EP1697896A2 - Procede de cartographie des defauts sur un cuir - Google Patents
Procede de cartographie des defauts sur un cuirInfo
- Publication number
- EP1697896A2 EP1697896A2 EP04816489A EP04816489A EP1697896A2 EP 1697896 A2 EP1697896 A2 EP 1697896A2 EP 04816489 A EP04816489 A EP 04816489A EP 04816489 A EP04816489 A EP 04816489A EP 1697896 A2 EP1697896 A2 EP 1697896A2
- Authority
- EP
- European Patent Office
- Prior art keywords
- phase
- leather
- image
- likelihood ratio
- defects
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 230000007547 defect Effects 0.000 title claims abstract description 125
- 239000010985 leather Substances 0.000 title claims abstract description 92
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000013507 mapping Methods 0.000 title claims abstract description 17
- 238000004458 analytical method Methods 0.000 claims abstract description 46
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 29
- 238000012545 processing Methods 0.000 claims abstract description 21
- 238000012896 Statistical algorithm Methods 0.000 claims abstract description 8
- 238000001514 detection method Methods 0.000 claims description 34
- 238000004364 calculation method Methods 0.000 claims description 15
- 101000859758 Homo sapiens Cartilage-associated protein Proteins 0.000 claims description 11
- 101000916686 Homo sapiens Cytohesin-interacting protein Proteins 0.000 claims description 11
- 101000726740 Homo sapiens Homeobox protein cut-like 1 Proteins 0.000 claims description 11
- 101000761460 Homo sapiens Protein CASP Proteins 0.000 claims description 11
- 101000761459 Mesocricetus auratus Calcium-dependent serine proteinase Proteins 0.000 claims description 11
- 102100024933 Protein CASP Human genes 0.000 claims description 11
- 230000004807 localization Effects 0.000 claims description 9
- 238000012512 characterization method Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 230000005540 biological transmission Effects 0.000 claims description 3
- 239000003086 colorant Substances 0.000 claims description 2
- 230000000875 corresponding effect Effects 0.000 description 23
- 238000005520 cutting process Methods 0.000 description 8
- 238000004519 manufacturing process Methods 0.000 description 5
- 230000007847 structural defect Effects 0.000 description 5
- 238000013459 approach Methods 0.000 description 4
- 210000003462 vein Anatomy 0.000 description 4
- 230000037303 wrinkles Effects 0.000 description 4
- 241001465754 Metazoa Species 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 241001674048 Phthiraptera Species 0.000 description 2
- 230000004075 alteration Effects 0.000 description 2
- 238000012217 deletion Methods 0.000 description 2
- 230000037430 deletion Effects 0.000 description 2
- 230000006866 deterioration Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 210000004209 hair Anatomy 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000003908 quality control method Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 238000009987 spinning Methods 0.000 description 2
- 206010063409 Acarodermatitis Diseases 0.000 description 1
- VYZAMTAEIAYCRO-UHFFFAOYSA-N Chromium Chemical compound [Cr] VYZAMTAEIAYCRO-UHFFFAOYSA-N 0.000 description 1
- 208000032544 Cicatrix Diseases 0.000 description 1
- 241000447727 Scabies Species 0.000 description 1
- 208000002474 Tinea Diseases 0.000 description 1
- 241000893966 Trichophyton verrucosum Species 0.000 description 1
- 230000002378 acidificating effect Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 239000001055 blue pigment Substances 0.000 description 1
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 238000011960 computer-aided design Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000004043 dyeing Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000002513 implantation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 244000045947 parasite Species 0.000 description 1
- 230000024241 parasitism Effects 0.000 description 1
- 229920001296 polysiloxane Polymers 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 208000005687 scabies Diseases 0.000 description 1
- 231100000241 scar Toxicity 0.000 description 1
- 230000037387 scars Effects 0.000 description 1
- 238000003307 slaughter Methods 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/44—Resins; Plastics; Rubber; Leather
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30124—Fabrics; Textile; Paper
Definitions
- the present invention relates to a method for qualifying a leather using a mapping of the defects of the leather by image processing, as well as a device making it possible to implement this method.
- leather will be used to designate skin or leather whatever its stage of manufacture and its mode of presentation: whole, segmented or cut in thickness.
- Leather the raw material of tanning and tanning, is heterogeneous in its structure and in its surface appearance.
- Leather has a natural grain due to the implantation of the animal's hairs, the homogeneity of the grain depending on their size and distribution. Depending on the species, leathers have different characteristics of size, suppleness, fineness of the grain.
- defects can be caused during the stages: -of breeding, due to the sanitary state, parasitism, mechanical or genetic lesions; - slaughter, due to the transport and restraint of animals, the skin of hides; conservation, due to the implementation of the process, and the storage conditions; - tanning-tanning, due to mechanical operations on the machine or to the effects of chemicals.
- the diversity of the nature of the defects implies a constant adaptation of the leather transformation process, certain defects modifying the appearance of the finished leather, others altering the structural properties of the leather itself.
- the position of the defect on the leather and the possibility of integrating it into the assembly margins at the level of the manufacture of the finished product must be taken into account to assess the value of the leather.
- the detection of faults is therefore of decisive interest, both for the tanner, for quality control, optimization of his process according to each skin, and the establishment of fault histories, as for manufacturer of finished leather products, for quantitative and qualitative sorting, maximizing the use of the surface for cutting leathers.
- wet-blue designates a leather after chrome tanning, giving it a blue color. The leather at this stage is wet.
- Cutting leather is a step requiring experience and optimizing the placement of parts on the leather.
- the permanent compromise that the cutter must perform begins to be computerized with automatic placement software, coupled with digital cutting tables.
- the phase of entering faults and quality zones remains manual, which penalizes the overall profitability and productivity.
- Attempts to automate the detection of faults on leathers were made, without success however, because these approaches did not take into account either an image quality sufficient to highlight the faults, or the highly noisy appearance of the images obtained, ie the lack of reproducible geometric structures on the leathers.
- these solutions do not allow structural defects to be revealed.
- Doganzic A the disclosure of Doganzic A.
- P628 to 635 describes a treatment method of image comprising at least one phase consisting in applying to a picture a statistical algorithm of the generalized likelihood ratio, making it possible to calculate, the likelihood ratio of the presence of a defect in a first region, relative to a reference noisy region and thus obtaining a probability image indicating for each pixel the value of this likelihood ratio.
- the solution proposed in this document does not allow local variations in noise on a skin to be taken into account.
- the aim of the present invention is to provide a method as well as an automatic device for mapping faults on a leather enabling these technical problems to be resolved by processing an image of sufficient quality to highlight the faults and taking into account the noisy appearance of the images obtained and the absence of reproducible structure, and moreover making it possible to detect both the appearance defects and the structural defects, taking into account the local variations in the sound effects of the image.
- the subject of the present invention is a method of qualifying a leather using a mapping of the leather defects, implementing a mapping device, comprising means for acquiring a digital image of the leather, and a image processing obtained comprising at least one phase consisting in applying to the image of the leather, after possible preliminary steps, a statistical algorithm of the generalized likelihood ratio or GLRT, making it possible to calculate, in a window analysis centered on a given pixel of the image and comprising a mask with at least two regions, the likelihood ratio of the presence of a defect in a first region relative to a second region and thus obtaining an image of probability indicating for each pixel the value of this likelihood ratio.
- the method comprises a phase for detecting linear type faults, and a phase for detecting circular type faults.
- the phase of detection of the linear defects uses a statistical algorithm of the generalized likelihood ratio or GLRT in its linear form, that is to say by using a form of mask for the analysis window comprising a central region in the form of a band passing through the center of the analysis window and oriented at an angle relative to the abscissa axis, and two lateral regions and situated on either side of the central region, the two regions lateral and completing the analysis window, by carrying out several calculations of the likelihood ratio according to several values of the angle and by conserving for each pixel: - the most important value of the likelihood ratio in a probability image and, - the value of the angle which gave the most important value of the likelihood ratio in an orientation plane.
- the circular defect detection phase uses a statistical algorithm of the generalized likelihood ratio or GLRT in its circular form, that is to say by using a mask form for the analysis window comprising a central region concentric with the analysis window, rectangular in shape and a region completing the analysis window, by carrying out several likelihood ratio calculations by deforming the central region to increase its size, in the direction of the abscissas and ordinates while keeping for each pixel : - the most important value of the likelihood ratio in a probability image, and - the dimensions on the abscissa and on the ordinate of the central region having given the most important value of the likelihood ratio stored respectively in two dimensioning images on the abscissa and on the ordinate.
- the method comprises a phase for locating circular defects, comprising the steps consisting in: - searching for local maxima of the likelihood ratio in the clusters or sets of points of the image having a likelihood ratio greater than a threshold value, - define a neighborhood around this local maximum, the dimensions of which on the abscissa and on the ordinate are determined using values coming from a detection phase, and - find a second local maximum of the likelihood ratio in the residual part of the cluster, if part of the cluster is outside the neighborhood thus defined, - a neighborhood is also defined around this second cluster whose dimensions on the abscissa and on the ordinate are determined using values coming from a detection phase.
- the values coming from a detection phase making it possible to determine the dimensions on the abscissa and on the ordinate of a neighborhood around this local maximum are the values corresponding to the position of the maximum in the dimensioning images.
- the method includes a phase for locating the linear defects, comprising the steps consisting in: - searching for local maxima of the likelihood ratio in the clusters or sets of points of the image having a likelihood ratio greater than a threshold value, - define a neighborhood around this local maximum whose dimensions correspond to those of the phase region, this neighborhood being oriented using the value corresponding to the position of the maximum in the orientation plane and, - find a second local maximum of the likelihood ratio in the residual part of the cluster, if a part of the cluster is outside the neighborhood thus defined, - a neighborhood also being defined around this second cluster whose dimensions correspond to those of the phase region, this neighborhood being oriented using the value corresponding to the position of the maximum in the plane of d ' orientation.
- the phase of locating the linear defects also includes a step consisting in merging the close neighborhoods, and to isolate the neighborhoods of surface greater than a threshold given as a region of altered texture.
- the method comprises a characterization phase comprising a step consisting in determining the precise shape of the contour of the resolved defects using an active polygonal statistical contour algorithm or CASP.
- the characterization phase comprises a step consisting in calculating geometric and photometric parameters characteristic of each defect.
- the method comprises an identification phase consisting in classifying the defects into categories.
- the method includes a phase of selecting the faults to be taken into account by a user.
- the method comprises a classification phase of the leather, making it possible to assign a level of quality, as a function of the position, the type and the surface of the defects on the leather.
- the characteristic parameters used in the calculation of the likelihood ratio are the mean and the standard deviation.
- the model chosen to represent the probability densities of the pixel intensity is the Gaussian model.
- the method also comprises a phase consisting in carrying out a geometric analysis of the leather to identify the quality zones and the load-bearing lines of elasticity of the leather.
- the method further comprises a phase consisting in determining a segmentation of the leather into regions of homogeneous hues using the “Division and Fusion” algorithm using a Tetra-tree.
- the method comprises several passes in the phase of detection of circular defects, corresponding to different sizes of analysis window.
- the invention also relates to a mapping device for implementing the method described above, comprising means for conveying the leathers, making it possible to route them to an acquisition station comprising a digital camera as well as lighting means, characterized in that the lighting means comprise a light source by transmission through the part of the conveyor constituting the leather support and a light source by reflection, the angle of the direction of the light source by reflection relative to the plane of the leather being between 15 ° and 35 °.
- the device comprises calculation means allowing software image processing.
- the device includes dedicated electronic calculation means for performing image processing.
- FIG. 1 is a schematic side view of the device according to the invention.
- FIG. 2 is a representation of the flow diagram of the image processing carried out.
- FIG. 3 represents an image of the support and of the outline of a leather.
- FIG. 4 represents an image identifying the carrying lines of the lender's directions.
- FIG. 5 represents an image identifying the quality zones of the leather.
- Figure 6 shows schematically the principle of division into parcels.
- Figure 7 represents a data structure in Tetra-tree.
- Figure 8 illustrates the generalized maximum likelihood ratio algorithm.
- Figures 9 to 12 show analysis window forms for a maximum generalized linear likelihood ratio algorithm.
- FIG. 13 represents the evolution of the form of analysis window for an algorithm of the maximum of the generalized circular likelihood ratio.
- FIG. 14 represents neighborhoods used in determining the location of resolved circular faults.
- Figures 15 to 17 show the successive stages of the polygonal statistical active contour algorithm.
- Figure 18 is a portion of an image of leather.
- FIG. 19 is a portion of a probability image, in the context of a linear GLRT, corresponding to the processing of the image portion of FIG. 18.
- FIG. 20 is a portion of a probability image, in the frame of a circular GLRT, corresponding to the processing of the image portion of FIG. 18 with a mask for a large defect.
- FIG. 21 is an image portion representing polygonal neighborhoods V for large circular resolved defects.
- FIG. 22 is an image portion representing the contours of large resolved defects obtained by a CASP algorithm.
- FIG. 25 is a portion of the binary image of the defects preserved, corresponding to the image portion of FIG. 18.
- the device 2 for mapping the defects represented in FIGS. 1 and 2, comprises three main elements: the conveying module 3, the image acquisition station 4 and calculation means 5 allowing the execution of the image processing software 6.
- the conveying module 3 comprises a translucent mat 7 driven on rollers 8, the route of this mat 7 forming a first plane 9 inclined at 45 ° relative to the vertical, then a horizontal portion 12 separated from the first inclined plane 9 by a roller 8 located at the upper edge of the plane 9, this horizontal portion 12 corresponding to the part of the mat 7 located at the acquisition station 4, then a second plane 10 inclined at 45 ° with respect to the vertical, symmetrical with the first plane 9 with respect to the acquisition station 4, separated from the portion horizo ntale 12 by a roller 8 located at the upper edge of the plane 10.
- the mat 7 forms a drive support in a direction of travel P for the leathers 13 which can be deposited on the first inclined plane 9 and then driven towards the acquisition station 4 then towards plane 10 where they can be unloaded from the mat 7.
- the inclination of planes 9 and 10 facilitates the loading and unloading of leathers 13, making it possible to view them in a reduced space.
- the mat 7 is silicone-coated, in order to prevent the leather 13 from sliding on the inclined planes 9 and 10.
- the acquisition station 4 comprises an articulated strip 14 located above the horizontal portion 12 of the mat 7 making it possible to press the leather 13 and to lay it flat before the acquisition of an image 16 by a digital camera 15 situated above the horizontal portion 12 of the belt 7 downstream from the strip 14 in the direction of travel P and directed vertically downwards.
- a first light source by reflection 17 of the leather 13, is located above the horizontal portion 12 of the mat 7, downstream of the camera 15, emitting diffuse light, directed at an angle angle of 25 ° relative to the plane of the leather 13. This particular orientation makes it possible to bring out both the structural defects and the appearance defects of the leather 13.
- the source 17 is covered by a cover 18 so as to avoid direct lighting of the camera 15 by this source 17 and a disturbance of the image 16 obtained.
- a second source of transmission light 19 arranged under the translucent mat 7, facing the camera 15, makes it possible to remove the shadow cast and to increase the contrast between the leather 13 and the mat 7 in the image 16.
- the camera 15 has a resolution greater than 6,000 pixels and a 35 mm lens.
- the camera 15 captures an image of the leather 13 with a discretization of each pixel on 1,024 gray levels. The image obtained is therefore of excellent quality, with a resolution of 0.24 mm per pixel.
- Control means not shown make it possible to control the driving of the leather 13 by the carpet 12 under the camera 15 and the acquisition of the image 16 by the camera 15.
- the camera 15 is a so-called linear sensor camera and comprises a single row of sensor cells each corresponding to a pixel arranged perpendicular to the direction of travel P. The movement over time of the leather 13 relative to this camera makes it possible to obtain a two-dimensional image 16.
- a pre-processing phase is carried out, making it possible to remove the artefacts from the image 16.
- the camera 15 with linear sensor causes artefacts appearing in the image in the form of columns perpendicular to the direction of movement of the leather, these columns appearing due to the difference in sensitivity of the sensors of the camera 15.
- the meadow treatment consists of compensating for these differences in sensitivity.
- An estimate of the average per column of the image processed by the same sensor cell is calculated, then all the estimates of the average are grouped into a curve from which the control points of a Bézier curve are calculated.
- phase a to h of the treatment shown in FIG. 2, allowing the qualification of the leather are described below, these phases are as follows: - a: geometric analysis, - b: segmentation into regions of homogeneous shade, - c: detection of faults, - d: location of faults, - e: characterization of faults, - f: identification of faults, - g: selection by the user, - h: classification, steps a to e constituting more particularly a mapping of defaults.
- the first phase a of the processing of image 16 is a geometric analysis of the leather 13 making it possible to determine the quality zones and the load-bearing lines 27 of the lending direction.
- the direction of lender corresponds to the direction of maximum elasticity of the leather at a given point.
- the quality zones are as follows: • The rump 23 • The collar 24 • The abutment 25 • The flanks 26
- the first phase a of image processing is based on a geometric analysis of the outline of the leather 13. Firstly , an image represented in FIG. 7 of the closed and unitary outline 28 of the leather 13 is obtained from the image of the support 29 of the leather, represented in FIG. 6.
- the support is obtained by thresholding the values of the pixels of image 16, this operation being facilitated by the lighting by the source 19 which accentuates the contrast between the leather 13 and the carpet 7.
- Different characteristic points of the leather 13 are then determined, and in particular the center of gravity 30 of the leather 13, the extremal points of the legs 32, 33, 34, 35, the maximum notches of the sides 36, 37 as well as the minimum and maximum points of the abutment 38, 39 and the collar 40, 42.
- Geometric laws known in the state of the art allow the boundaries of the quality zones 23, 24, 25, 26 to be drawn as well as the lines carrying the lending direction 27 as shown in the figures
- results of this geometric analysis phase a are: - an image 20, an example of which is shown in FIG. 5, which groups the borders of the different quality zones of the leather.
- This image 20 of the quality zones is useful for determining the severity of the defect because it depends on its positioning on the leather
- - an image 22, an example of which is shown in FIG. 5 represents the load-bearing lines 27 of the directions of lending.
- Image 22 is generated for example for the automatic placement of shoe parts.
- a shoe part represented in a computer format coming from a Computer Aided Design or CAD software includes the information of the lender's direction. It is therefore possible to make the lending direction of the leather 13 given by the image 22 coincide with that of the shoe part.
- the second phase b of the processing of the image 16 is a segmentation into regions of homogeneous shade R t of the leather 13.
- This step is an optional step intended for example for cutting parts for a shoe. In this case, it is desirable not to cut a piece on two regions of different colors.
- a filtering of the original image 16 by a filter calculating an average on a mask, of size 15 by 15 pixels in this embodiment, is carried out.
- This first step makes it possible to attenuate the contrast of the small defects and does not influence the result to be obtained because the aim of this step is the detection of regions Rt of extended surface. These Rt regions have a high contrast between them, the quality of the image 16 obtained amplifying this contrast compared to human visualization.
- This phase b implements in a second step a “Division and Fusion” algorithm using a Tetra-tree.
- This algorithm is broken down into two stages.
- the first so-called division step consists of a recursive splitting of the image into plots p
- the second so-called merging step is an iterative process aimed at grouping these plots p into RT regions having common properties, that is to say say, in this mode of implementation, a neighboring shade.
- a region of the image during the division process will be called plot p
- region RT a grouping of plots made during the fusion
- the division consists of a recursive splitting of the image 16 in accordance with a criterion of pixel homogeneity. If the criterion is not satisfied for a plot p of the image, the latter is then divided into four sub-parts of equal sizes and the criterion is again evaluated in each of them. This process is repeated recursively until all the plots p satisfy the given criterion. At the start of the division, the criterion is evaluated on the entire image.
- the division into plots p is represented in FIG. 6, at a given stage of the algorithm.
- the plot pi satisfying the homogeneity criterion is not divided on the contrary, the plot p 2 must be divided because it does not meet the criterion.
- each plot p is assigned to a region R ⁇ .
- each plot represents an RT region.
- Each pair of plots p is then considered.
- the fusion criterion is evaluated on the union of the two RT regions to which these two plots belong. If this is satisfied, the two RT regions are merged into one. The two plots p considered are now attached to this new RT region. The merger is complete when all the couples have been treated.
- the algorithm uses the average pixel intensity as a criterion.
- the division and fusion threshold identical and equal to 10 gray levels. This threshold must be configurable according to the level of quality desired by the customer.
- the minimum dimension on the abscissa and ordinate is fixed at 5 pixels for each parcel p of image.
- phase b is an image 44 showing the regions of homogeneous hue RT.
- the processing then comprises a phase c on the image 16 relating generally to the detection of faults D grouping two phases and c 2 concerning respectively: - ci: the detection of linear shape defects grouping the resolved linear defects DL as well as the regions of altered texture RA, the linear defects corresponding in particular to wrinkles, veins or physical defects of linear shape such as tears or scars, the regions of altered texture R A corresponding for example to an extensive alteration of the leather due to wrinkles, a network of veins, crimps or a concentration of resolved defects and - c 2 : the detection of resolved defects of circular shape or circular defects D c , corresponding in particular to structural defects such as defects caused by parasites such as lice, ringworm, or scabies, or of mechanical origin such as a hole or appearance defects such as stains.
- An analysis window F runs through image 45, as shown in the figure 8.
- the analysis window F is divided into two regions Ri and R 2 of different sizes Ni and N 2 , Ni and N 2 representing the number of pixels present in each region. Let and l 2 be two samples of pixels contained in each of the two regions of the window, I the union of these two samples. For a position of the analysis window F, the following two hypotheses are considered: • Ho: the window F covers a homogeneous area.
- the characteristic parameters of the texture are therefore the same in the two regions Ri and R 2 and are symbolized by ⁇ .
- Hi the window F is positioned on a defect D located in the region R ⁇
- the characteristic parameters of the texture in the two regions Ri and R 2 are therefore different and are equal to ⁇ i and ⁇ 2 .
- the two corresponding decisions are therefore: “ Ho: there is no default.
- Hi a fault is present.
- Neyman-Person strategy makes it possible to build the optimal decision rule, because it gives the best probability of detection for a fixed probability of false alarm. This rule consists in comparing the likelihood ratio ⁇ :
- ⁇ i represents the parameters of standard deviation ⁇ and of average mi over the first region
- ⁇ 2 represents the parameters of standard deviation ⁇ 2 and of mean m 2 over the second region
- ⁇ represents the parameters of standard deviation ⁇ 0 and of average mo over the entire analysis window F.
- the GLRT algorithm is used according to two variants described below: the linear GLRT and the GLRT circular, these two variants differing mainly by the shape of the analysis window chosen.
- linear is used in the phase Ci of detection of resolved defects, linear resolved defects DL OR regions of altered texture RA.
- the circular GLRT is used in phase c 2 of detection of circular resolved defects D c .
- the linear GLRT is characterized by the internal shape or “mask” of the analysis window F which is broken down, as shown in FIGS.
- the angle A1 varies by an angular step of 15 ° between 0 ° and 180 °.
- the size of the analysis window F best suited to the defects of the leathers 13 is a size of 51 by 51 pixels. A smaller size causes the grain of the leather 13 to be detected as a defect.
- a larger size gives too low and too wide probabilities for a defect because the union of the regions R 2 'and R 2 "is too large and can include other faults in the neighborhood, thus reducing the statistical difference between the defect and
- the selected width of the strip forming the region Ri is 5 pixels and the angular step is 15 °.
- the configurations of the mask of the analysis window F are shown in FIGS. 9 to 12 for an angular step of 45 ° and an angular range of 0 ° to 180 ° For each dimension of the analysis window F, a bandwidth forming the region Ri makes it possible to obtain detection of all of the defects.
- the linear GLRT algorithm provides as result: - a probability image 45 as defined above, and - an orientation plane 46 indicating, for each pixel of the original image 16, the value of the angle A1 of the mask having given the highest likelihood ratio ⁇ value.
- - a dimensioning image in x 47 giving for each of the pixels a constant value corresponding to the width of the region R-
- - A dimensioning image in y 48 giving for each of the pixels a constant value corresponding to the thickness of the strip forming the region Ri.
- the circular GLRT makes it possible to determine the presence in each pixel of the image of the leather 13 of a circular type defect D c .
- the analysis window F must have an interior shape or mask adapted to the form of the defect De.
- the window F is therefore divided into a region Ri, rectangular and concentric with the analysis window F, of variable size and dimension, and a region R 2 completing the analysis window F.
- the region Ri is deformed for each position of the window F. Indeed, the greatest value of the likelihood ratio ⁇ is obtained for a given defect D c , when the region Ri has dimensions corresponding to the dimensions of the circular defect D c .
- the region R- is deformed incrementally according to changes E x in x and E y in x shown in FIG.
- the region Ri is defined by the following parameters: - The minimum dimension of the region Ri in x and y, - The maximum dimension of the region Ri x and y, The pitch between two dimension values in x or y of the region R - must be weak when the region Ri is small because a defect D c of a few pixels must have a region Ri very close to its real form to obtain an optimal likelihood ratio ⁇ . Indeed, a small defect D c detected by a poorly adapted mask greatly reduces its probability. In addition, it is likely to find several small defects in a large area.
- a dimension of the region Ri ill-suited influences the probability less because the size of the sample I is larger.
- the step for detecting large faults can therefore be greater.
- three passes are therefore carried out for small, medium, and large region dimensions Ri.
- the step has a different value, small, medium, and large, all these values being between 5 and 10 pixels.
- the maximum size of the region Ri is 130 pixels for the pass of large defects.
- the analysis window F has dimensions equal to the maximum dimensions of the region Ri plus 3 pixels in this exemplary embodiment.
- the circular GLRT algorithm provides as result: - a probability image 45 as defined above. - An orientation plane 46 containing a zero constant value, not further processed.
- - a dimensioning image in x 47 giving for each of the pixels of image 45 an integer value of the dimension in x of the region Ri which gave the value of the likelihood ratio ⁇ the highest.
- a dimensioning image in y 48 giving for each of the pixels of image 45 an integer value of the dimension in y of the region Ri which gave the value of the likelihood ratio ⁇ the highest.
- FIGS. 19 and 20 represent two image portions 45 corresponding to the processing of the image portion 16 of FIG. 18 by a linear and circular GLRT respectively used in phases Ci and C 2 .
- the purpose of the fault location phase d is to obtain a polygonal neighborhood V surrounding the circular and linear faults using the data provided by the detection step.
- the localization phase d brings together two phases di and d 2 relating respectively to: - di: the localization of the defects of linear form gathering the linear resolved defects DL as well as the regions of altered texture R A , and - d 2 : the localization of the resolved circular defects or circular defects D c ,
- the phase d 2 of location of circular faults D c is illustrated in FIG. 14, being limited to the X axis, and described below.
- the probability image 45 is thresholded with the threshold t set by the user to obtain a binary image grouping together clusters of likelihood greater than the threshold, that is to say the set of points of l image 45 having a likelihood ratio ⁇ greater than the threshold.
- a local maximum M of the likelihood ratio ⁇ per cluster is determined.
- a neighborhood V around this local maximum is defined whose dimensions in x and y are determined by using the values corresponding to the position of the maximum in the dimensioning images 47, 48 in x and y. If a part of the cluster is outside the neighborhood V thus defined, a second local maximum M of the likelihood ratio ⁇ is sought in the residual part of the cluster, then also surrounded by a neighborhood V whose the dimensions in x and y are determined using the values corresponding to the position of the second maximum M in the dimensioning images 47, 48 in x and y.
- This approach makes it possible to isolate circular defects D c of large size in the form of a single circular defect D c , and to correctly isolate two small defects De close to one another as shown in FIG. 16.
- phase di is similar to phase d 2 , however in this case the polygonal neighborhood V consists initially of a rectangle whose dimensions, given by the dimensioning planes, correspond to the dimensions of the region Ri of the analysis window F of this phase.
- this rectangular neighborhood V is oriented using the value of the orientation plane for the point corresponding to the maximum probability.
- phase di includes a merging step making it possible to merge the close rectangular neighborhoods V. This merging step consists in running the contour of a neighborhood V through a circular shape of fixed diameter, and in merging the neighborhoods V reached by this shape with the V neighborhood whose circular shape runs around the outline.
- phase e of image processing consists in characterizing the resolved defects D R. This phase includes: - a first step to obtain the precise shape of the contour C of each resolved fault DR, whether this fault is a resolved fault linear D or circular Dc.
- an active polygonal statistical contour algorithm or CASP is used in the first step.
- the input parameters of this algorithm are an image containing the DR defect and a contour C initialized by the neighborhood V.
- the CASP makes it possible to deform this contour C to obtain the precise shape of the contour C of the DR defect.
- the CASP algorithm is used in this embodiment, only for defects D c of large size. Indeed, it is not useful to obtain the precise form of a defect De small.
- this defect De is 4 pixels and the rectangular neighborhood V encompassing it surrounds it perfectly in the broad sense, the approximation of the contour by the rectangular neighborhood V causes a negligible loss of 2 or 3 pixels or about 1mm 2 of the leather 13.
- the principle of CASP, illustrated in FIGS. 15 to 17, is to minimize the energy starting from an initial contour formed by 4 nodes K corresponding to the vertices of the rectangular neighborhood V.
- the nodes K are moved either randomly or in order determined by an adjustable step on the amplitude of the deformations from 1 to 15 pixels, then the energy is calculated. so as the energy decreases, the nodes K are displaced and the energy calculated again. If this becomes constant or if it increases the previous iteration, it provides the best possible segmentation of the contour C of the defect D R.
- the energy calculated in this type of active contour is a function of the statistical parameters of the image, here the variance and the mean.
- the CASP is applied to an initial contour composed only of four nodes K.
- the CASP determines a contour C closer to the defect but composed than of 4 nodes K therefore of imprecise shape.
- 15 knots K are then added so that the shape of the contour C can be refined.
- the CASP algorithm is reiterated to obtain a more precise shape of the C contour.
- This multi-step approach makes it possible to speed up the computation time because the addition of 15 nodes K from the first step for an initial C contour distant from the DR defect real would involve the displacement of 19 knots K over long distances and therefore would require a much higher computation time.
- the step of approximation of the contour is only carried out on 4 K nodes and the 15 K nodes added move only over short distances.
- the parameters calculated for each resolved fault DR are: - geometric parameters, in particular the position of the center of gravity of the fault, circularity, total area of the fault, areas of the faults included in the different areas of quality, perimeter, elongation, stretching, mean and variance of the orientation of the linear defects, - photometric parameters, in particular the mean, variance and local contrast, and - parameters extracted from the GLRT algorithms having highlighted the defect.
- the result of phase e is a set 49 of resolved faults DR, containing the information relating to their contour C and the associated parameters.
- the next phase f is a fault identification phase.
- the purpose of this phase is to assign to each defect D a defined category, a quality zone and a level of visibility.
- the categories defined are three in number: - wrinkle, - vein, and - physical defects. This last category groups together all the structural defects causing an alteration of the flower and the appearance defects.
- the category of a fault is determined according to minimum and maximum limits of a parameter set, consisting of the parameters calculated for each fault in the previous phase.
- the quality zone 23, 24, 25, 26, to which the defect belongs is determined from the position of the defect on the support. Visibility is determined using the contrast parameter defined in phase e for each defect D.
- phase f is an assignment to each defect of a category, a quality zone and a level of visibility .
- the optional phase g consists of a selection by the user of the faults D to be kept in the subsequent phase. In this embodiment, this selection can be made by the user by means of functions for sorting faults on the basis of their category, their visibility and their area.
- the result of phase f is a set 52 of faults to be taken into account.
- Phase h consists of a classification of the defects present in category: - wrinkle, - vein, and - physical defects, and in position in a quality zone: - croupon 23, - collar 24, - abutment 25, and - sides 26. The number, area and percentage of defects present on each part is also calculated.
- the user can fill in a table defining the sorting criteria defining the level of quality or choice of leather. For example, a choice 1, of better quality, is defined by a maximum number of faults in each category per quality zone. From all these data, the system attributes a choice or level of quality to the leather 13. The result of this phase g is therefore the attribution of a level of quality to the skin, as well as an image of the defects preserved, this image can be used for a later cutting phase, as a binary image 53. For example, FIG. 23 represents a portion of image 53 corresponding to a portion of image 16 of FIG. 18. Certain phases image processing 13 are optional depending on the desired use.
- phase b can therefore be omitted in this case.
- Phase a can be simplified in this case. It should be noted that phases a to h do not necessarily follow sequentially.
- the phases di and d 2 for detecting faults D are carried out in several passes, one pass for detecting linear faults, and three passes for detecting circular faults, corresponding to different analysis window sizes F , starting with a large analysis window size F and decreasing this size.
- the phases d of localization and e of determining the contour C must be carried out between each pass to allow the detection of the defects D detected from the image support before making a second pass, so as not to detect the same defect D twice .
- the neighborhood rectangular V is not usable to remove the defect of the image, because the defect DR can be partially outside the neighborhood V.
- the sequence can be, for an image 13: - phase a, - phase b, - phase c 2 of detection of circular faults, for a pass concerning large faults, - phase d 2 of localization of circular faults D c , - phase e of characterization, making it possible to obtain a precise contour for each fault D c , then elimination of this support defect in image 13, - phase c, of detection of linear faults, single pass - phase d, of localization of linear faults, - phase e of characterization of faults, making it possible to obtain a contact precise for each defect D 2 , deletion of the linear defects and of the altered texture zones Ri of the support 29 of the image 13, - phase c 2 of detection of circular defects, for a pass concerning the defects of medium size, - phase d 2 , - ph ase e, then deletion of medium-sized defects D c of the support 29, - phase c 2 , for the pass relating to small-scale defects D c - phase
- the image processing can be carried out, by following the same method, not by image processing software 6, but by a dedicated electronic card according to a technique known elsewhere.
- This second embodiment makes it possible to reduce the computation time necessary to operate the treatment.
- the device will be completed by the addition of a chemical developer, increasing the contrast of the defects causing deterioration of the flower.
- the variations in humidity of the leather at the wet-blue stage generate large variations in color and almost 20% of the defects are not visible at this stage by a sorter such as juicy hairs or lice bites.
- the addition of a chemical developer facilitates the automatic detection of open flower D defects by significantly improving their contrast.
- the blue pigments are only fixed on the parts where the flower is more or less altered.
- the coloring then quickly disappears by itself, the pH of the leather being acidic at this stage.
- the configuration of the device 2 can be modified as a function of the sector chosen, this sector possibly being the tanning-tanning trade or the shoe manufacturers or leatherworkers.
- the tannery acquisition station can be positioned at the inlet or outlet of the wringer.
- the leather is wet, its weight is important and the flattening is not optimal.
- the color is uniform and allows a better identification of the defects D.
- the leather is laid flat, the folds have been crushed and the color regions are different according to the thickness of the leather. Detection is less effective.
- the acquisition station 4 When used for leather goods or the manufacture of shoes or in furniture or in cars, the acquisition station 4 may be offset relative to the production line, or be integrated within the cutting installations themselves. In the latter case, the inclination of the planes of the belt 7 can vary for integration within the cutting installations.
- the invention is not limited to the described embodiment, on the contrary it embraces all variants.
- the lighting angle cu of the leathers 13 can vary.
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR0315464A FR2864668B1 (fr) | 2003-12-26 | 2003-12-26 | Procede de cartographie des defauts sur un cuir |
PCT/FR2004/003377 WO2005069220A2 (fr) | 2003-12-26 | 2004-12-23 | Procede de cartographie des defauts sur un cuir |
Publications (1)
Publication Number | Publication Date |
---|---|
EP1697896A2 true EP1697896A2 (fr) | 2006-09-06 |
Family
ID=34639639
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP04816489A Withdrawn EP1697896A2 (fr) | 2003-12-26 | 2004-12-23 | Procede de cartographie des defauts sur un cuir |
Country Status (3)
Country | Link |
---|---|
EP (1) | EP1697896A2 (fr) |
FR (1) | FR2864668B1 (fr) |
WO (1) | WO2005069220A2 (fr) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107516310A (zh) * | 2017-07-13 | 2017-12-26 | 法视特(上海)图像科技有限公司 | 一种口罩视觉污点检查及排废方法 |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AT509382B1 (de) | 2010-01-18 | 2011-12-15 | Wollsdorf Leder Schmidt & Co Gmbh | Prüfeinrichtung zur bestimmung der qualität von leder |
WO2012060726A1 (fr) | 2010-11-04 | 2012-05-10 | Couro Azul - Indústria E Comércio De Couros, Sa | Procédé pour détecter des défauts dans du cuir |
CN102249086B (zh) * | 2011-03-25 | 2013-01-30 | 江苏连港皮革机械有限公司 | 皮革分拣系统及其使用方法 |
US10297018B2 (en) * | 2017-07-14 | 2019-05-21 | Lear Corporation | Method of digitally grading leather break |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE4216469A1 (de) * | 1992-05-19 | 1993-11-25 | Diehl Gmbh & Co | Einrichtung zum Klassifizieren von Fehlern in Häuten |
-
2003
- 2003-12-26 FR FR0315464A patent/FR2864668B1/fr not_active Expired - Lifetime
-
2004
- 2004-12-23 WO PCT/FR2004/003377 patent/WO2005069220A2/fr active Application Filing
- 2004-12-23 EP EP04816489A patent/EP1697896A2/fr not_active Withdrawn
Non-Patent Citations (1)
Title |
---|
See references of WO2005069220A3 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107516310A (zh) * | 2017-07-13 | 2017-12-26 | 法视特(上海)图像科技有限公司 | 一种口罩视觉污点检查及排废方法 |
CN107516310B (zh) * | 2017-07-13 | 2020-11-10 | 法视特(上海)图像科技有限公司 | 一种口罩视觉污点检查及排废方法 |
Also Published As
Publication number | Publication date |
---|---|
FR2864668A1 (fr) | 2005-07-01 |
WO2005069220A3 (fr) | 2006-10-12 |
FR2864668B1 (fr) | 2006-05-26 |
WO2005069220A2 (fr) | 2005-07-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114359270B (zh) | 基于计算机视觉的汽车发动机油路铜套缺陷检测方法 | |
Bennedsen et al. | Performance of a system for apple surface defect identification in near-infrared images | |
Mehra et al. | Maturity and disease detection in tomato using computer vision | |
CN111667455B (zh) | 一种刷具多种缺陷的ai检测方法 | |
EP3234913B1 (fr) | Procede de detection de defaut sur une surface de pneumatique | |
CN115082683A (zh) | 一种基于图像处理的注塑缺陷检测方法 | |
KR101800058B1 (ko) | 웨브-기반의 재료 내의 불균일성의 고속 처리 및 검출 | |
EP1298588A1 (fr) | Procédé de traitement d'images pour l'extraction automatique d'éléments sémantiques | |
CN111179263B (zh) | 一种工业图像表面缺陷检测模型、方法、系统及装置 | |
US10726535B2 (en) | Automatically generating image datasets for use in image recognition and detection | |
Bandara et al. | Automated fabric defect detection | |
CN113034476A (zh) | 皮革瑕疵检测方法、系统、储存介质及计算机设备 | |
TW202403603A (zh) | 用於偵測在晶圓之成像資料集中的異常的電腦實施方法與使用此方法的系統 | |
EP1697896A2 (fr) | Procede de cartographie des defauts sur un cuir | |
WO2015132531A1 (fr) | Procede d'analyse d'echantillons sedimentaires avec reconnaissance automatique de nannofossiles | |
Villar et al. | A new approach for wet blue leather defect segmentation | |
CN112669300A (zh) | 瑕疵检测方法及装置、计算机设备和存储介质 | |
Kim et al. | Automatic crack detection for concrete infrastructures using image processing and deep learning | |
WO1995016247A1 (fr) | Systeme de surveillance automatique | |
Rezaei et al. | Detecting sunburn in pomegranates using machine vision | |
Hassoon | Classification and Diseases Identification of Mango Based on Artificial Intelligence: A Review | |
CN114612415A (zh) | 基于深度残差网络与多通道图像融合的表面缺陷分类方法 | |
Aleksi et al. | Features extraction and texture defect detection of sawn wooden board images | |
CA2997335C (fr) | Generation automatique d'ensembles de donnees image destinee a la reconnaissance et la detection d'image | |
WO2017005930A1 (fr) | Détection d'objets par traitement d'images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
17P | Request for examination filed |
Effective date: 20060703 |
|
AK | Designated contracting states |
Kind code of ref document: A2 Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU MC NL PL PT RO SE SI SK TR |
|
AX | Request for extension of the european patent |
Extension state: AL BA HR LV MK YU |
|
PUAK | Availability of information related to the publication of the international search report |
Free format text: ORIGINAL CODE: 0009015 |
|
RIC1 | Information provided on ipc code assigned before grant |
Ipc: G01N 33/44 20060101AFI20061025BHEP Ipc: G06T 7/00 20060101ALI20061025BHEP |
|
RIN1 | Information on inventor provided before grant (corrected) |
Inventor name: VIGIER, CEDRIC Inventor name: DARRAGON, JEROME |
|
17Q | First examination report despatched |
Effective date: 20070117 |
|
DAX | Request for extension of the european patent (deleted) | ||
GRAP | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOSNIGR1 |
|
INTG | Intention to grant announced |
Effective date: 20160713 |
|
RAP1 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: CENTRE TECHNIQUE CUIR CHAUSSURE MAROQUINERIE |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
18D | Application deemed to be withdrawn |
Effective date: 20161124 |