Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T5/00—Image enhancement or restoration
  • G06T5/40—Image enhancement or restoration using histogram techniques
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T5/00—Image enhancement or restoration
  • G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T5/00—Image enhancement or restoration
  • G06T5/90—Dynamic range modification of images or parts thereof
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/10—Image acquisition modality
  • G06T2207/10032—Satellite or aerial image; Remote sensing
  • G06T2207/10044—Radar image

Definitions

• the present invention relates to the modeling of digital images made up of pixels and representing a scene having at least one area of interest of particular radiometry, on a non-uniform background which will be called "clutter", in order to better show areas of interest or differentiate areas of interest of different natures.
• radiometry will designate a parameter representative of each point of the scene represented by the image and making it possible to constitute an image the pixels of which are each assigned a value representative of the radiometry.
• radiation can notably designate:
• Radiometry will generally be represented either by an amplitude, or by an energy or intensity (square of the amplitude).
• the invention finds a particularly important application in the modeling and interpretation of the images provided by a synthetic aperture radar, called SAR or RSO.
• These images show a speckle that is found in any image obtained by illumination in coherent waves, for example by laser or in ultrasound imaging.
• the RSO uses a number L of images, or views of the same site, acquired almost at the same time and statistically independent, which makes it possible to generate an L-views image less resolved than a single view image, but with less speckle.
• Two types of images are used: intensity images and amplitude images; For the latter, two types exist:
• the invention aims in particular to provide an image modeling method making it possible in particular to reduce the effects of noise and to better differentiate the zones of interest between them or with respect to a noisy background.
• the invention notably proposes a digital image modeling method consisting of pixels, according to which each pixel of coordinates i, j is defined by the values of parameters ⁇ ij, ⁇ 1 and ⁇ 2 representing the histogram Hvij of the radiometries on a determined neighborhood, of identical size for each pixel, in a representation in the form of a mixture.
• N is an integer greater than or equal to 2
• L is a positive number chosen according to the nature of the image or the number of views, known or estimated, in the case of mutil-viewed images.
• the numbers ⁇ k denote a set of N positive coefficients, with a sum equal to 1.
• the numbers ⁇ k designate a set of N parameters to be developed, representative of N complementary areas of the neighborhood,
• the parameters ⁇ , ⁇ 1 and ⁇ 2 on the different pixels are estimated from intensity histograms on the neighborhoods, generally of identical sizes, by solving the system of equations which results from the calculation of the cumulants of order 1, 2 and 3.
• the invention also aims to provide methods for exploiting the modeled image.
• the pixels are assigned to one or other of finite number zones, for example 2, of different natures
• operating modes include:
• the values of the parameters ⁇ , j, ⁇ 1, memo ⁇ 2, j of the mixing law can be used, at any point ij of the radar image, to separate the components a folding of the radar image, if we know a digital terrain model or DTM and the shooting geometry.
• FIG. 1 shows an example of a scene image comprising two distinct zones of different radiometric characteristics, affected by noise, and a particular size of neighborhood;
• FIG. 2 shows a sequence of processing performed on the image to arrive at modeling and possibly filtering
• FIG. 3 is a flowchart showing the method of obtaining ⁇ concerned, ⁇ 1 forbidden, ⁇ 2 forbidden
• the figure schematically shows an image Lvue, consisting of a rectangular matrix of pixels Lij belonging to the i th line and to the j th column.
• Each neighborhood is assigned a neighborhood 12, identical for all the pixels.
• it will generally be insufficient to adopt a neighborhood of 9 pixels.
• we will adopt a square neighborhood of at least 5 x 5 pixels.
• Each pixel of the image has a radiometric value x.
• For each pixel we calculate and optionally store log x, (log x) 2 and (log x) 3 .
• the table thus obtained constitutes a multiple histogram (step 14 in FIG. 2). From there, a neighborhood histogram is produced on each pixel ij.
• the constitution of neighborhood histograms implies, for each pixel, to do as many times the operations indicated in Figure 3 as there are points in the neighborhood. For each of the pixels belonging to the neighborhood, the operations shown in FIG. 3 give rise to three parameters, called "cumulants" A, B and C. The cumulants have, for each neighborhood pixel of size N,
• the operations carried out involve: - square or cube elevations and multiplications by constants (octagonal boxes) and
• a test makes it possible to determine whether the neighborhood of a determined pixel corresponds to a binary mixture of laws G ( ⁇ , L). For this, we solve a quadratic equation. Depending on the case, this equation has roots or not. If the discriminant of this equation is positive, this means that the vicinity of the pixel considered covers two different areas. If the discriminant is negative, this means that the neighborhood does not meet the conditions indicating the presence of a mixture of two zones.
• the backscatter coefficient ⁇ can be determined by the flowchart in Figure 6, where p represents ⁇ 1 or p2 and ⁇ represents ⁇ 1 or ⁇ 2, as the case may be.
• Figure 7 shows such a decomposition in two laws, while the calculation of a traditional G law would lead to the mean law represented in thick lines.

Landscapes

• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Image Processing (AREA)

Applications Claiming Priority (3)

Publications (1)

Family

ID=8867832

Family Applications (1)

Country Status (3)

Families Citing this family (1)

• 2001
  • 2001-10-02 FR FR0112648A patent/FR2830356B1/fr not_active Expired - Fee Related
• 2002
  • 2002-10-02 WO PCT/FR2002/003359 patent/WO2003030102A2/fr not_active Ceased
  • 2002-10-02 EP EP02800175A patent/EP1433136A2/de not_active Withdrawn

Non-Patent Citations (2)

Also Published As

Similar Documents

Legal Events