IL98614A - Shape sorting particularly for small objects - Google Patents

Abstract

In order to provide an accurate sort into many different classes, objects are passed through a viewing zone and viewed at the same instant by a number of viewers. Signals from the viewers are processed to provide signals representative of the maximum, minimum and mean of the blockiness and symmetry of the object, and the maximum and mean of the convex hull deviance of the object. The signals, together with an edge-breakthrough count are subjected to a linear transformation to provide a normalised shape parameter which is then assigned a value of 0 to 15, for each class being sorted, on the basis of the expected occurrence of the value of the parameter in that class. Each pair of secondary shape parameters so determined is used to derive a decision value from the table specific for that pair of secondary shape parameters. The shape class of the object is ascertained on the basis of a majority vote for all the shape decision values. The tables are "learnt" by the machine by feeding a statistically significant number of samples representative of a number of shape classes through the machine. [GB2246230A]

Description

SHAPE SORTING PARTICULARLY FOR SMALL OBJECTS τηι ο mis "ai? THE APPLICANTS: BERSAN ESTABLISHMENT A LIECHTENSTEIN COMPANY Aeulestrasse 5, 9 90 Vaduz, LIECHSTENSTEI THE INVENTORS: 1. Cook, Nigel Roger 17 Rylston Close, Cox Green, Maidenhead, Berkshire SL6 3HT ENGLAND S. Osgood, Timothy James 73 Haz ley Heath, Hartley Witney, Basingstoke, Hampshire RG27 SNA ENGLAND 3. Ho11away, Stephen Peter 9 Arncliffe, Wildridings, Bracknell, Berkshire RG12 SA ENGLAND . Bowler, Ian William 50A Hilltop View, Lang ley Park, Durham, DH7 9YU ENGLAND M&C FOLIO: 230P61343 WA GDOC: 01481 SHAPE SORTING Background of the Invention This invention relates to a method of ascertaining the classification of the shape of an object based upon deriving a set of values for features representative of the shape of the object.

For example the method may be used with apparatus described in GB-A-2184832. This discloses an apparatus having a viewing zone through which the object is fed. The object is llumlnated whilst in the viewing zone and viewed by a number of fixed electronic viewers spaced around the viewing zone. The images- of the object derived by the viewers are all derived at substantially the same time. The signals from each viewer comprise video pictures which are subsequently normalised (white made true white, black made true black) by selecting a voltage threshold between black and white, and digitised. Digital data for each viewer repreeentive of the edges of the object is then derived by tracking all the points in the image where white (which in this case represents the background, as the objects are viewed dark against a light field) goes to black or black goes to white. Basic shape parameters of the object are then derived for each viewer.

GB-A-2184832 also discloses a method of ascertaining the shape class of an object using the apparatus by processing the average values of the basic shape parameters for all the viewers using a decision tree process. The present application relates to an alternative and more competent method of using the basic shape parameters derived by the viewers.

GB-A-2184832 also discloses a method of detecting ' edge-breakthrough' . The specification can provide apparatus for sorting transparent objects, which give problems due to light shining through the edge zones of the objects. Areas of highly irregular re-entrants in the edge of the image are produced by edge-breakthrough, and are detected as they have a very high rate of change in direction of each incremental length of the edge.

Edge-breakthrough zones are detected in the image and a corrected edge is provided by joining up the ends of the breakthrough zone by a smooth line.

The basic shape parameters derived may be: Approximation of the object to a spherical shape (blocklnese ) -the area of the image is determined and this area is divided by the square of the length of the edge. Images having a high value represent a higher approximation to a spherical shape.

Symmetry in the plane of the viewers- the centroid of each image is determined, . and the image is divided into two parts along a line passing through the centroid. One part of the image is rotated through 180* to superimpose it on the other part and the mismatch area is compared with the overlap area. Images having a substantially higher overlap area than mismatch area have a high degree of mirror symmetry about said plane.

Reentrants or convex hull deviance- re-entrants in the edge may be evaluated by comparing the length of the edge to the length of a line extending all around the edge, like an elastic band, passing straight across any re-entrants. Images having a high difference between actual edge length and ' elastic band' length are s ikey. Alternatively, convex hull deviance may be measured by measuring the distance of a line extending around the edge like an elastic band and the point of the surface furthest from it.

Preferably at least four, more preferably nine viewers are used. Thus) the apparatus of GB-A-2184832 can produce a set tf values of block!nees, symmetr and convex hull deviance for each of the nine viewers. In addition to thejse 27 parameters, an additional i .. parameter, the total count of edge-breakthroughs detected, can bjs derived. i The Invention The invention provides a method of ascertaining the ■ . ■ ( shape class of an obect, comprising: deriving a set bif, primary Bhape parameters representative of the shape of the object, taking a group of two or more of the primary shape parameters to provide coordinates for deriving from a table a decisloi value for said group, the table for said group bein† fixed for all the objects; repeating the process of deriving a decision value for other groups o 'two or more primary shape parameters, using a specific said table for each group; and ascertaining from the resulting set of decision values the shape class) of the object.

The invention c¼n improve the accuracy of I classification! and can enable the number of classes to ! be increased. \ Preferably the process of deriving decision values is repeated using g oups with the same number of primary shape parameters, and preferably all the possible remaining combinations are used to obtain the maximum information.

Preferably the tjiethod is performed electronically in apparatus for sorting a succession of objects, the apparatus being substantially according to the invention Of GB-A-2184832. However, the number of viewers can be reduced, and it is possible to use one viewer for sorting some obj ects, or two viewers, though a larger number is preferred, for instance three, four or more, the illuminatior. is not restricted to visible light and may be for instance infra-red. The machine may just classify, e. g. providing a total of the objects in each class, or may physically separate different classes of ob ects.

Although the metihod is preferably used with a machine substantially according to the invention of GB-A-2184832, any other machine capable of measuring i . · primary shape parameters may be used.

Preferably the primary shape parameters used comprise the maximum, mini!mura and average values for all the viewers of at leafst two basic shape paraneters.

Suitable basic s ape parameters are blookiness, symmetry and convex hull deviance as set out in GB-A-2184832, and I a satisfactory classification can be achieved on the i basis of these three basio shape features, possibly also with edge break-through (see below). However, other basic shape features that can be used are for instance: i central moments; aspect ratio; j straightness of edge measure; i convex hull deviance normalised with respect to object size; ' convex hull deviance normalised with respect to arc length of missing boundary; area of convex httll to real area; pixel spectrum (Reeling off one laye of pixels at a ons (the relationship between the views from different viewers); I any of the foregoing extended into three dimensions. i i Preferably a trahs ormation is provided for transforming said primary shape parameters onto secondary shape parameters having a fixed range of discrete values.

Preferably the decision as to shape class is made by which decision value is most conunonly identified by all the possible different tables.

The shape class decision may also be made by a hierarchical decision process.

It is preferred that said tables are generated by a training procedure in which, for each shape class, a statistically significant sample of objects falling within that class are fed through apparatus for classifying or sorting, further programmed to derive said table. However, it is not necessary to use a training procedure, and tables derived on another apparatus, or even by a computational method, may be used instead. Said groups are preferably pairs, but it is possible to form the table on the basis of groups of three or more primary shape parameters.

Description of the Preferred Embodiment The invention will be further described by way of example and with reference to the accompanying drawings in which: Figure 1 shows a typical frequency histogram for the linearly transformed values of one of the primary shape parameters in the training process; Figure 2 shows a cummulatlve frequency histogram for the same values in the training process; Figure 3 shows an occurrence map table from the training process; Figure 4 shows a table for a pair of secondary shape parameters (showing shape identifications for only some of the pairs of parameter values); Figure 5 shows a flow chart for the shape classification process; Figure 6 shows a decision tree for the shape class decision process.

In order to operate the method of the invention according to the preferred embodiment, three sets of fixed information will have to be provided and can be stored in the local memory of the sorting apparatus or machine : A. A linear transformation, for transforming the values of the primary shape parameters from the sorting machine to normalised shape parameter values.

B. A non-linear mapping of normalised shape parameter values onto secondary shape parameter values. The primary shape parameters may take any value from a continuum and this non-linear mapping maps regions of the continuum onto discrete values of secondary shape parameter. The secondary shape parameters preferably take values 0, 1, 2, ... up to 15.

C. Decision value map tables (class maps).

The apparatus or sorting machine of Figure 1 of GB-A-2184832 is provided for producing electronic signals representative of shape parameters. Preferably there are nine viewers, the images from each of which are processed to give three basic shape parameters/ namely blockiness, symmetry and convex hull deviance. Thus 27 signal8 are produced by this machine. A further signal representative of the total count of edge-breakthroughs may be provided; any view with edge-breakthrough is marked as invalid, and the signal is suppressed but the fact of edge-breakthrough is recorded; however, if all views of the object (i.e. the view from each viewer) show edge-breakthrough, the object is rejected. A microprocessor derives a smaller set of primary shape i irameters, namely the maximum, minimum and mean for all of the viewers for each of the basic shape features, except for convex hull deviance. As the minimum value of convex hull deviance is usually zero, the minimum convex hull deviance signal need not be provided, and a ninth parameter can be provided by the edge-breakthrough count.

It is preferred that the three sets of information A, B and C are derived for each sorting machine by a training procedure.

Training Procedure The sorting machine can be set up to allow signals from the machine to be fed into a training programme which generates the sets of information A, B and C as set out above.

The information is generated by oompiling results for each class of shape. A statistically-viable sample of a given shape class (say 6000 objects from the mid-range of the class and typical of that class) is fed through the machine to provide for each object the nine primary shape parameter signals as set out above. The data is stored on a computerised data storage system with each file of the storage system containing data for the many objects of the same class. The transformation A for normalising the signals from the sorting machine is now generated. This puts the signals into a more suitable form for reading by the following part of the training procedure. For each of the primary shape parameters the maximum value, Nm_a„ . and the minumum value Nmi. _ for all the objects of that class are taken and given the values 0 and 1023 respectively. The rest of the values N for each primary shape parameter are transformed linearly into values N' in the range 0 to 1023 by the following relation: N = (N - Nmin) x 1023 A Nmax"Nmin The relation is information A referred to above, and is fed into the sorting machine.

A histogram as shown in Figure 1 is then generated showing the frequency of occurrence of each value from 0 to 1023. This histogram is then integrated to give a cummulative frequency histogram as shown in Figure 2. The range of the normalised parameters from 0 to 1023 is then divided into sixteen successive intervals, labelled 0 to 5, each interval having approximately the same number of occurrences - the labels of these intervals are the secondary shape parameters. For a given information loss, the secondary shape parameters can be quanticized more coarsely than the primary shape parameters.

The non-linear transformation of the normalised parameter values lying in the range 0 to 1023, to the secondary shape parameters is the information B referred to above, and is fed into the sorting machine.

This process is repeated for as many classes of shape as are required for the classification or sorting being undertaken. For instance, in sorting rough diamonds, one can sort into nine classes of sawables and seven classes of makeables (sixteen classes in all), namely: Sawables: octahedral perfect crystals octahedral imperfect crystals octahedral stones (ie not pure single crystals) long perfect crystals long imperfect crystals long stones irregular stones shaped stones cubes - irregular and concave (ie waisted) Makeables: maccles (triangular) chips (broken) longs (long chips) flats near sawables (between sawables and makeables) cubes - rounded cubes - geometrically perfect There can also be three classes of rejects, namely: misfeede; no vote (where the stone is of a type unrecognised by the machine); undecided (where the stone is borderline between classes ).

Data for all the classes is now compiled by drawing up shape classification occurrence maps. These may be in the form of tables as in Figure 3 in which the rows represent the values from 0 to 15 of one of the nine secondary shape parameters (for example mean value of symmetry) and the columns represent values from 0 to 15 of a second, different secondary parameter (for example mean values of blockiness). Tables of th s form are generated for all of the possible different groups or combinations of two secondary shape parameters. A formula for the total number n of combinations of a different values from a set of N values is given (N - a)l x a! In the present case there will therefore be gC2 combinations, that is; n = 9! /2! .7! = 9 x 8 2 = 36 tables Tables are completed by entering into each square the frequency with which the two different secondary shape parameters occurred together out of the total 6, 000 objects, listed for each class. The sum of frequencies for each class across a row or down a column should be 6, 000 divided by 16 ie. 375, because of the way the secondary shape parameters are derived from the primary shape parameters. There is a reading in each square of the table for each of the classes tested.

Shape classification map tables (class maps) as in Figure 4 are then generated from the occurrence map tables of Figure 3 by deriving a shape identification for each square of the table.

Class decisions for each block of the tables are based on: If (Class 1 > Ypi) then Class 1 (Class 2 If (Class 2 > Yp2) then Class 2 (Class 1 where Ypl and Yp2 are yield factors based on the purity of the sort required, i. e. the target error rates. The training procedure can be re-run with different target error rates, possibly several times, until a suitable sort is achieved. The shape classification space maps are stored in a computerised memory and are the information C referred to above; they are fed into the sorting machine.

Sorting Once the machine has been trained as above or supplied with the necessary information from another source, it can be used to ascertain the shape class of an object, using the physical sorting apparatus disclosed in GB-A-2184832, in which compressed air nozzles are provided to direct an object whose shape has been determined and which is leaving the shape measuring zone to an appropriate shape bin, a rapid succession of objects being processed. A microprocessor operating according to the invention activates the compressed air supply of the nozzles by a solenoid in order to direct the object into the bin corresponding to its shape class.

Figure 5 shows a flow chart for the shape classification process.

In operation, the obj ct is fed through the detecting zone, and at 1 the signals from the viewers are processed to give 27 basic shape parameter readings, plus a reading representing the total number of edge-breakthroughs, which readings are in turn processed as set out above at 2 to give nine primary shape parameters a to i. These primary shape parameters a to i are then transformed at 3 by transformation A followed by transformation B to give secondary shape parameters a' to i' having values between 0 and 15. Secondary shape parameters are then taken in pairs at 4 and a shape decision value is read off from the appropriate shape classification map table at 5. Means 8 for holding all the possible shape classification map tables, are provided in the form of a RAM or computerised memory; the tables can be stored on disk. This shape decision value will just be a class identification, and it is stored in a memory at 6. The process is then repeated for all the remaining possible different combinations of two different secondary shape parameters. Using nine primary shape parameters, a total of 36 shape decision values are produced for each object. The final shape decision, which ascertains the shape class of the object, is made at 7 and is based upon a majority vote system: If (Class 1 > Eai) then Class 1 Class 2 If (Class 2 > E<¾2) then Class 2 Class 1 where: Ed^ and Ed2 are experimentally derived factors to produce the required sort characteristics. With this system there will be some ' undecided' or ' no vote' results, and one or two bine will be provided for them. These are then hand-sorted.

The operations at 1, 2, 3, 4, 5 and 7 in the flow chart above will be carried out by electronic computing elements in the form of microcomputers or (personal) computers.

If the objects are to be sorted into more than two non-reject shape classes, a decision tree, as shown in Figure 6, may be used. The secondary shape parameters are fed into a sequence of classifiers, each classifier being set up according to Figure 5 to classify the object into one of two shape class groups or into a mis feed/no vote/undecided class. Figure 6 shows an example of a decision tree for six shape classes.

Secondary shape parameters are collected at 9 by a computing element such as a microprocessor or computer and passed to the first classifier 10. This decides whether the object belongs to a group of classes 1, 2 and 5 or to the group of classes 3, 4 and 6, or is mis fed, no-vote or undecided.

If the object belongs to one of the groups of classes the information is then fed to classifier 11 or 12, according to which group of classes the object belongs to. Classifier 11 has three outputs: objects belonging to class 1 or class 2, object belonging to class 5 and undecided. Similarly, classifier 12 has the outputs: objects belonging to class 3, objects belonging to class 4 or 6 and undecided.

If the object is found to belong to class 1 or 2, or to class .4 or 6, the information is passed to classifier 13 or 14 respectively which classify the object as undecided, class 1 or class 2 in the case of classifier 13, and class 4, class 6 or undecided in the case of classifier 14.

For a stone to be assigned to class 4, the information must be passed from classifier 10 to classifier 12, and thence to classifier 14, the outputs being, in order: "class 3, 4, 6", "class 4 or 6", "class 4".

Each individual classifier has its own target error rate values (Yp and E^).

The foregoing description is applicable to any sorting machine with three or more viewers. With two viewers, the mean values of the primary shape parameters are not derived. With one viewer, a single primary shape parameter is produced for each of say blockiness, symmetry and convex hull deviance.

The present invention has been described above purely by way of example, and modifications can be made within the spirit of the invention.

Claims (18)

1. 98614/3 C L A I M S :- 1. A method of ascertaining the shape class of an object, comprising: deriving a set of primary shape parameters representative of the shape of the object; taking a combination of n of the primary shape parameters to provide coordinates for deriving from a s-previously established table a decision value for said combination, n being a fixed integer which is two or more, the table having decision values corresponding to said combination; repeating the process of deriving a decision value for all remaining possible different combi nations^f n primary shape parameters, using previously established tables having decision values, each table corresponding to a combination of n primary shape parameters; and ascertaining from the resulting set of decision values the shape class of the object,

2. The method of Claim 1, further comprising the steps of: feeding the object through a viewing zone; illuminating the object as it passes through the viewing zone, using at least one viewer viewing substantially the whole of the profile of the object as presented to the viewer; deriving from the viewer signals representati e of substantially the whole of the profile of the object as viewed at a particular instant by the viewer; processing the signals to provide the set of primary shape parameters. 98614/3

3. The method of Claim 2, wherein a plurality of viewers spaced in one plane around the viewing zone is used, and the primary shape parameters are derived by taking the maximum, mean and minimum values of each of at least two basic shape parameters representative of the edges of the object.

4. The method of Claim 1, wherein the primary shape parameters are transformed by a mapping including a linear transformation onto a set of normalised shape parameters having values lying in a fixed range.

5. The method of Claim 1, wherein the primary shape parameters are transformed onto secondary shape parameter taking values from a fixed set of values, by a transformation including a non-linear mapping.

6. The method of Claim 1, wherein the primary shape parameters are taken in pairs for deriving said decision value, wherein a table having rows and columuns is provided for each pair of primary shape parameters; the rows of the table representing all the possible values derived from one of the primary shape parameters and the columns of the table representing all the possible values derived from the other primary shape parameter, and the spaces in the table containing a shape identificat on; wherein the values of the primary shape parameters derived for the object are used to read a shape identification from the table.

7. The method of Claim 1, wherein the shape class of the object is ascertained by a majority vote system based on the number of times each decision value is derived from all the tables.

8. The method of Claim 1, wherein the method is used to sort the object into one of two classes, or to reject the object.

9. The method of Claim 1, in which each decision value in each table comprises a vote for the object belonging to a firt class, or to a second class, or no vote for the object belonging to either class.

10. The method of Claim 1, wherein deriving the primary shape members includes the step of deriving a basic shape parameter representative of any optical edge breakthrough at the profile of the object and joining up edges on either side of the breakthrough.

11. The method of Claim 1, wherein deriving the primary shape parameters includes the step of deriving a basic shape parameter representative of the approximation of the object to a spherical shape.

12. The method of Claim 1, wherein deriving the primary shape parameters includes the step of deriving a basic shape parameter representative of the approximation of the object to symmetry.

13. The method of Claim 1, wherein deriving the primary shape parameters includes the step of deriving a basic shape parameter representative of re-entrants in the image .

14. The method of Claim 3, wherein a primary shape parameter is derived representative of the total number of edge breakthroughs observed for all the viewers.

15. The method of Claim 1, wherein n is less than the number of primary shape parameters.

16. A method of ascertaining the shape class of an object, comprising: 98614/3 feeding the object through a viewing zone; illuminating the object as i t passes through the viewing zone, using at least one viewer viewing substantially the whole of the profile of the object as presented to the viewer; deriving from the viewer signals representative of substantially the whole of the profile of the object as viewed at a particular instant by the viewer; processing the edge signals to produce a set of basic shape parameters, the set of basic shape parameters including a parameter representative of the approximation of the object to a sphere, a parameter representative of the approximation of the object to symmetry, a parameter representative of the convex hull deviance of the object and a signal representative of the number of edge breakthroughs for the view; deriving a set of primary shape parameters by taking the maximum, average and minimum for all the views of each of the said basic shape parameters of approximation to a sphere and to symmetry, the maximum and average values of convex hull deviance and the total count of edge breakthroughs for all the views; transforming the primary shape parameters onto norma ised shape parameters having values lying in a fixed range; transforming the normalised shape parameters onto a set of secondary shape parameters, taking values from a fixed set of values; taking a pair of the secondary shape parameters as co-ordinates for deriving from a table a decision value for said pair, the table being fixed for all the objects for said pair, the rows of said table represen- 9S614/S ting all the possible values of one of said pair of secondary shape parameters, the columns of the table representing all the possible values of the other of said pa r of secondary shape parameters, and the spaces in the table representing a shape identification in the form of a vote for one of two shape classes, or no vote for either; repeating the process of deriving a decision value for all the remaining possible different pairs of secondary shape parameters using a specific said table for each pair; and ascertaining from the resulting set of shape identification the shape clas of the object on the basis of a majority vote system.

17. A classifying machine for classifying a succession of objects according to shape, comprising: a viewing zone through which each successive object will be fed; means for illuminating the object as i t passes through the viewing zone; at least one electronic viewer for generating viewer signals upon viewing the object as it passes through the viewing zone; means for deriving, from the viewer signals, edge signals representat ve of the edge of the object as viewed by the viewer; means for deriving from the edge signals a set of primary shape parameters representative of the shape of the object; means for holding decision value tables for all groups of n of the primary shape parameters when such groups are used to provide coordinates to the respec- 98614/3 tive tables, n being a fixed integer which is two or more; means for providing said coordinates to respective said tables and deriving from respective said tables decision values for the respective groups; and means for ascertaining from the resulting set of decision values the shape class of the object.

18. The classifying machine of Claim 17, wherein a plurality of viewers spaced around the viewing zone is used: the edge signals being processed to produce a set of basic shape parameters, the set of basic shape parameters including a parameter representative of the approximation of the object to a sphere, a parameter representative of the approximation of the object to symmetry, a parameter representative of the convex hull deviance of the object and a signal representative of the number of edge breakthroughs for that view; the primary shape parameters being the maximum, average and minimum for all the views of said basic shape parameters of approximation to a sphere and symmetry, the maximum and average values of convex hull deviance and the total count of edge breakthroughs for all the views; the primary shape parameters being transformed onto normalised shape parameters having values lying in a fixed range for the machine; the norma ised shape parameters being mapped onto a set of secondary shape parameter, taking values from a set of values defined for the machine; said means for providing coordinates to said tables providing said secondary shape parameters in pairs; #•644/$ tables for all possible different pairs of secondary shape parameter being stored in a memory, the rows of said tables representing all the possible values of one of said pair of secondary shape parameters, the columns of the table representing all the possible values of the other of said pair of secondary shape parameters, and the spaces in the table representing a shape identification, in the form of a vote for one of two shape classes, or no vote for either; the means for deriving decision values from said tables for the respective pairs having means to read a shape classification vote from said table using the secondary shape parameters as coordinates; the means for as certaining the shape class of the object comprising a memory in which all the shape class votes provided by said means for deriving decision values are stored and means for classifying the object as belonging to the first class, belonging to the second class or undecided on a majority vote basis using said shape class votes. For the Applicant, Simon Lavie Patent Attorney