Classifications

• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
  • G06V10/00—Arrangements for image or video recognition or understanding
  • G06V10/40—Extraction of image or video features
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
  • G06V10/00—Arrangements for image or video recognition or understanding
  • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
  • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
  • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
  • G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching

Definitions

• the invention relates to a method according to the preamble of claim 1.
• the object of the invention is the similarity of two objects based on two
• Each object image is used in a number of images for evaluation, i. E. at least two images, preferably a plurality of images, of each of the two objects are made available or used for the similarity comparison.
• the respective image sets of the two subject images in the form of hyperspectral images with predetermined spectral resolution and / or in the form of images in different color channels, eg XYZ, ClELab, ClELuv, RGB and / or CHS color channels , present.
• spectral resolution e.g XYZ, ClELab, ClELuv, RGB and / or CHS color channels
• correspondingly many values of vectors can be created, which contributes to the accuracy of the similarity comparison.
• the method according to the invention can be used advantageously if the method for comparing the similarity of actual images of objects with target images stored in stored form is used, in particular for comparing an actual image, e.g. a printing unit, a surface design, an article surface, a fabric, a wallpaper, a veneer plate, a patterned article surface, with a corresponding, a target value specifying comparison image.
• an actual image e.g. a printing unit, a surface design, an article surface, a fabric, a wallpaper, a veneer plate, a patterned article surface, with a corresponding, a target value specifying comparison image.
• the invention further relates to a data carrier with a program stored on this program for carrying out the method according to the invention.
• Fig. 1 shows two sets of images to be compared.
• Fig. 3 shows a vector v of a certain length and its angle ⁇ with respect to a reference vector v ref .
• 4 shows a point cloud formed by the lengths of the vectors v and v ', respectively.
• Fig. 5 shows a point cloud formed by the difference angles of the vectors v and v ', respectively.
• Fig. 1 two sets of object images 1, 2, 3, 4, 5 and 1 ', 2', 3 ', 4', 5 'are shown, which originate from identical or congruent image areas. These images or image areas have pixels or pixels X ⁇ 1 , x 2 yi, x 3 , yi ..., xy 2 , x 2 y 2 > ⁇ 2, ••• •
• the individual images of the two image sets are using different recording techniques taking into account different image parameters, for example by recording the object in different wavelength ranges and / or different color channels.
• a vector is now formed in which the intensity values determined for the respective pixels in the individual images, ie gray values or color values, are entered. Is obtained in this manner, a number of vectors v ⁇ y containing as values of the color intensities of the mutually corresponding image points or pixels of the images of the two object images.
• All values of the pixels with the same x and y values of the individual images of each of the two image sets each yield a vector v * y (z,) with z elements, where z is the number of images in the image set or eg the number of color channels can be.
• Each vector can be interpreted as a point in z-dimensional space, as shown in FIG.
• the procedure is such that for each of these vectors v , v 'its length
• FIG. 3 explains the determination of the difference angles ⁇ and ⁇ 'of the vectors v and v ' to a predetermined reference vector v r ⁇ f .
• the length and angle of the vectors determined for the individual pixels of the image areas of the two image sets are possibly different due to any differences or deviations of the color values and thus the vectors are not the same length and / or include a different angle to the reference vector v ref . If one plots the lengths L and L 'of the vectors v and v ' obtained for corresponding pixels of the image areas of the two image sets against one another in a diagram, ie one forms points with the coordinates (L, L '), one obtains those shown in FIG 4 illustrated point cloud.
• the vector v g can be represented as the tuple (L, j , ⁇ , j ) or the vector v , / as (L 1J 1 , ⁇
• One is still at the position ij of the two sets of images and now the lengths L and L 'as well as the difference angles (Q 11 and ⁇ ,,') are to be compared.
• the lengths L are now plotted against the lengths U for the corresponding pixels in the diagram of FIG. 4 for the two vectors.
• the length Ly is plotted against the length Lj /, ie the x-section of the point in the diagram is determined by Ly, the y-section by Ly '.
• the resulting point has the coordinates (Ly, Lj /).
• angles ⁇ are plotted against the angles ⁇ 'in a second diagram according to FIG.
• the angle ⁇ y is plotted against the angle ⁇ y ', i. the resulting point in the diagram has the coordinates ( ⁇ y, ⁇ y '). This procedure is performed for all the pixels in the two sets of images, yielding two point clouds in the two diagrams of FIGS. 4 and 5.
• a characteristic straight line is laid through each of the two point clouds, or in each of the two point clouds, the respectively contained points are fitted or approximated with a straight line.
• the slopes and the abscissa sections of the two obtained straight lines, in particular the vector formed with these values, are then used as a measure for the evaluation of the similarity of the two object images. In this way one obtains a vector with numerical values or at most a single number, if the length of this vector is taken as a measure of the similarity between the two objects.
• the vector formed by the two gradients and the two abscissa sections of the straight lines in normal form is normalized or standardized by standard methods or the length of the normalized vector is determined and the number obtained as a measure of the similarity of the object images to be compared, especially the actual image and the target image, is considered. It can further be provided that the two slopes and the two abscissa sections are weighted and / or inverted such that their maximum value is 1.
• the corresponding images of the image sets correspond not only to the image areas but also to the wavelength ranges for which they were taken.
• the images to be compared or the corresponding images should have the same spatial resolution, otherwise a corresponding conversion of the image coordinates would have to take place in order to be able to compare corresponding images or congruent image regions.
• the selection or the creation of congruent image areas is carried out in particular in a preprocessing of the images to be compared with one another. It is advantageous if, for the purpose of creating corresponding identical image areas, the object images or the images of the image sets are subjected to preprocessing, e.g. a rotation, equalization, resizing and / or displacement.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Multimedia (AREA)
• General Physics & Mathematics (AREA)
• Physics & Mathematics (AREA)
• Evolutionary Computation (AREA)
• General Health & Medical Sciences (AREA)
• Medical Informatics (AREA)
• Software Systems (AREA)
• Databases & Information Systems (AREA)
• Computing Systems (AREA)
• Artificial Intelligence (AREA)
• Health & Medical Sciences (AREA)
• Image Analysis (AREA)

Applications Claiming Priority (2)

Publications (1)

Family

ID=39323058

Family Applications (1)

Country Status (2)

Cited By (3)

Family Cites Families (3)

• 2007
  • 2007-01-04 AT AT0002207A patent/AT505885B1/de not_active IP Right Cessation
• 2008
  • 2008-01-03 WO PCT/AT2008/000001 patent/WO2008080185A2/fr active Application Filing

Also Published As

Similar Documents

Legal Events