DK167462B2 - Method and plant for use in treating a meat subject - Google Patents
Method and plant for use in treating a meat subject Download PDFInfo
- Publication number
- DK167462B2 DK167462B2 DK199101504A DK150491A DK167462B2 DK 167462 B2 DK167462 B2 DK 167462B2 DK 199101504 A DK199101504 A DK 199101504A DK 150491 A DK150491 A DK 150491A DK 167462 B2 DK167462 B2 DK 167462B2
- Authority
- DK
- Denmark
- Prior art keywords
- image
- values
- areas
- sum
- light source
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims description 30
- 235000013372 meat Nutrition 0.000 title claims description 18
- 210000003484 anatomy Anatomy 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 18
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000001914 filtration Methods 0.000 claims description 10
- 230000009466 transformation Effects 0.000 claims description 6
- 230000004807 localization Effects 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 2
- 235000012054 meals Nutrition 0.000 claims 1
- 230000003287 optical effect Effects 0.000 claims 1
- 238000012216 screening Methods 0.000 claims 1
- 238000011282 treatment Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 210000002414 leg Anatomy 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 241000950638 Symphysodon discus Species 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- HOQADATXFBOEGG-UHFFFAOYSA-N isofenphos Chemical compound CCOP(=S)(NC(C)C)OC1=CC=CC=C1C(=O)OC(C)C HOQADATXFBOEGG-UHFFFAOYSA-N 0.000 description 2
- 238000001454 recorded image Methods 0.000 description 2
- 241000283690 Bos taurus Species 0.000 description 1
- 241001050985 Disco Species 0.000 description 1
- 241000667653 Duta Species 0.000 description 1
- 241001508691 Martes zibellina Species 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 210000001217 buttock Anatomy 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000006735 deficit Effects 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000007688 edging Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000945 filler Substances 0.000 description 1
- 210000003194 forelimb Anatomy 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 235000015277 pork Nutrition 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 210000002374 sebum Anatomy 0.000 description 1
- 238000003307 slaughter Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A22—BUTCHERING; MEAT TREATMENT; PROCESSING POULTRY OR FISH
- A22B—SLAUGHTERING
- A22B5/00—Accessories for use during or after slaughtering
- A22B5/0017—Apparatus for cutting, dividing or deboning carcasses
- A22B5/0058—Removing feet or hooves from carcasses
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- 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
- G06T2207/30128—Food products
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Food Science & Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Description
W DK 167462 B2 l : ,-ϋ- ·***£*·<«·.-·, «· -i-Γί:· *> Den foreliggende opfindelse angår en fremgangsmåde ved behandling af et kødernnfe.'ved hvilken fremgangsmåde kødemnet belyses med en lyskilde, og der optages et videobillede ved hjxlp af el videokamera, hvilket billede registreres og derefter databehandles i en beregningsenhed for al lokalisere bestemte områder på emnet, såsom bestemte anatomiske 5 områder, hvorefter beregningsenheden afgiver et af lokaliseringen afhængigt signal til brug ved efterfølgende behandling af kødemnet.The present invention relates to a method of treating a meat additive by which method the meat item is illuminated with a light source and a video image is recorded by means of an electric camcorder, which image is recorded and then processed in a computation unit for all locating certain areas of the subject, such as certain anatomical areas, after which the computing unit emits a signal dependent on the location for use. by subsequent treatment of the meat subject.
I forbindelse med automatiseret behandling af kødemner er det blevet foreslået, at der anvendes videooptagelser til fastlæggelse af emnets ydre karakteristika, hvorefter behandlingen gennemføres på grundlag af de fundne karakteristika. Videooptagelser kan fx anvendes i 10 forbindelse med halv- eller helautomatisk klassificering af slagtekroppe. Et videobillede af kroppen registreres, og det registrerede billede databehandles i en computer for at fremhæve og isolere form- og farveværdier, der er af betydning for slagtekroppens klasse.In the context of automated processing of meat subjects, it has been suggested that video recordings are used to determine the external characteristics of the subject, after which the treatment is carried out on the basis of the characteristics found. Video recordings can be used, for example, in connection with semi-or fully automatic classification of carcasses. A video image of the body is recorded and the recorded image is processed in a computer to highlight and isolate shape and color values that are important to the carcass class.
En anden anvendelse er bestemmelse af kød/spækforholdet i en slagtekrop. En videooptagelse af snitfladen i en flækket slagtekrop underkastes en databehandling, idet de grå områder regnes 15 for at være kød, mens de lyse arealer repræsenterer spæk. De sorte områder ses der bort fra,idet de udgør baggrunden. , '•'V. * ·Another use is the determination of the meat / pork ratio in a carcass. A video recording of the cut surface in a speckled carcass is subjected to data processing, the gray areas being considered to be meat, while the light areas represent blubber. The black areas are disregarded as they form the background. , '•' V. * ·
Videooptagelser kan også anvendes i forbindelse med halv- eller helautomatisk bearbejdning af kødemner, fx til automatisk indstilling af en sav, der skal partere en flækket slagtekrop. Ved passende databehandling af et billede af kroppen kan man beregne positionen af forud bestemte 20 anatomiske dele, der har en sammenhæng med den ønskede placering af snittet. Savens eller kroppens stilling kan derefter justeres i overensstemmelse med den fundne position, si snittet bliver lagt korrekt. Anatomiske dele, der kan anvendes til delte formål, er fx den flækkede krops forben eller rygrad.Video recordings can also be used in connection with semi-or fully automatic processing of meat items, for example for automatically adjusting a saw to partition a speckled carcass. By appropriate data processing of an image of the body, one can calculate the position of predetermined 20 anatomical parts which are related to the desired location of the incision. The position of the saw or body can then be adjusted according to the position found, if the cut is laid correctly. Anatomical parts that can be used for shared purposes are, for example, the forelimb or spine of the fractured body.
For det meste anvendes det registrerede billede kun til bestemmelse af emnets kontur, og 25 lokaliseringen af det anatomiske område må derfor udføres på grundlag af konturens oplysninger. Det er imidlertid begrænset, hvor nøjagtige og forskelligartede behandlinger, ejer kan udføres på kødemner på basis af sådanne lokaliseringer. . \ I WO-A-90/l 0915 er beskrevet en melode lil filtrering af et registreret billede opbygget af billedelementer, fx et billede af et fly. Metoden kan forstærke konturer, der forekommer i billedet. ' 2 DK 167462 B2 DK B 157.579 beskriver en fremgangsmåde og et anlæg til bestemmelse af kvaliiets-5 egenskaber ved individuelle kreaturslagtekroppe, hvorved der tilvejebringes en høj kontrast mellem en mørk krop og en lysudsendende baggrund for opnåelse af et nøjagtigt videobillede af kroppens kontur. Der tilvejebringes desuden el videobillede af slagtekroppen, medens den er belyst, med henblik på bestemmelse af hvor stor en andel af overfladen, der er dækket af talg.For the most part, the recorded image is used only to determine the outline of the subject, and therefore the location of the anatomical area must be performed on the basis of the outline information. However, it is limited where accurate and diverse treatments can be performed on meat items based on such locations. . WO-A-90 / l 0915 describes a melody for filtering a registered image made up of image elements, for example, an image of an aircraft. The method can enhance contours that appear in the image. 2 DK 167462 B2 DK B 157.579 discloses a method and apparatus for determining the quality characteristics of individual cattle carcasses, thereby providing a high contrast between a dark body and a light transmitting background for obtaining an accurate video image of the body contour. Additionally, video images of the carcass while illuminated are provided for the purpose of determining the proportion of the surface covered by sebum.
10 EP Bl 20.417 beskriver et elektronisk billcdbchandlingssystcm med et højpasfiller, der reducerer støjen. I publikationen nævnes desuden liniedetektorer. Spalte 1-3 beskriver teknikkens stade vedrørende elektroniske billedbehandlingssystemer med elektriske kredsløb, der virker som lav- og højpasfiltre. De kan anvendes til at detektere karakteristika i billedet.EP EP 20,417 describes an electronic image processing system with a high pass filler that reduces noise. The publication also mentions line detectors. Columns 1-3 describe the state of the art regarding electronic imaging systems with electrical circuits that act as low and high pass filters. They can be used to detect characteristics in the image.
Filtrene virker på lignende måde som en matrixmaske, der finder karakteristika, såsom 15 vandrette og lodrette linier (én pixel brede) eller kantovergange.The filters work similarly to a matrix mask that finds features such as 15 horizontal and vertical lines (one pixel wide) or edge transitions.
Computer Vision, DanaH. Ballord, Christopher M. Brown, Printicc-Hall Inc. 1982, s. 65-88 og s. 123-131 beskriver forskellige teknikker ved billedbehandling (billedfiltrering, kantdetektion og kurvedetektion).Computer Vision, DanaH. Ballord, Christopher M. Brown, Printicc-Hall Inc. 1982, pp. 65-88 and pp. 123-131 describe various techniques of image processing (image filtering, edge detection and curve detection).
I manualen for et program, der kan udføre billedbehandling, er beskrevet, hvorledes 20 programmet kan udføre Sobelfiltrering og linietransformation, se IPA - 150 ITEX Programmer’s Manual, Post Number 47-515002-02, s. 5-49/5-50 og 5-83/5-85, september 1988, Imaging Technology Inc.The manual for a program that can perform image processing describes how the 20 program can perform Sable filtering and line transformation, see IPA - 150 ITEX Programmer's Manual, Post Number 47-515002-02, pages 5-49 / 5-50 and 5 -83 / 5-85, September 1988, Imaging Technology Inc.
De indre anatomiske dele, der fx kan ses på flækkede slagtekroppe, udgør en mere præcis kilde for automatiseret behandling af slagtekroppe end oplysninger om konturen. Fx kunne viden om 25 placeringen af de enkelte rygradsled give en betydeligt støne nøjagtighed i fastlæggelsen af snitpositioneme ved tredeling af flækkede slagtekroppe end det fx er muligt ved hjælp af konturen.The internal anatomical parts, which can be seen on, for example, broken carcasses, provide a more accurate source for automated carcass processing than contour information. For example, knowledge of the location of the individual backbone joints could provide a significantly greater accuracy in the determination of the cut positions for triplicate fractured carcasses than is possible, for example, by means of the contour.
3 DK 167462 B2 n^s j7 Problemet er imidlertid, at det er særdeles vanskeligt al tilvejebringe el videobillede, i hvilket den søgie anatomiske del kan identificeres og placeres med tilstrækkelig sikkerhed. Dets kan der være relativ dårlig kontrast i billedet mellem de enkelte dele, dels kan slagtedyrene have forskellig bygning, og dels kan der i en del tilfælde forekomme løse hinder eller kødtrævler, 5 som ved flækkeoperationen er trukket ind over de søgte anatomiske dele, og som derved udvisker eller skjuler deres tilstedeværelse.However, the problem is that it is extremely difficult to provide any video image in which the searchable anatomical part can be identified and placed with sufficient security. There may be a relatively poor contrast in the picture between the individual parts, partly the animals may have different structures, and in some cases loose obstructions or meat turbines may be present, which were pulled over the anatomical parts searched during the splicing operation and which thereby blurring or hiding their presence.
Det er formålet med den foreliggende opfindelse at tilvejebringe en fremgangsmåde, ved hvilken anatomiske dele kan lokaliseres med god sikkerhed under slagteriforhold.It is the object of the present invention to provide a method by which anatomical parts can be located with good security under slaughter conditions.
Formålet opfyldes af fremgangsmåden ifølge opfindelsen, der er ejendommelig ved, at 10 databehandlingen af det registrerede billede i beregningsenheden omfatter en retningsfiltrering, der fremhæver en forud bestemt retning i billede, at der for hver linie, der kan tegnes i billedet parallelt med den bestemte retning udføres en summation af billedelementernes lysværdier, og at de opnåede sumværdier anvendes til områdelokalisering.The object is fulfilled by the method according to the invention, characterized in that the data processing of the registered image in the calculation unit comprises a directional filtering which emphasizes a predetermined direction in the image, that for each line that can be drawn in the image parallel to the particular direction a summation of the light values of the pixels is performed and that the sum values obtained are used for area localization.
jj
Ved hjælp af den omhandlede rctningsfiltrering sker der en tydeliggørelse af de anatomiske 15 strukturer, der forløber i samme retning som den forud fastlagte retning i billedet, og ved hjælp af sumværdieme tydeliggøres positionen af de anatomiske dele yderligere. Positionen af de anatomiske dele, der er knyttet til sådanne strukturer, kan derved fastlægges med forøget sikkerhed, og på den måde kan det blive muligt at foretage en automatiseret behandling af kødemner på slagterier.With the aid of the directional filtration, the anatomical structures which extend in the same direction as the predetermined direction in the image are clarified, and the position of the anatomical parts is further clarified by means of the sum values. The position of the anatomical parts attached to such structures can thereby be determined with increased certainty, and in this way it can be possible to carry out an automated treatment of meat items in slaughterhouses.
20 Fremgangsmåden ifølge opfindelsen adskiller sig således på to væsentlige punkter fra teknikken, der beskrives i EP Bl 20.417, og tilvejebringer derved en markant forbedret positionsbestemmelse. Den fremhæver strukturer, der forløber i den forud bestemte retning 1 billedet, således al disse strukturer kommer til at fremstå tydeligere. Ved hjælp af del opnåede billede med fremhævede sturkturer dannes desuden en kurve over summerne af pixelværdieme 25 i linier, der er parallelle med den forud bestemt retning. Sumdannclsen reducerer støjen,Thus, the method according to the invention differs in two essential points from the technique described in EP Bl 20.417, thereby providing a markedly improved positioning. It highlights structures that extend in the predetermined direction in the image, thus all of these structures will become more apparent. In addition, by means of image obtained with highlighted structures, a curve of the sums of pixel values 25 is formed in lines parallel to the predetermined direction. Sum noise reduces noise,
Teknikken ifølge EP BI 20.417 kan ikke fremhæve (hele) skrukturer i form af anatomiske dele på ct kødemne, idet de angivne matricer er beregnet til at detektere kanten af strukturer.The technique according to EP BI 20.417 cannot emphasize (whole) structures in the form of anatomical parts on a meat subject, since the indicated matrices are intended to detect the edge of structures.
- ' Ή ·;' : · . ./fa &: • . ' i ‘ -1: DK 167462 02 .- 'Ή ·;' : ·. ./few &: • . 'i' -1: DK 167462 02.
, ..... ri..·.:,, i. 4 ,v;f , ' :'Η··;ί, ..... ri .. ·.: ,, i. 4, v; f, ':' Η ··; ί
Strukturernes midte bidrager ikke til det opnåede billede. Dciektion af kanter er højst usikker i billeder af kødemner. Matrixbehandlingcn tilvejebringer desuden både et positivt og el negativt bidrag, som i visse situationer kan ophæve hinanden, hvis der dannes en kurve over summerne af pixelværdier i linier, der løber i en bestemt retning.The center of the structures does not contribute to the image obtained. Edging of edges is highly uncertain in images of meat subjects. The matrix processing additionally provides both a positive and a negative contribution which in some situations can cancel each other if a curve is formed over the sums of pixel values in lines running in a particular direction.
5 Ved hjælp af ct videosystem har det ved den her omhandlede fremgangsmåde vist sig muligt at lokalisere visse anatomiske områder på svineslagtckroppe med tilstrækkelig nøjagtighed til, al el opskxringsanlæg, styret af de opnåede lokaliseringsdata, kan foretage automatisk tredeling af en slagtekrop i forende, midterstykke og skinke.5 By means of a video system, the method of the present invention has proved possible to locate certain anatomical areas of pig carcasses with sufficient accuracy that all electrical cutting systems, controlled by the obtained location data, can automatically triple a carcass at the front, center piece and ham.
Ved forsøg har skæringerne vist sig al blive udført lige så nøjagtigt som de i dug udførte 10 manuelle skæringer.In experiments, the cuts have been shown to be carried out just as accurately as the 10 manual cuts performed.
· -f '· -F '
Anatomiske områder, der har vist sig egnede ved tredeling af en slagtekrop, er rygradens led, s; /- især diskosserne mellem specifikke ryghvirvler samt diskossen mellem sidste ryghvirvel og første halehvirvel (rygradens knæk).Anatomical areas that have proved useful in triplicate a carcass are the backbone joints, s; / - especially the discus between specific vertebrae and the discus between the last vertebra and the first vertebra (vertebral crack).
Den omhandlede målemetode er fordelagtig ved, at den ikke kræver bemanding, og desuden er 15 den ikke destruktiv, dvs. der sker ikke nogen værdiforringelse af kødet som følge af målingen.The method of measurement in question is advantageous in that it does not require manning, and in addition it is non-destructive, ie. there is no impairment of the meat as a result of the measurement.
Det har vist sig, at den omhandlede fremgangsmåde kan udføres tilstrækkelig hurtigt til, at der kan måles og behandles fx 360 slagtekroppe i timen, hvilket er tilfredsstillende under slagteriforhold.It has been found that the method in question can be carried out sufficiently quickly to allow, for example, 360 carcasses per hour to be measured and processed, which is satisfactory under slaughterhouse conditions.
20 Krav 2-8 angiver foretrukne udførelsesformer af fremgangsmåden ifølge opfindelsen.Claims 2-8 specify preferred embodiments of the method of the invention.
Opfindelsen angår også ct anlæg til anvendelse ved behandling af et kødemne, jf. indledningen til krav 9. Det nye og særegne ved anlægget er, at beregningsenhedens databehandl i ngsdel omfatter en retningsfiltreringsproces, der fremhæver en forud bestemt retning i billedet og er indrettet til - for hver linie, der kan tegnes i billedet parallelt med den bestemte retning · derefter 25 at udføre en summation af billedelementernes lysværdier og anvende de opnåede sumværdier til områdelokaliscring.The invention also relates to a plant for use in processing a meat subject, cf. the preamble to claim 9. The new and unique feature of the plant is that the computing unit's data processing in part comprises a directional filtering process which emphasizes a predetermined direction in the image and is adapted to - for each line that can be drawn in the image in parallel with the particular direction · then 25 to perform a summation of the pixel light values and apply the obtained sum values to area localization.
- »·· ·... · ...- »·· · ... · ..
5 DK 167462 B25 DK 167462 B2
Del omhandlede anlæg kan med forbedret sikkerhed lokalisere anatomiske dele, der har strukturer, som forløber parallelt med den fastlagte retning. På den måde kan den af lokaliseringen styrede, efterfølgende behandling af kødemnet ske med forøget nøjagtighed.Partially referred installations can, with improved certainty, locate anatomical parts having structures that run parallel to the determined direction. In this way, the subsequent treatment of the meat can be controlled by the locator with increased accuracy.
Krav I0-I2 angiver foretrukne udførelsesformer af anlægget ifølge opfindelsen.Claims I0-I2 specify preferred embodiments of the plant according to the invention.
5 Opfindelsen forklares nærmere i det følgende under henvisning til tegningen, hvori - Hg. I viser et opmålingsanlæg ifølge opfindelsen til brug ved partering af flækkede svincslagtekroppe, - fig. 2 et videobillede oplaget af skinkeområdel med indtegnet rygradxkurve, • fig. 3 et databehandlet deludsnit af billedet, 10 - fig. 4 samme udsnit efter remingsfiltrering, : n' - fig. 5 en kurve over sumværdier og - fig. 6a-d databehandlede kurver.The invention is explained in more detail below with reference to the drawing, wherein - Hg. Fig. 1 shows a metering system according to the invention for use in partitioning of split pig carcasses; 2 shows a video image of the ham area portion with the backbone curve plotted; FIG. 3 is a partial sectional view of the image; FIG. 4 shows the same section after belt filtration: n '- fig. 5 is a graph of sum values; and FIG. 6a-d data-processed curves.
Opmålingsanlæggct i fig. 1 omfatter en transportør med et sort transportbånd i. på hvilket i? flækkede svineslagtekroppe 2 er anbragt med sværsiden nedad. Kroppene fremføres løbende 15 i retning af pilen P med ryggen forrest. Over båndet er anbragt tre CCD-videokameraer 3, hvis billedfelter dækker hhv. forenden, skinkedelen og bagbenet af kroppen. Forende- og skinkekameraeme er forsynet med grønfiltre, mens bagbenskameraet har el rødfilter. Tre lyskilder 4 belyser slagtekroppen. To af lyskildemes lysretning danner en vinkel på ca. 45 grader med vandret plan, således at der dannes et skyggeområde i slagtekroppens hulhed, og den 20 ene side af skyggeområdet kommer til at grænse umiddelbart op (i) slagtekroppens rygrad. Den tredic lyskilde lyser direkte ned på slagtekroppen under en vinkel på 90 grader.The surveying system of FIG. 1 comprises a conveyor with a black conveyor belt in. On which i? split pig carcasses 2 are positioned face down. The bodies are continuously advanced 15 in the direction of the arrow P with the back at the front. Above the tape are three CCD video cameras 3, whose image fields cover respectively. front end, buttocks and hind legs of the body. The front and butt cameras are fitted with green filters, while the rear leg camera has an electric red filter. Three light sources 4 illuminate the carcass. Two of the light sources' light direction forms an angle of approx. 45 degrees with a horizontal plane, so that a shadow area is formed in the carcass cavity, and the 20 one side of the shadow area will immediately adjoin (i) the carcass backbone. The thirty light source glows directly onto the carcass at an angle of 90 degrees.
En "framegrabber" 5 er forbundet til hvert af kameraernes video-udgang. Framegrabberen lagrer et videobillede, når der fra en central beregningsenhed 6 afgives et elektronisk udløsersignal.A "frame grabber" 5 is connected to each of the cameras video output. The frame grabber stores a video image when an electronic trigger signal is output from a central computing unit 6.
Det kan fx initieres af et ved båndet anbragt lysrelæ 7, der detekterer tilstedeværelse af en 25 slagtekrop i kameraets billedfelt eller ved et signal fra fremføringsstyringen.It may, for example, be initiated by a tape relay 7 located on the tape, which detects the presence of a carcass in the camera's image field or by a signal from the feed control.
6 DK 167462 B26 DK 167462 B2
Beregningsenheden 6 omfalter en styre- og beregningsdel, der henter duta fra den aktuelle grabber og behandler dem efter en fonid fastlagt proces. Herved kan om ønsket anvendes andre måledata, fx oplysning om den aktuelle vægt eller kpdApæktykkelser målt med en sonde.The computing unit 6 wraps a control and calculation part which retrieves the duta from the current grabber and processes them according to a phonid determined process. If desired, other measurement data can be used, for example, information on the current weight or kpdA thicknesses measured with a probe.
Processen resulterer i et signal, der er et udtryk for skærepositionen. Det anvendes som-5 styresignal til automatisk indstilling af en efterfølgende båndsav til korrekt skæreposition i forhold til de anatomiske dele.The process results in a signal that reflects the cutting position. As a control signal, it is used to automatically adjust a subsequent band saw to the correct cutting position relative to the anatomical parts.
I det følgende er nærmere beskrevet behandlingen af skinkebilledet, og fastlæggelsen af skinkesnitlets referencepunkt (den skarpe overgang mellem rygrad og hale, der undertiden er. betegnet med "knækket").The following describes the treatment of the ham image and the determination of the point of reference of the ham cut (the sharp transition between the spine and the tail, sometimes referred to as "cracked").
10 Søgning efter rygrad På grund af den særlige skrå belysning af kroppen er rygraden fuldt belyst, mens kødområdet, der støder op til den ene side af rygraden, ligger i skygge, se fig. 2.10 Searching for a backbone Due to the special oblique illumination of the body, the backbone is fully illuminated, while the meat area adjacent to one side of the backbone is in shadow, see fig. 2nd
Det optagne og lagrede billede er opbygget af billedelementer (pixels), der er anbragt i el rasler i rækker og søjler med ensartet indbyrdes afstand. I de 20 første søjler i billedet søges det 15 område, hvori findes elementerne med de mindste lysværdicr inden for et område på 15*20 pixels. Fra dette skyggeområde findes rygraden over et 10 pixels bredt område som det sted, hvor der er den største positive ændring (gradient) i lysvxrdien, og hvor de gennemsnitlige lysværdier før gradient har en given størrelse.The recorded and stored image is made up of image elements (pixels) placed in electric rattles in rows and columns of uniform spacing. In the first 20 columns of the image, 15 areas are searched, in which the elements with the smallest light values are found within a range of 15 * 20 pixels. From this shadow area, the backbone over a 10 pixel wide area is found as the place where there is the largest positive change (gradient) in the light value and where the average light values before the gradient have a given size.
Når dette sled er fundet søges rygraden inden for et interval på +AI5 pixels. Rygradspunktet 20 defineres som del punkt, hvor der er den største positive gradient, og hvor de gennemsnitlige pixelværdier før gradienten har en given størrelse.When this sled is found, the spine is searched within a range of + AI5 pixels. The backbone point 20 is defined as the part point where there is the largest positive gradient and where the average pixel values before the gradient have a given size.
Når rygradspunktet er fundet søges næste punkt efter samme kriterier som skitseret ovenfor.When the backbone point is found, the next point is searched for the same criteria as outlined above.
Findes koordinaten ikke, sættes den lig med den foregående koordinat. Koordinaterne for rygraden midies, inden de anvendes i de følgende beregninger. De fundne punkter er indtegnet 25 som en rygradskurve i fig. 2.If the coordinate is not found, it is set equal to the previous coordinate. The coordinates of the spine are midies before being used in the following calculations. The points found are plotted 25 as a backbone curve in FIG. 2nd
• · ·· ·· -·: V.: . >: DK 167462 B2 ;· ;'v\ , 7 ;;V' i Beregning af foreløbig position for "knæk" ' .:.¾ ^ :• · ·· ·· - ·: V.:. >: DK 167462 B2; ·; 'v \, 7 ;; V' in Calculating preliminary position for "crack" '.:. ¾ ^:
Knækkets foreløbige position bestemmes som den position, hvor ændringen i rygradskurvens krumning er størst. Den er markeret med linien Ύ" på fig. 2.The preliminary position of the tear is determined as the position where the change in the curvature of the spine curve is greatest. It is marked with the line Ύ "in Fig. 2.
Billedbehandling af rygraden 5 Ved hjælp af rygradskurven dannes der et billedudsnit på 50 gange 300 pixels, indeholdende rygraden og "knækket". Udsnittets øvre kant svarer til kurven, Der foretages en korrektur for forvrængningen, der fremkommer som følge af opretningen af udsnittet.Image processing of the spine 5 Using the spine curve, an image section of 50 times 300 pixels is created, containing the spine and "cracked". The upper edge of the slice corresponds to the curve. A correction is made for the distortion that results from the alignment of the slice.
Det opnåede delbillede af rygraden, der er vist i Hg. 3, underkastes en retningfiltrering, der fremhæver strukturer vinkelret på rygraden, og desuden eventuelt strukturer af en bestéiiil 10 bredde. Hertil anvendes en talmatrix af følgende udseende:The obtained image of the spine shown in Hg. 3, a directional filtration which emphasizes structures perpendicular to the spine and, optionally, structures of a best width 10 is subjected. For this, a number matrix of the following appearance is used:
-1 I 2 I-I -112 1-1 -2 2 4 2-2 -1 I 2 I -I 15 -I ! 2 I -I-1 I 2 I-I -112 1-1 -2 2 4 2-2 -1 I 2 I -I 15 -I! 2 I -I
En 5*5 talmatrix hentes Fra pixel-lysværdicme i det ene hjørne af dclbilledct. Skalarproduktet uf de to matricer udregnes af de fundne talværdier og indsættes i stedet for de oprindelige pixel -værdieri det lagrede billedudsnit. En ny 5*5 talmatrix dannes af pixelværdieme, der ligger et enkelt billedelement til højre for den første talmatrix. Produktet dannes af denne matrix og den 20 ovenfor viste talmatrix, hvorefter fundne værdier indsættes i det lagrede billedudsnit, i stedet for de oprindelige værdier. På denne måde fortsættes til billedudsnittets højre kant. Der flyttes et billedelement op, hvorefter proceduren gentages. Når hele rækken af pixels i dette nivtfati er behandlet, flyttes igen et billedelement op. og således fortsættes indtil hele billedudsnittets pixel-lysværdier har været databehandlet med den retningsfremhævende matrix. Billedet har 25 nu det i fig. 4 viste udseende, i hvilket diskosplademc mellem rygradens led træder tydeligere frem end på originalbilledet (fig. 3).A 5 * 5 number matrix is retrieved From pixel light value in one corner of dclbilledct. The scalar product of the two matrices is calculated from the found numerical values and inserted instead of the original pixel values in the stored image section. A new 5 * 5 number matrix is formed by the pixel values, which is a single image element to the right of the first number matrix. The product is formed by this matrix and the number matrix shown above, after which values found are inserted into the stored image section, instead of the original values. In this way, continue to the right edge of the image section. An image element is moved up and the procedure is repeated. Once the entire row of pixels in this level fat has been processed, an image element is moved up again. and thus is continued until the pixel light values of the entire image section have been processed with the directional highlighting matrix. The image now has that of FIG. 4, in which the disc space between the spine joints appears more clearly than in the original image (Fig. 3).
8 DK 167462 B28 DK 167462 B2
En fem pixels bred kant skæres af billedudsniitct hele vejen rundt, hvorefter der foretages en simpel sammenlægning af pixel-lysværdierne i hver af billedets søjler af pixels. Sumkurven er vist i fig. 5.A five pixel wide border is cut by the image layout all around, then a simple merging of the pixel light values into each of the image's columns of pixels is made. The sum curve is shown in FIG. 5th
Behandling af sumkurve 5 I fig. 6u-b er vist to sumkurver, der ofte forekommer i praksis. Det søgte "knæk" vides at ligge i nærheden af position 50, men for at bestemme "knækkets" nøjagtige position må kurverne databehandles, hvorved det udnyttes, at afstanden mellem pladerne i rygraden i det store og hele er ens inden for samme individ.Treatment of sum curve 5 In fig. 6u-b are shown two sum curves that often occur in practice. The "crack" sought is known to be in the vicinity of position 50, but in order to determine the "crack" exact position, the curves must be computed, taking advantage of the fact that the distance between the plates in the spine is broadly the same within the same individual.
På fig. 2 kan det anes, at diskospladcmc fremtræder som hvide striber mellem mørke ben. Ved 10 anvendelse af gradicntfiltrcring af kurverne i Tig. 6a-b opnås et positivt signal ved starten af en stribe og ct negativt signal ved stribens afslutning. Pladen antages at ligge dér hvor kurven på sin vej fra top til bund krydser gennemsnitsniveauet.In FIG. 2, it may be considered that the disco plate cmc appears as white stripes between dark legs. Using gradient filtering of the curves in Tig. 6a-b, a positive signal is obtained at the start of a strip and ct negative signal at the end of the strip. The plate is assumed to be where the curve on its path from top to bottom crosses the average level.
På kurverne i fig. 6a-b udføres først transformationen p2(x)=abs(p(x-3)-p(x+3)) 15On the curves of FIG. 6a-b, first the transformation p2 (x) = abs (p (x-3) -p (x + 3)) is performed
Efter transformationen har kurverne det i fig. 6c-d viste udseende.After the transformation, the curves have the 6c-d.
p2(x) cr stor, når p(x-3) er stor og/eller p(x+3) er lille. Det ses, at der er kommet en spids de steder, hvor der sker el stort fald over 6 enheder (pixels) i vandret billcdrelning.p2 (x) cr is large when p (x-3) is large and / or p (x + 3) is small. It is seen that there has been a spike in the places where there is a major fall over 6 units (pixels) in horizontal image rotation.
Der forekommer store fald ved pladerne og små fald mange steder som følge af støj. Hvis el 20 fald cr dobbelt så stort som ct andet, bør dets betydning ikke blot være dobbelt så stort, men fremtræde endnu kraftigere. For at reducere støjen anvendes følgende ikke-lineærc transformation af p2(x): p3(x)=p2(x)3There are large falls at the plates and small falls in many places due to noise. If el 20 falls cr twice as large as ct other, its significance should not only be twice as large, but appear even more powerful. To reduce the noise, the following non-linear transformation of p2 (x) is used: p3 (x) = p2 (x) 3
Pladernes positioner er nu temmeligt tydelige med en afvigelse svarende til højst én pixel.The positions of the plates are now fairly clear with a deviation equal to at most one pixel.
øv.: · ; -·:· * V 'j. ? : V i! ? ; 4 DK 167462 B2 9Ex .: ·; - ·: · * V 'j. ? : V i! ? ; 4 DK 167462 B2 9
For at fremhæve svage spidser i støjsvage områder foretages en lokal omnormering ved foldning med funktionen k: k(s)=exp(-abs(s/a)) p4(x)=k o p3(x)= J k(s) p3(x-s) ds 5To emphasize weak points in quiet areas, a local renaming is done by folding with the function k: k (s) = exp (-abs (s / a)) p4 (x) = ko p3 (x) = J k (s) p3 (xs) ds 5
Dc opnåede kurverer vist i fig. 6 e-f.The obtained curves shown in FIG. 6 e-f.
Beregning af diskospladernes positionCalculation of the position of the discs
Funktionen p4(x) matches med en skabelon med 6 ækvidistante kanter, der skal repræsentere positionerne af diskospladerne mellem ryghvirvlerne. Ideen er, at kurven efter passende 10 forskydning og skalering skal ligne en gennemsnitskurve, der således tjener som prototype eller skabelon. Når kurven tolkes, strækkes og skubbes den inden for visse på forhånd fastsatte grænser. Den transformation, der giver del største overlap med skabelonen, er den korrekte.The function p4 (x) is matched with a template with 6 equidistant edges to represent the positions of the discs between the vertebrae. The idea is that the curve after appropriate 10 offset and scaling should be similar to an average curve, thus serving as a prototype or template. When interpreted, the curve is stretched and pushed within certain predetermined limits. The transformation that causes some overlap with the template is the correct one.
Overlappets størrelse er et udtryk for, hvor sikkert profilen er blevet tolket.The size of the overlap reflects how secure the profile has been interpreted.
Den anvendte beregningsprocedure foretager således en mønster-genkendelse baseret på en 15 helhedsvurdering.Thus, the calculation procedure used makes a pattern recognition based on an overall assessment.
Det fastlægges på forhånd, inden for hvilke grænser den første og sidste plade skal søges. Inden for grænserne gennemregnes alle mulige positioner, idet skabelonen strækkes mere eller mindre, og ud fra en helhedsvurdering findes skabelonen, der giver størst overensstemmelse med kurven. Positionen af diskospladcme, herunder den i Fig. 6b, 6d og 6f usynlige plade ved 20 overgangen mellem rygrad og hale (’'knækket"), er dermed blevet bestemt med stor sikkerhed.It is determined in advance within which limits the first and last record should be sought. Within the boundaries, all possible positions are calculated, the template being stretched more or less, and on the basis of an overall assessment, the template is found that gives the most conformity to the curve. The position of the disc disk, including that of FIG. 6b, 6d and 6f invisible plate at the transition between the spine and tail ('' cracked '), has thus been determined with great certainty.
Pladernes positioner er indtegnet i fig. 2 som lodrette streger. Den anvendte foreløbige position af "knækket" kan nu erstattes med den mere præcise position, der er markeret med ”K".The positions of the plates are plotted in FIG. 2 as vertical lines. The preliminary position of the "crack" used can now be replaced by the more precise position marked "K".
Hele den ovenstående procedure er gennemført automatisk i beregningsenheden ved elektronisk databehandling af det lagrede billede af slagtekroppens skinkedel. På grundlag af den fundneThe entire above procedure is performed automatically in the calculator by electronically processing the stored image of the carcass ham. On the basis of the found
Claims (7)
Priority Applications (6)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DK199101504A DK167462B2 (en) | 1991-08-23 | 1991-08-23 | Method and plant for use in treating a meat subject |
| GB9217299A GB2258916B (en) | 1991-08-23 | 1992-08-14 | Method of and apparatus for individually treating pieces of meat |
| FR9210081A FR2680449B1 (en) | 1991-08-23 | 1992-08-17 | METHOD AND INSTALLATION FOR INDIVIDUAL TREATMENT OF MEAT PARTS. |
| NL9201472A NL9201472A (en) | 1991-08-23 | 1992-08-18 | METHOD AND APPARATUS FOR USE IN THE INDIVIDUAL TREATMENT OF PIECES OF MEAT. |
| IE260392A IE922603A1 (en) | 1991-08-23 | 1992-08-21 | Method and apparatus to be used by individual treatment of¹pieces of meat |
| DE4228068A DE4228068A1 (en) | 1991-08-23 | 1992-08-24 | METHOD AND DEVICE FOR THE INDIVIDUAL TREATMENT OF MEAT PIECES |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DK150491 | 1991-08-23 | ||
| DK199101504A DK167462B2 (en) | 1991-08-23 | 1991-08-23 | Method and plant for use in treating a meat subject |
Publications (4)
| Publication Number | Publication Date |
|---|---|
| DK150491D0 DK150491D0 (en) | 1991-08-23 |
| DK150491A DK150491A (en) | 1993-02-24 |
| DK167462B1 DK167462B1 (en) | 1993-11-01 |
| DK167462B2 true DK167462B2 (en) | 1999-11-01 |
Family
ID=8105530
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| DK199101504A DK167462B2 (en) | 1991-08-23 | 1991-08-23 | Method and plant for use in treating a meat subject |
Country Status (6)
| Country | Link |
|---|---|
| DE (1) | DE4228068A1 (en) |
| DK (1) | DK167462B2 (en) |
| FR (1) | FR2680449B1 (en) |
| GB (1) | GB2258916B (en) |
| IE (1) | IE922603A1 (en) |
| NL (1) | NL9201472A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001015538A3 (en) * | 1999-08-27 | 2001-07-19 | K J Maskinfabriken As | Arrangement for laying-down alf carcasses, and based automatic cutting-up system in connection therewith |
| US6860804B2 (en) | 1999-08-27 | 2005-03-01 | Kj Maskinfabriken A/S | Laying-down system and vision-based automatic primal cutting system in connection therewith |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5793879A (en) * | 1992-04-13 | 1998-08-11 | Meat Research Corporation | Image analysis for meat |
| GB9510171D0 (en) * | 1995-05-19 | 1995-07-12 | Univ Bristol | A method of and apparatus for locating a spine in a half-carcass |
| DE60129556T2 (en) * | 2000-05-30 | 2008-04-03 | Marel Hf. | METHOD FOR MEAT PROCESSING AND FOR EDITING INFORMATION |
| DE102007017899B4 (en) * | 2007-04-13 | 2017-02-16 | Innotech Ingenieursgesellschaft Mbh | Apparatus and method for cutting food material |
| CN112384768B (en) * | 2018-05-04 | 2022-09-20 | 艾克斯波特西溶液公司 | Scale for determining the weight of a living being |
| DE102020006482A1 (en) | 2020-10-14 | 2022-04-14 | Innotech Ingenieursgesellschaft Mbh | Device for cutting agricultural products and central processing unit with at least one data memory for controlling the device |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE2728913A1 (en) * | 1977-06-27 | 1979-01-18 | Hans Breitsameter | METHOD AND DEVICE FOR CLASSIFYING MEAT |
| DK157380C (en) * | 1986-11-06 | 1991-08-12 | Lumetech As | METHODS OF OPTICAL, BODY-FREE MEASUREMENT OF MEAT TEXTURE |
| FR2608899B1 (en) * | 1986-12-29 | 1990-02-23 | Simonet Andre | PROCESS FOR QUALIFYING CARCASSES OF BUTCHER ANIMALS, AND CORRESPONDING INSTALLATION |
| DK676487A (en) * | 1987-12-22 | 1989-06-23 | Slagteriernes Forskningsinst | PROCEDURE FOR DETERMINING QUALITY CHARACTERISTICS OF INDIVIDUAL CREATURE GENERATOR AND PLANT FOR USE IN DETERMINING THE PROPERTIES |
-
1991
- 1991-08-23 DK DK199101504A patent/DK167462B2/en not_active IP Right Cessation
-
1992
- 1992-08-14 GB GB9217299A patent/GB2258916B/en not_active Expired - Fee Related
- 1992-08-17 FR FR9210081A patent/FR2680449B1/en not_active Expired - Fee Related
- 1992-08-18 NL NL9201472A patent/NL9201472A/en not_active Application Discontinuation
- 1992-08-21 IE IE260392A patent/IE922603A1/en not_active IP Right Cessation
- 1992-08-24 DE DE4228068A patent/DE4228068A1/en not_active Ceased
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001015538A3 (en) * | 1999-08-27 | 2001-07-19 | K J Maskinfabriken As | Arrangement for laying-down alf carcasses, and based automatic cutting-up system in connection therewith |
| US6860804B2 (en) | 1999-08-27 | 2005-03-01 | Kj Maskinfabriken A/S | Laying-down system and vision-based automatic primal cutting system in connection therewith |
Also Published As
| Publication number | Publication date |
|---|---|
| DK150491D0 (en) | 1991-08-23 |
| DK167462B1 (en) | 1993-11-01 |
| GB2258916B (en) | 1995-08-02 |
| NL9201472A (en) | 1993-03-16 |
| FR2680449B1 (en) | 1994-05-20 |
| FR2680449A1 (en) | 1993-02-26 |
| IE922603A1 (en) | 1993-02-24 |
| DK150491A (en) | 1993-02-24 |
| GB2258916A (en) | 1993-02-24 |
| GB9217299D0 (en) | 1992-09-30 |
| DE4228068A1 (en) | 1993-03-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105513077B (en) | A kind of system for diabetic retinopathy screening | |
| US5944598A (en) | Method and apparatus for using image analysis to determine meat and carcass characteristics | |
| AU665683B2 (en) | Image analysis for meat | |
| Mahfouz et al. | Fast localization of the optic disc using projection of image features | |
| Grau et al. | Automatic localization of cephalometric landmarks | |
| US6055326A (en) | Method for orienting electronic medical images | |
| DK167462B2 (en) | Method and plant for use in treating a meat subject | |
| US20030072472A1 (en) | Image analysis systems for grading of meat, predicting quality of meat and/or predicting meat yield of an animal carcass | |
| US20070211944A1 (en) | Apparatus for detecting feature point and method of detecting feature point | |
| WO2004044829A1 (en) | System for uniquely identifying subjects from a target population | |
| CA2001429A1 (en) | Pedotopography apparatus and method using moire fringe analysis to measure foot shapes | |
| BR112021016335A2 (en) | PORTIONING/TRIMMING OF PRIMARY RIB CUTS | |
| CA2085124A1 (en) | Automatic carcass grading apparatus and method | |
| DK160903B (en) | PROCEDURES FOR LOCAL AREA LOCATION, ISRAEL FISH, BY LIGHTING | |
| EP1827116B1 (en) | Data acquisition for classifying slaughtered animal bodies and for their qualitative and quantitative determination | |
| Savolainen et al. | Application of machine vision to assess involved surface in patients with psoriasis | |
| DE102004047773A1 (en) | Method for determining physiological quantities of an animal carcass | |
| WO1991014180A1 (en) | Evaluating carcasses by image analysis and object definition | |
| DE4131556A1 (en) | Half carcass conformation determn. - by image processing using chine as datum point | |
| EP1174034A1 (en) | Method for trimming pork bellies | |
| JP2020183876A (en) | Feature point recognition system and workpiece processing system | |
| CN1882248A (en) | Method for determining the quality of a slaughtered animal carcass and quantities therein | |
| Petrascu et al. | Automatic on‐line electronic portal image analysis with a wavelet‐based edge detector | |
| Costa et al. | Innovative automated landmark detection for food processing: the backwarping approach | |
| Rahayu et al. | Fetal head and femur detection from USG image to estimate gestational age |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| B1 | Patent granted (law 1993) | ||
| PPF | Opposition filed | ||
| B2 | Patent amended (law 1993) | ||
| PBP | Patent lapsed |
Country of ref document: DK |