METHOD OF CLASSIFYING THE AESTHETIC QUALITY OF PLANAR MATERIALS, AND SYSTEM FOR AUTOMATICALLY SORTING PLANAR MATERIALS USING SUCH A METHOD
TECHNICAL FIELD
The present invention relates to a method of classifying the aesthetic quality of planar materials, and to a system for automatically sorting planar materials using such a method.
The present invention may be used to advantage, though not exclusively, for automatically sorting and classifying the aesthetic quality of planar materials such as wood, ornamental stone tiles, ceramic tiles or similar, to which the following description refers purely by way of example .
BACKGROUND ART
As is known, planar materials, such as tiles, are currently sorted (as a function of aesthetic quality) manually by a sorter, who removes the tiles off a conveyor belt and arranges them in stacks, each containing tiles of the same aesthetic quality.
More specifically, the sorter performs a first
operation in which he inspects the surface aesthetic characteristics of each tile (e.g. grain or staining) and compares them with those of reference tiles, each representing a given quality class e.g. first, second or reject.
The sorter then performs a second operation in which, on the basis of the above comparison, he determines the quality class of the tile and places it on a stack of tiles all of the same class (first, second, third, etc. ) .
The above method of sorting planar materials as a function of aesthetic quality obviously has the drawback of involving a good deal of time-consuming manual labour at the final production stage, thus increasing the production cost of the tiles.
Moreover, quality classification varies depending on the sorter, i.e. is highly subjective.
In other words, sorting and classification depend to a large extent on the personal opinion of the sorter, the negative effect of which, from the commercial standpoint, is obviously that of failing to guarantee consistent aesthetic quality of the tiles.
DISCLOSURE OF INVENTION
It is therefore an object of the present invention to provide a method of classifying the aesthetic quality of planar materials, designed to eliminate the aforementioned drawbacks .
According to the present invention, there is
provided a method of classifying the aesthetic quality of planar materials, as claimed in Claim 1.
It is a further object of the present invention to provide a system for automatically sorting planar materials using such a method, and designed to eliminate the aforementioned drawbacks.
According to the present invention, there is provided a system for automatically sorting planar materials, as claimed in Claim 7. BRIEF DESCRIPTION OF THE DRAWINGS
A non-limiting embodiment of the present invention will be described by way of example with reference to the accompanying drawings, in which:
Figure 1 shows a schematic view of a system for automatically sorting planar materials in accordance with the teachings of the present invention;
Figure 2 shows a flow chart of a method of classifying the aesthetic quality of planar materials implemented by the Figure 1 system; Figure 3 shows, by way of example, a first parameter range which may be used in the Figure 2 method of classifying the aesthetic quality of planar materials;
Figure 4 shows, by way of example, a second parameter range which may be used in the method of classifying the aesthetic quality of planar materials in accordance with the teachings of the present invention;
Figure 5 shows a flow chart of part of the method of classifying the aesthetic quality of planar materials
implemented by the Figure 1 system.
BEST MODE FOR CARRYING OUT THE INVENTION
Number 1 in Figure 1 indicates as whole a system for automatically sorting planar materials. System 1 may be used to advantage, though not exclusively, for automatically sorting planar materials such as wood, ornamental stone tiles 2, ceramic tiles 2 or similar, to which the following description refers purely by way of example . More specifically, system 1 provides for sorting tiles 2 as a function of the aesthetic quality class CL of the surface material of each tile 2.
System 1 comprises a conveyor belt 3 for feeding each tile 2 in a given direction; and a loading device 4 located upstream from conveyor belt 3 and for transferring each tile 2 from a stack 16 of tiles 2 on to conveyor belt 3 (with the surface of the tile visible) .
System 1 also comprises a lighting device 6 located over conveyor belt 3 to illuminate the surface of tiles 2 on conveyor belt 3; and an image acquisition device 7 located next to lighting device 6 to pick up an image of the surface of each tile 2 , and for supplying an information signal INF coding the surface image of tile 2 in matrix format . System 1 also comprises a central control device 8 for coordinating and controlling operation of system 1.
More specifically, central control device 8 receives and processes the information signal INF from image
acquisition device 7, and supplies a control signal COM coding a specific aesthetic quality class CL of tile 2.
System 1 also comprises a sorting device 9 located downstream from conveyor belt 3 and for selectively directing each tile 2 in a given direction relating to a specific class CL of tile 2 according to the control signal COM from central control device 8.
System 1 also comprises a stacking machine 10 for placing each tile 2 on a respective stack 11 comprising a number of tiles 2 of the same class CL; and a number of conveyor lines 12 interposed between sorting device 9 and stacking machine 10, and each for receiving each tile 2 directed by sorting device 9 and for feeding it to the respective stack 11 comprising tiles 2 of the same class CL.
Central control device 8 is of known type, and may advantageously be defined by a computer comprising a memory 13, and a microcontroller 14 for coordinating loading, classification and sorting of tiles 2. More specifically, central control device 8 supplies loading device 4 with a signal CON for controlling loading of tiles 2 on to conveyor belt 3.
In more detail, loading device 4 may advantageously be defined by a robot arm operated by central control device 8 to load tiles 2 appropriately on to conveyor belt 3, which feeds tiles 2 to sorting device 9 at a speed controlled by central control device 8.
Sorting device 9 is interposed between the output of
conveyor belt 3 and the inputs of conveyor lines 12, and may be defined by a deflector operated by central control device 8 to selectively direct each tile 2 to a given conveyor line 12 for feeding tile 2 to the stack 11 relative to a given aesthetic quality class CL.
Central control device 8 also controls stacking machine 10, which is located downstream from conveyor lines 12 and may advantageously be defined by a robot arm by which central control device 8 coordinates the loading of each tile 2 from a given conveyor line 12 on to a respective stack 11.
Image acquisition device 7 may advantageously be defined by a digital television camera operating in the visible spectrum to acquire a surface image of each tile 2 in matrix format, and to supply central control device 8 with information signal INF containing information relative to each pixel in the surface image of tile 2.
More specifically, for each pixel in the image, image acquisition device 7 supplies a vector VET containing a number of values, each indicating a component of the colour (e.g. red, green, blue) of the pixel .
More specifically, in information signal INF, the surface image of tile 2 is coded in a three-dimensional matrix format having a first and second dimension defined by the number of pixels representing a first and second side of the image respectively, and a third dimension defined by the number of vector VET values.
In other words, in addition to colour information, the three-dimensional matrix also comprises the position in space of each pixel in the image .
Central control device 8 cooperates with image acquisition device 7, and performs a first numeric processing operation on the three-dimensional matrix image to calculate a predetermined number of parameters
Pjc and functions Fjc(Pjc) representing the aesthetic characteristics of the surface material of tile 2. Central control device 8 also performs a second processing operation to determine whether parameters Pjc and/or functions Fjc (Pjc) have a predetermined relationship with a number of reference parameters PjR and functions FjR(PjR) relating to each class CL and memorized in memory 13.
Central control device 8 also codes in the signal
COM the class CL of tile 2 as function of the above check, and operates sorting device 9 to direct tile 2 to a given conveyor line 12, i.e. to the stack 11 relating to the aesthetic quality class CL of tile 2.
Central control device 8 automatically sorts tiles 2 using the classification method shown in Figure 2 and described below.
The method of classifying the aesthetic quality of tiles 2 commences with a start block 100.
Block 100 is followed by a block 110 (described in detail later on) in which aesthetic quality classes CL are characterized'.
More specifically, block 110 selects a method of analysis (of the surface of tile 2) by which to establish a number of parameters Pj and functions Fj(Pj) relating to the surface properties of tile 2 and which together characterize the aesthetic quality class CL of tile 2.
Block 110 also processes the three-dimensional matrix images of a predetermined number N,^ of reference tiles relating to a given class CL, so as to obtain and store in memory 13, for each class CL, a range of parameters Pj and functions Fj(Pj) characterizing each class CL.
These parameters Pj and functions Fj(Pj) are hereinafter referred to as reference parameters PjR and reference functions FjR(PjR). For example, the range of reference parameters PjR characterizing each class CL ranges between a minimum value PJHmin and a maximum value j^x defining a one- dimensional interval in space (Figure 3) ; whereas the range of reference functions FjR(PjR) falls within a two- dimensional space AR (Figure 4) defined, for example, by the envelope of reference functions FjR(PjR).
In connection with the above, it should be pointed out that the analysis method selected in block 110 may comprise, for example, statistical processing employing as reference parameters PjR and/or functions FjR(PjR) the statistical distribution (average and covariance) of the image pixel colour component values (vector VET) ; or may comprise stationary geostatistical analysis employing a
variogram, spatial covariance, or generalised covariance as reference parameters PjR and/or functions FjR(PjR); or may comprise morphological processing employing reference parameters PjR and/or functions FjR(PjR) indicating dilation, morphological openings and closures, or granulometry of optical-phase.
In connection with the above, it should also be pointed out that characterization is performed using a number of reference parameters PjR and functions FjR(PjR) (e.g. 10-15 parameters) sufficient to characterize tile
2.
Reference parameters PjR and functions FjR(PjR) are therefore used, or not, depending on the type of surface material of tile 2. Block 110 is followed by a block 120 which acquires the three-dimensional matrix image of the surface of a tile 2.
More specifically, in block 120, central control device 8 receives, by means of signal INF from acquisition device 7, the three-dimensional matrix image of the surface of the tile 2 on conveyor belt 3.
Block 120 is followed by a block 130, in which central control device 8 processes signal INF coding the three-dimensional matrix of the surface of tile 2, i.e. processes the position and vector VET of each pixel and calculates the parameters Pjc and/or functions Fjc(Pjc) characterizing the tile 2 to be classified.
The calculated parameters Pjc and/or functions
FJc(pJc) are obviously those determined using the analysis method selected in block 110, i.e. correspond to reference parameters PjR and/or functions FjR(PjR) .
Block 130 is followed by a block 140 which starts a comparison cycle of the calculated parameters Pjc and/or functions Fjc(Pjc) and the reference parameters PjR and/or functions FjR(PjR) stored in memory 13.
More specifically, block 140 initializes a class indicator Ic=l unequivocally identifying each class CL(IC) .
More specifically, in block 140, central control device 8 reads the reference parameters PjR and/or functions FjR(PjR) relative to a first class CL. For example, when Ic=l, class CL may be the one indicating the first-quality class CL(1).
Block 140 is followed by a block 150, in which central control device 8 compares the calculated parameters Pjc and/or functions Fjc(Pjc) with the reference parameters PjR and/or functions FjR(pjR) relating to class CL(IC) .
More specifically, block 150 determines whether the values of the calculated parameters Pjc and/or functions
FJc(p c) fall within the ranges defined by respective reference parameters PjR and/or functions FjR(PjR) relating to class CL(IC) .
For example, a check is made to determine whether each calculated parameter Pjc (Figure 3) satisfies the equation PjRmin<pJc <pJ, Rmaχ ar whether each calculated
function Fjc(Pjc) (Figure 4) satisfies the equation Fjc(Pjc)DAR, i.e. whether the calculated function Fjc(Pjc) falls within multidimensional space A-,.
If the value of at least one of the calculated parameters Pjc and/or functions Fjc(Pjc) is outside the ranges defined by respective reference parameters PjR and/or functions FjR(PjR) relating to class CL(IC), block 150 is followed by a block 160, which determines whether the class indicator Ic has reached a maximum value ICMAX indicating the last class CL (ICMAX) (e.g. reject class) stored in memory 13.
Conversely, if the values of calculated parameters Pjc and/or functions Fjc(Pjc) are all within the ranges defined by respective reference parameters PjR and/or functions FjR(PjR) relating to class CL(IC), block 150 is followed by a block 170, which assigns class CL(IC) to tile 2.
It should be pointed out that, in block 170, central control device 8 supplies a control signal COM coding the deflection to be imparted by sorting device 9 to tile 2 to direct it to the conveying line 12 relating to the assigned class CL(IC).
If
block 160 is followed by a block 180, in which the value of indicator I
c is increased according to the equation I
C=I
C+1, and central control device 8 reads the reference parameters Pj
R and/or functions Fj
R(Pj
R) relating to another class CL. For example, when I
c=2, class CL may be
' the one indicating the second-quality
class CL (2 ) .
If IC>ICMAX, block 160 is followed by a block 190, which performs a "discriminatory analysis" of calculated parameters Pjc and/or functions Fjc(Pjc). Block 180 is followed by block 150, in which the above comparison is repeated between calculated parameters Pjc and/or functions Fjc(Pjc) and the ranges defined by reference parameters PjR and/or functions FJ (PJR) relating to class CL(IC+1) following class CL(IC) .
In block 190, central control device 8 performs a discriminatory analysis of tile 2, i.e. classifies tile 2 on the basis of a perfected statistical processing operation. More specifically, the calculated parameters Pj
c and/or functions Fj
c(Pj
c) are processed jointly to obtain a known multivariate function which, applied to the values of calculated parameters Pj
c and/or functions
indicates the "match distance" of tile 2 from each class CL(I
C) .
At this point, central control device 8 assigns tile 2 the best-match class CL(IC), i.e. the class CL(IC) at a minimum "match distance" from tile 2.
In block 190, as in block 170, central control device 8 supplies a control signal COM coding the deflection to be imparted by sorting device 9 to tile 2 to direct it to the conveying line 12 relating to the assigned class CL.
Block 190 is followed by a block 200, in which characterization of the class CL most closely matching the last tile 2 classified by discriminatory analysis is updated. More specifically, block 200 updates the reference parameters PjR and/or functions FjR(PjR) of the best-match class CL(IC) determined in block 190.
More specifically, in block 200, the ranges of reference parameters PjR and/or functions FjR(PjR) of the best-match class CL are "widened" to also include the calculated parameter Pjc and/or function Fjc(Pjc) values which were formerly outside the ranges.
Blocks 170 and 200 are followed by a block 210, which determines whether another tile 2 is to be classified.
More speci ically, if central control device 8 does not receive another image by means of information signal INF, block 210 is followed by a block 220, which terminates the classification procedure. Conversely, block 210 goes back to block 130 which performs another classification cycle on another tile 2.
Figure 5 shows in detail the characterization step performed in block 110 in Figure 2.
Firstly, a block 300 starts the characterization of classes CL.
It is assumed the tiles 2 being characterized all belong to the same class CL, i.e. all belong, for example, to first-quality class CL(1).
Block 300 is followed by a block 310, which enters a number of information items relating, for example, to the type of surface material of tiles 2, the class CL of reference tiles 2, and the number N,^. of reference tiles to be supplied to system 1 to characterize class CL.
Block 310 is followed by a block 320, which determines the best processing method according to the type of material entered.
More specifically, block 320 determines the reference parameters PjR and/or functions FjR(PjR) by which to characterize the material and respective aesthetic quality class CL entered in block 310.
Block 320 is followed by a block 330, which initiates a counter N=l indicating the number of reference tiles 2 supplied to system 1.
Block 330 is followed by a block 340, in which acquisition device 7 acquires an image of the reference tile surface and transmits it, in matrix format by means of information signal INF, to central control device 8. Block 340 is followed by a block 350, in which central control device 8 processes the three-dimensional matrix image to obtain the values of reference parameters PjR and/or functions FjR(PjR) .
Block 350 is followed by a block 360, in which central control device 8 stores the values of reference parameters PjR and/or functions FjR(PjR) in memory 13, thus updating and defining the ranges of reference parameters PjR and/or functions FjR(PjR) .
Block 360 is followed by a block 370, which determines whether the supplied reference tile counter N is less than or equal to the maximum number N^ entered in block 310 (N<=NMAX) . If reference tile counter N is less than or equal to the maximum number N,^ (N<=NMAX) , block 370 is followed by a block 380 which increases counter N.
Block 380 is followed by block 340, which acquires another surface image of another reference tile 2. If reference tile counter N is greater than the maximum number N^ (N>NMAX) , block 370 is followed by a block 400, which terminates the characterization of class CL.
In connection with the above, it should be pointed out that the characterization step is repeated for a number of classes CL, e.g. for the first-quality, second- quality and reject classes.
More specifically, for each class CL, the surface images of a predetermined number of tiles similar to class CL are processed, and the ranges of the reference parameters PjR and/or functions FjR(PjR) characterizing class CL are determined and stored each time in memory 13.
Operation of system 1 is obvious from the above description, and therefore requires no further explanation.
The system described for automatically sorting tiles and planar materials has the advantage of involving no
labour, thus reducing production cost.
The system described for automatically sorting tiles and planar materials also has the advantage of monitoring correct/incorrect performance of the production process on the basis of information relative to the aesthetic quality of the tiles and planar materials.
Finally, the system described for automatically sorting tiles and planar materials also has the advantage of providing for unequivocal, repeatable classification, thus ensuring consistent quality of the tiles and planar materials supplied to customers.