FR2543457A1 - METHOD AND DEVICE FOR SORTING OBJECTS ACCORDING TO THEIR EXTERNAL APPEARANCE, PARTICULARLY FOR COLORIMETRIC DRAWING OF OBJECTS - Google Patents

Abstract

THE INVENTION CONCERNS A PROCESS AND A DEVICE FOR SORTING OBJECTS AS A FUNCTION OF PARAMETERS CONCERNING THEIR EXTERNAL APPEARANCE. THIS PROCESS CONSISTS OF GENERATING AT LEAST ONE VIDEO IMAGE OF EACH OBJECT USING A VIDEO 3, 4 CAMERA, IN REAL TIME ENTERING THE ANALOGUE VIDEO SIGNAL OF EACH IMAGE AND IN CONVERTING IT INTO A MATRIX OF DIGITAL CODES REPRESENTATIVE OF THE LEVELS OF GRAY DOTS, TO BE MEMORIZED AND PROCESSED BY CALCULATION EACH MATRIX OF NUMERICAL CODES IN A WAY TO GENERATE A SERIES OF NUMERICAL INFORMATION REPRESENTATIVE OF THE OBJECT'S PARAMETRIC VALUES, AND TO BE CARRIED OUT ON THESE SERIES OF USEABLE CLASSIFICATION INFORMATION.

Description

METHOD AND DEVICE FOR SORTING OBJECTS INTO

  FUNCTION OF THEIR EXTERNAL APPEARANCE, PARTICULARLY FOR

A COLORIMETRIC SORTING OF OBJECTS.

  The invention relates to a method of sorting objects according to parameters relating to their external appearance;

  it extends to a device for the implementation of said program.

  assigned The invention may in particular be applied for

  read a calorimetric sorting of objects, in particular fruits,

  in order to classify them according to calorimetric criteria.

  Many devices are known which aim to

  read a colorimetric sorting of objects The majority of these devices performs the sorting by sending a signal to the object

  or a given radiation, by capturing said signal or radiating-

  ment after reflection on the object or passage through it and by carrying out an analog measurement of this last signal or radiation to deduce certain characteristics of the object One could, for example, refer to French patents NO 2 183 128 , 2 183 130, 2 235 365, 2 319 894, 2 327

  526, 2 459 620, etc, which describe several types of dis-

  However, these arrangements

  only provide average information about

  the object, which is completely incapable of taking into account

  possible surface heterogeneities thereof; this in-

  poor and fleeting training only offers the possibility of ef-

  make very rough classifications where the object is considered

  overall, independently of its particular details. In addition, these systems require specific means of lighting objects and their structure is made considerably more complex. In addition, such systems are fixed for a specific application (objects of a natural determined, nature and sorting criteria determined) and have

  no adaptability to be used in an application

  different cation, same neighbor It should be noted, however, that the patents 2 363 097 and 2 473 910 describe systems

  with phototransistors which have possibilities of adaptation

  more important than other systems due to the

  presence of a programmable microprocessor provided for analysis

  information from phototransistors.

  Furthermore, another type of device is described in French patent No. 2,269,255; this device performs the sorting by projecting on each object a striated pattern, by generating a video image of the surfaces of said pattern projected on the object and by identifying on this video image elements characteristic of the deformation of the projected pattern, in view

  to acquire information concerning the form or the orientation

  tion of said object However, such a device has defects similar to those previously mentioned. It is frozen for a specific application of shape recognition or

  without any possibility of adaptation to

  In addition, its principle of operation

  allows it to perform sorting only according to

  form or orientation criteria and not colorimetric sorting

  In addition, the "on the fly" analysis of the analog video signal to define the characteristics of the object only allows sorting based on a small number of

  simple parameters It should be noted that the need to project

  ter a pattern on each object implies a relative device-

  complex and limits its application to objects of su-

  sufficient area, the disparities of form of which are not

too important.

  The present invention proposes to overcome

  faults in existing systems.

  An objective of the invention is in particular to provide a system capable of carrying out an extremely fine sorting of objects, making it possible in particular to differentiate objects having local heterogeneities of surfaces such as local irregularity, spot spot, local differences

  of color, etc., even if the average parametric values

  nes of these objects do not allow to distinguish them.

  Another object of the invention is to provide

  nin a system benefiting from a great flexibility of adaptation

  tion in order to make it usable for different applications

  o the nature of the sorting to be carried out and its parameters

can be very different.

  Another objective is to provide a simple structure system capable of allowing sorting rates.

very high.

  To this end, the sorting process targeted by the

vention essentially consists of: -

  generating at least one video image of each object by means of at least one video camera,

  to enter, in real time, the analog signal

  that video of each image and to convert it into a matrix of digital codes, each representative of the gray level of the corresponding point of the image,

  to memorize, in real time, the matrix (s)

  these digital codes corresponding to each object, to be processed, by calculation, each matrix of digital codes as a function of the sorting parameters, so as to generate a series of digital information representative of the parametric values of the object,

  * to be carried out on the series or series of information

  numerical information concerning each object, function tests of the sorting to be carried out in order to deliver information of

exploitable classification.

  It should be noted that by "video image", we mean

  broadly extends not only a standard television picture, but also any picture represented by a signal

  electric generated by a two-dimensional electronic scan-

  object (X-ray, U or infrared scanning, ima-

  ge supplied by a charge transfer circuit "C C D", etc.).

  The invention is therefore characterized essential-

  by the combination of the following three points used

  tion of a video image of each object, which is converted

  into a digital matrix; real-time storage of this-

  the matrix which remains available for the entire duration

  treatment; processing in two successive calculation steps

  sive functions of the sorting to be carried out, a first stage during which are stored, in the form of intermediate calculation results, the information of the matrix useful for carrying out the sorting considered, a second stage o

the sorting decision is made.

  Thus, a complete and fine representation of each object is available before treatment and during

  the entire duration of it, so the process of the in-

  vention allows to take into account all the characteristics

  whatever aspect of the object they are, whatever they affect

  attempt the object as a whole or on the contrary that they are limited to an area thereof The two stages of

  treatment are programmable according to each application

  tion and allow according to it, in the first place,

  to keep, under contracted forms, the information used

  the sorting envisaged, then to carry out the taking tests

  decision on this information or on specific information

  picks of these These cascading steps allow you to

  benefit, at the same time, from a good speed of treatment and a great simplicity of means thanks to the elimination of useless information taking into account the sorting to be carried out and this without harming the flexibility of adaptation of the process since

  this deletion, which is done from a numerical matrix

  that very rich, is programmed according to sorting.

  Video images can be generated in black and white by standard video cameras at low cost; the digital conversion then consists in assigning to the points of the image codes with several bits representative of the gray levels of these poirts among a

  number of predetermined levels between white and black.

  If applicable, video images may

  can be generated in color, the conversion being carried out, in a manner identical to that mentioned above, for each chromatic component (three in number in a signal

standard color video).

  According to another characteristic of the pro-

  transferred, the objects to be sorted are brought to scroll along at least one sorting line and two video images of each object seen from different angles are generated by means of two cameras arranged facing each sorting line, the daux

  images of the same object being converted, stored and processed

  tees with a time difference so as to deliver successively

  two series of digital information corresponding

  said object, series of information on which are

  then carried out the sorting tests to be carried out.

  It is thus possible to take into account the appearance of the entire surface of the object by reducing

the hidden parts of it.

  The invention makes it possible in particular to

  read a sorting of objects according to colorimetric parameters.

  The processing of each matrix of digital codes can then consist in calculating the weight of each gray level by counting the number of digital codes corresponding to the various gray levels, with a view to delivering

  words in number equal to the number of gray levels programmed-

  més, the tests carried out on this series of words consisting

  to calculate statistical quantities characteristic of the

  said series, such as mean, standard deviation, variance and at

  compare them to thresholds to deliver class-

ment exploitable.

  The number of gray levels taken into account

  is programmed according to the required analysis finesse

  by sorting, in particular: 8 levels (three-bit words), 16 ni-

  calves (four-bit words) or, in the case of very sophisticated sorting 64 levels (six-bit words) For example, using 16 gray levels, experiments have shown that it was

  possible to sort apples according to colorimetric standards

  international, with an error rate of less than 5%

  and at rates higher than 5 fruits per second per li-

sort sort.

  In the event that two video images are captured

  entered, entered and digitized for each object, each of the two corresponding numerical matrices is, in the first stage of processing, treated independently with a view to obtaining the words representative of the weight of the various gray levels taken into account The tests are then carried out on both

  series of words obtained at the end of this first step, pre-

  preferentially in the following way t calculation of three sets of statistical quantities, characteristic, on the one hand, of the first

  series of words, on the other hand, of the second series of words, en-

  end, of the union of these two series, calculation, in a space with 3 N dimensions (n equal to the number of statistical quantities), of the distances separating the point having as coordinates the three sets of statistical quantities mentioned above and of the points of

47457

  benchmarks representative of class boundaries,

  determination of the corresponding benchmark

  laying at the minimum distance and providing classification information attached to said benchmark. Such treatment allows, with means

  relatively light calculation, to use all the information

  two useful video images without introducing distortion

  due to possible overlap areas.

  The invention extends to a device for

  sorting for the implementation of the above process

  ni This device essentially comprises: at least one video camera arranged to successively generate video images of the objects to be sorted,

  associated with each video camera, a module called "pro-

analysis ceaser "including:

  an analog / digital conversion unit

  camera receiving the video signal from the camera and

  adapted to ensure the capture of this signal and deli-

  check a matrix of digital codes as a function of the amplitude of said sampled video signal,

  a storage memory, capable of memorizing

  be at least one matrix of numeric codes and

  mandated by a direct memory access unit capable of managing the storage of digital codes in memory,

  a calculation unit suitable for assu-

  rer the processing of digital codes contained in the storage memory and associated, on the one hand, with

  a program memory containing an operating program

  on the other hand, to a logical

  working memory for the provisional storage of

  results, arranged to receive information from the various analysis processors, a module called "processor

  decision "including a calculation unit adapted to perform

  kill tests on the information received and associated, on the one hand, with a program memory containing the tests to be carried out,

  on the other hand, to a working memory for pro-

  view of test results, means of transmission capable of transmitting

ES 43457

  the information from the various analysis processors to the decision processor,

  a communication interface for class information

  ; ment from the decision processor, to one or more operating means. Each calculation unit can advantageously be

  constituted by a microprocessor.

  Other features, purposes and benefits of

  l 0-the invention will emerge from the following description with reference to

  this in the appended drawings, which show an unrelated example thereof.

mitative; on these drawings

  Figure 1 is a schematic view of a dis-

  positive sorting according to the invention, Figure 2 is a partial block diagram of this device, Figure 3 is a block diagram of a mode

  of one of the assemblies of the device.

  The sorting device shown by way of example in the figures comprises several sorting lines T 1, T 2,

  T 3.ç 7 along which the objects to be sorted are brought to defi-

  ler by suitable drive means (Figure 1) In the example, the objects consist of apples which are arranged on conveyor belts, such as 1; each carpet

  may consist of rollers (not shown) which rotate

  keep the apples on their own as they move.

  At the start of the mat, each sorting line is equal

  presence sensor 2 adapted to detect the presence

  of apples and generate a presence signal when an apple

  file at its level This sensor 2 is of any type known per se (photosensor, electromagnetic sensor,) In addition, two black and white video cameras 3 and 4 are arranged opposite each sorting line and offset

  relative to each other along the direction of travel

  carpets, so as to provide an image of two faces of each apple (this one having turned on itself during its journey between the cameras) 0

  These standard type cameras each provide

  ne to an analysis processor 5 an analog video signal; sure

  presence signal control from sensor 2 (after intro

  duction of a delay corresponding to the offset of the camera

  stunned and sensor 2), the video signal is captured, converted into a matrix of digital codes, then processed to deliver a series of digital information, sorting function

  and representative of the parametric values of the object.

  The series of information from the various pro-

  analysis halters are issued, through means of trans-

  asynchronous mission 6, to a decision processor 7, suitable for carrying out sorting tests on said series The two sets of information, representative of the same object and coming from the two cameras 3 and 4 of the same sorting line, are

  converted, stored and processed with a time difference of

  finished, in the two corresponding analysis processors,

  then delivered successively to the decision processor 7.

  After carrying out the tests, the latter delivers exploitable classification information which is routed through a communication interface 8, to an operating device (display means, printer, interface

  manufacturer for the control of a mechanical switching system

ge O).

  Figure 2 is a block diagram showing

  the general structure of an analysis processor 5 and of the process

decision maker 7.

  The analysis processor 5 comprises a conversion unit 9 which receives the video signal from the camera 3 and the presence signal from the detection sensor 2; in the presence of a fruit, this unit 9 captures the video signal and assigns to the points of the image codes with several bits

  representative of the gray levels of these ptints by -i a name-

  bre of predetermined levels between white and black, deli-

  thus showing a matrix of digital codes representative of

said image.

  It should be noted that most of the apps

  cations does not, in practice, require a definition as fine as that provided by a standard video image; preferably the conversion unit is suitable, as we will see more

  far, to select certain lines distributed among the ea-

  seems to be lines of the frame (for example 1 line on 2) and, for each selected line, a certain number of points (for example 100 points per line) This reduces the storage and calculation means of the analysis processor without harming ,

  in practice, the finesse of the sorting to be carried out.

  The analysis processor also comprises a storage memory 10 capable of storing a matrix of digital codes on command of a direct access unit

  memory 11, a computing unit 12 composed of a microprocessor

  a program memory 13 and a working memory 14

associated with the microprocessor.

  Program memory 13 contains the pro-

  gram of the logical operations to be performed by the microprocessor

  sor 12, logical operations which are a function of sorting to reali-

  ser and which allow, from the numerical matrix

  naked in the storage memory 10, to develop a series of in-

  much smaller training The program written in memory 13 is easily adaptable to each

  application without modifying the structure of the device.

  For example, for a calorimetric sorting with 16 gray levels, each series of information coming from the microprocessor 12 represents the weight of each gray level

  and includes 16 words of 16 bits each, which are delivered through

  towards the transmission means 6 to the decision processor 7.

  It should be noted that the processing of each

  that the digital code matrix by the microprocessor 12 is produced in real time during the period of time separating the end of storage in memory 10 and the entry of a

  new image in the conversion unit; the matrix of co-

  of digital is kept in memory 10 for this entire period of time In the above example, assuming that the microprocessor is of the "TEXAS TMS 9995" type, the processing lasts approximately 100 milliseconds, which is perfectly

  compatible with obtaining high sorting rates.

  Furthermore, the decision processor 7 receives the successive series of information from the analysis processors, which are distributed through

  from a bus 15 to a computing unit 16 constituted as a pre-

  cedently by a microprocessor; this is associated with a program memory 17 containing the calculations and tests

  to be performed and to a working memory 18.

  The program contained in memory 17

is suitable for the planned sorting.

  In most cases, a calculation is first performed by the microprocessor 16 to provide

  statistical quantities which represent, in contrac-

  ty, each set of information (mean, standard deviation, varian-

  ce) and are a function of the sort to be carried out; these are the great

  statistical data which are then compared with thresholds to deliver a signal to the communication interface 8

of exploitable classification.

  In the case of two cameras arranged on each sorting line, two sets of information are delivered by each object to the bus 15; these series are temporally

  shifted by a determined period of time and this characteristic

  that is used by the bus to match the

  series of information concerning the same object, in order to

  put the microprocessor 16 to make the appropriate calculations

  required For example, the statistical quantities mentioned above

  will be calculated, on the one hand, for each of the 2 sets of information

  training, on the other hand, for the whole constituted by the

  nion of these series and the tests will be carried out on these three sets of statistical quantities (in particular tests of minimum distance compared to reference points in a

  space with 3 N dimensions, where N is the number of static quantities

* statistics taken into account).

  FIG. 3 shows by way of example the block diagram of an embodiment of the unit of

conversion 9.

  In a calibrator 19, the component is separated

  video health and the synchronization signals of the video signal coming from the camera The calibrated video component is sent to a sampling and conversion module 20, which is suitable for temporally cutting said video component

  and to assign to each sample a numerical code represented

  signal amplitude sensitive This code is generated by

  comparison with a plurality of amplitude thresholds (8, 16 or 64)

representative of gray levels.

  The synchronization signals from the calibrator 19 are processed to control the start-up and

  coding stopped by module 20.

  A monostable flip-flop 21 extracts the

  signals from the start of the frame and delivers them to another flip-flop 22 which receives the signal from the presence sensor 2 after passing

ge in a delay circuit 23.

  This flip-flop 22 delivers a signal of

  validation when it is in the presence of both a start of frame signal and a signal representative of the presence of a fruit This validation signal is delivered to the validation input of another flip-flop 24 which receives the

  start signals of selected lines.

  These signals are generated from the

  general synchronization using a divider 25 which extracts certain line start signals and suppresses others For example, this divider can be provided to extract a signal

  start of line on 2, so that only the corresponding lines

  dantes will then be taken into account.

  These selected signals pass through a

  timed flip-flop 26, in order to eliminate the signal "black from ex-

  'which is at the start of each line.

  The flip-flop 24 receives the line start signals from the timed flip-flop 26, the validation signals from the flip-flop 22 and, on another input, conversion stop signals when the desired number of points has been coded for a given line This flip-flop 24 delivers, to a MAND gate 27, signals which authorize the coding of the lines selected from the start of a frame.

  Gate 27 simultaneously receives the

  of a clock 28 (which also synchronizes the diff-

  different components of the device) and delivers conversion control signals to the module 20 In the presence of a conversion signal, the module performs the sampling of the calibrated video signal of the line in question and the coding of the samples It delivers the digital codes to a buffer memory 29 which stores them temporarily before they are stored in the storage memory 10 of the analysis processor. The module 20 is connected to a counter 30,

  which it increments at each coded point This counter is programmed

  mé to provide a conversion stop signal

  the 24th when a certain number of points distributed on the line have been coded (one can for example take into account the O O

  points on each selected line).

  Thus, the conversion unit described above by way of example makes it possible to select certain lines from the set in the video component and / lcount and code

  a determined number of points on each line.

  This selection will in practice depend on

  the complexity and finesse of the sorting to be carried out; she per-

  puts to lighten the means of the device (means of memorization

  tion, calculation methods) and achieve processing speeds

higher.

  By way of illustration, the various characteristics of a prototype device which has

  was carried out on an experimental basis.

Objects to sort: apples

  sorting parameters: colorimetric sorting

  training in international standards (standards applicable from October 1, 19712 distributed by the Standardization Service "CTIFL" I Interprofessional Technical Center for Fruits and Vegetables: 4 standard categories in red and 6 in green or yellow) number of lines of sorting: 3 drive speed: 0.5 m / s overall sorting rate: 15 fruitfsecond divider 25 adapted to validate every other line, counter 30 adapted to take into account points per line, clock: frequency 16 M Hz module 20 programmed to code 16 gray levels (4 bits) storage memory 10: capacity 12 kx x 4 bits microprocessor 12 type "TEXAS TMS 999511 unit 11 type" TEXAS DMA, TMS 9911 "programmed memory 13: memory" EPROM "'of 6 k bytes programmed to control the weight scale for each gray level,

  working memory 18: RAM memory of 2 k oc-

  tets means of transmission: "UART units: TMS

9902 "

  microprocessor 16 identical to microprocessor

  sor 12 program memory 17: "EPROM" memory of 4 k bytes programmed to control the calculation of the average of the weights and variance and for their comparison with thresholds working memory 18: "RAM" memory of 2 k bytes interface communication 8: RS 232

  scanning time of an image: 1/50 s.

  duration of memorization in memory 10: milliseconds

  processing time in the analysis processor

  lysis 5: 100 milliseconds processing time in the decision processor 7: 10 to 20 milliseconds

  error rate less than 5%.

Z 5243457

Claims (10)

CLAIMS I / Method for sorting objects according to parameters relating to their external appearance, characterized in that it consists in: generating at least one video image of each object by means of at least one video camera (3, 4) , to capture, in real time, the analog video signal of each image and to convert it into a matrix of digital codes, each representative of the gray level of the point corresponding to the image, to memorize, in real time, the matrix or matrices of digital codes corresponding to each object, to be processed, by calculation, each matrix of digital codes as a function of the sorting parameters, so as to generate a series of digital information representative of the parametric values of the object, to be performed on the series or series of digital information relating to each object, tests of the sorting functions to be carried out in order to deliver usable classification information.   2 / Sorting process according to the claim   tion 1, in which the objects to be scrolled are brought along at least one sorting line (L 1, L 2), characterized in that   two video images of each object seen under images are generated   different types by means of two cameras (3, 4) arranged opposite each sorting line, the two images of the same   object being converted, stored and processed with a shift   time geometry so as to successively deliver two series of digital information corresponding to said object, series of information on which are then carried out sorting tests to be performed.   3 / Sorting method according to one of the   vendications 1 or 2, characterized in that one generates   black and white video images, digital conversion consists   both to assign to the points of the image codes with several bits representative of the gray levels of these points among   a number of predetermined levels between white and black.   4 / Sorting method according to one of the   vendications 1, 2 or 3, characterized in that one detects   543457   the presence of each object to be sorted independently of the ge-   video image generation and that we allow the capture   the video signal from the detection of the presence of an object. / Sorting method according to one of the   vendications 1, 2, 3 or 4, characterized in that the conversion   digital sion consists, for each video image: f of extracting from the video signal the video synchronization signal, O of generating from this synchronization signal, conversion control signals giving the instant of start of conversion of each line of the image to be converted, on command of these signals, to be produced in T   converting the video component of the video / sample signal   playing the said video component and assigning to each sample a digital code representative of its amplitude, counting down the number of samples thus coded and stopping the conversion of each line after coding of a determined number of them, temporarily storing each numeric code.   61 Sorting method according to claim-   tion 5, characterized in that during the digital conversion   risk: in the video component of the video signal, lines distributed among all of the   lines and we generate a single conversion command signal   ment for these lines selected so as to delete the coding of the other lines,   for each selected line,   count and code a determined number of samples.   71 Sorting method according to one of the   vendications 1, 2, 3, 4, 5 or 6, characterized in that the processing   Each stored matrix is produced in real time during the period of time between the end of storage of the considered matrix and the capture of a new image, the c, considered matrix being kept in memory for all this amount of time.   8 / Method according to one of the claims   1, 2, 3, 4, 5, 6 or 7, for sorting objects according to colorimetric parameters, characterized in that the processing of each matrix of digital codes con-   is to calculate the weight of each gray level in decomp-   as the number of digital codes corresponding to the various gray levels, in order to deliver words in a number equal to the number of programmed gray levels, the tests carried out on this series of words consisting in calculating statistical quantities characteristic of said series such as mean, standard deviation, variance and compare them to   thresholds for delivering exploitable classification signals.   9 / Sorting process according to the claims   tions 2 and 8 taken together, characterized in that the tests carried out on the two series of words corresponding to the two video images of the same object consist in: calculating three sets of statistical quantities, characteristic, on the one hand, of the first   series of words, on the other hand, of the second series of words, en- end of the union of these two series.   to be calculated, in a space with 3 N dimen-   sions (n equal to the number of statistical quantities), the dis-   tances separating the point having for coordinates the three en-   seemingly above mentioned statistical magnitudes and benchmarks representative of elasse boundaries,   to determine the appropriate benchmark   laying at the minimum distance and delivering information   of classification attached to said benchmark.   10 / Device for sorting objects in   view of the implementation of the method in accordance with one of the   previous sales, characterized in that it comprises: at least one video camera (3, 4) arranged to successively generate video images of the objects to be sorted, associated with each video camera, a module 5) called "analysis processor" including:   an analog / digital conversion unit   meric (9) receiving the video signal from the caoera and adapted to ensure the capture of this signal and   issue a matrix of digital codes based on C 543457   the amplitude of said sampled video signal   read, a storage memory (10), capable of memorizing at least one matrix of digital codes and controlled by a direct memory access unit (11) capable of managing the storage in memory of digital codes, a calculation unit (12) suitable for   ensure the processing of digital codes   held in the storage memory (10) and associated   ciée, on the one hand, to a program memory (13) containing a program of logical operations to be carried out, on the other hand, to a working memory (14) for the temporary storage of the results, arranged to receive the information from   various analysis processors (5), a module (7) called "process   decision generator "comprising a suitable calculation unit (16)   to test the information received and asso-   linked, on the one hand, to a program memory (17) containing the tests to be carried out, on the other hand, to a working memory (18) for the temporary storage of the test results,   transmission means (6) capable of transmitting   the information from the various analysis processors (5) to the decision processor (7),   an interface (8) for communicating information   classification from the decision processor (7) to one or more means of exploitation.   11 / Sorting device according to the   vendication 10, in which each computing unit (12, 16)   is made up of a microprocessor.   12 / Sorting device according to one   claims 10 or 11, characterized in that the unit   analog / digital conversion (9) includes:   a video signal calibrator (19)   with a calibrated video component and sync signals   nisation, a set of scales and doors   loqigues (21, 22, 24, 26, 27), receiving the syn-   timing and associated with a clock (28) and a counter (30) for delivering conversion control signals, and an analog / digital sampling and conversion module (20) receiving the calibrated video component and the control signals conversion to deliver   digital codes to a buffer memory (29), said mo-   dule being connected to the aforementioned counter (30) with a view to its increment statement for each coded point.   13 / Sorting device according to the re-   vendication 12, characterized in that a divider (25) is   associated with the set of scales and doors, in order to remove   sea of line start signals so as to remove the co- dage points of said lines.   14 / Sorting device according to one   claims 10, 11, 12 or 13 comprising means (1)   for sorting objects adapted to cause them to scroll along sorting lines (Li, L 2,), characterized in that a sensor for the presence of objects (2) is arranged on each sorting line and connected to the analysis processors (5)   of the line considered in order to authorize the entry of the   video signals generated by the cameras (3, 4) of said sorting line.