-i],[^n,«-i])
corresponds to a row shift x and/or column shift y vector, rounded to the nearest integers, between the current image n and the preceding image n-1 owing to the translation of the camera,
- for updating a normalization matrix N, with:
- and for normalizing the output of the temporal filter, with:


WO 2012/101355

PCT/FR2012/050055

20. A computer program comprising instructions for the implementation of the method as claimed in one of the preceding claims when this program is executed by a processor (PROC).
21. A device for processing image data, comprising an input for processing successive digital image data, acquired by a camera immersed in a liquid comprising turbulence phenomena causing an effect of apparent displacement of pixels of the images,
characterized in that it furthermore comprises calculation means for the implementation of the method as claimed in one of claims 1 to 19.
22. A use of the method as claimed in one of claims 1 to 19, for the
processing of image data acquired by a camera immersed in the water of a
nuclear installation, for filming fuel element assemblies, characterized in that
it comprises at least:
- the determination, from the acquired and processed images, of a state of
the fuel element assemblies.
23. The use as claimed in claim 22, characterized in that it comprises at
least:
- the determination, in said acquired and processed images, of at least one
identifier of a fuel element assembly.
24. The use as claimed in one of claims 22 and 23, characterized in that it
furthermore comprises:
- the measurement, in said images, of a spacing between fuel element
assemblies.
25. An installation for the implementation of a use as claimed in one of
claims 22 to 24, characterized in that it comprises at least one camera and
one device as claimed in claim 21.