Imaging a hazardous zone within a skinner

Abstract

For a hand-fed skinner machine, improvements as described to the optical system tend to render a gloved hand appearance occurring anywhere along a hazardous volume adjacent the blade, as viewed from either end, independent of position by reducing proximity distortion. Return mirrors are used to lengthen the optical path within the tunnel and the physical tunnel size may be reduced.

Description

[0001] Title: Imaging a hazardous zone within a skinner.

[0002] Field:

[0003] This invention relates to improvements to machinery used after slaughter to separate layered parts of a carcass; such as skin from underlying meat, by forcing it against a fixed blade. That machinery is called “a skinner”, a “derinder”, a “denuder”, or “a flaying machine”. More particularly this invention relates to a class of skinners or derinders that are used by an operator. Protection of a human user; in particular the hands, from damage inadvertently caused by the blade is required. This invention relates to improved arrangements for optical sensing of a hand when in a dangerous place.

[0004] Definitions:

[0005] Bundle of image-forming rays of light

[0006] This concept is abbreviated in the following text to “image light”.

[0007] Electronic camera

[0008] This name refers to cameras used in machine-vision applications, typically operated to send a constant stream of captured images to a processor. Currently preferred Teledyne “Blackfly” colour cameras are described at https: / / www.teledynevisionsolutions.com / en- 150 / products / blackfly-s-usb3 / ?model=BFS-U3-04S2C- CS&vertical=machine%20vision&segment=iis

[0009] Hazardous volume.

[0010] The hazardous volume of a skinner machine is the volume in which an operator’s hand is most likely to receive trauma from the blade of the skinner. It is located over the top of, and behind the driven feeder roller and in particular is just in front of, or includes an edge of, a fixed blade and is situated across the rear of an operator work area. A typical blade has a length of 500 mm, setting the minimum length of the hazardous volume. The height and depth of the hazardous volume are calculated according to an achievable skinner stopping time after detecting a finger within the hazardous volume before injury arises by touching the blade, along with a safety margin. For optical sensing, the hazardous volume is viewed horizontally by at least one electronic camera at each end, across the working surface of the skinner.

[0011] Prior Art:

[0012] The Applicant’s series of imaging optical sensing systems that are adapted to optically detect an operator's hand using images, close to, and especially in dangerous proximity to a cutting blade typically of a meat-processing machine began with WO2017196187A1 for a bandsaw. At least one electronic or machine-vision camera looks down on to the bandsaw table and a controller analyses image data and if necessary causes the bandsaw blade to be stopped within 20 milliseconds. The Applicants later described a skinner or derinder including an optical safety protection apparatus in WO2021 / 091399 Al (herein ‘399). In ‘399, electronic cameras are directed horizontally along the skinner blade from either side through an apparatus that extended the effective distance between the camera lens and the viewed object.

[0013] Bergman et al (US 2008 / 0079590) describes non-imaging protection for an operator when using a skinner. A processor determines the position of a physical sensor, worn on the operator’s body, that senses generated fields and sends an output to the processor. A magnetic field is generated sequentially in several axes.

[0014] Townsend (US 5,122,091) describes an electric sensor for a skinning machine, using a conductive glove.

[0015] Problem to be solved:

[0016] Selection of an imaging optical sensing system for hand proximity must overcome a skinner characteristic, that the working area will be concealed from above by parts of the object being processed. That is unlikely for a bandsaw, for instance, where a view from above a work table is satisfactory.

[0017] An end view is required along the length of the blade of the skinner, and from one or both ends, in order to monitor the hazardous volume. The position of one or more identifiable coloured objects (hands) in relation to one or more hazards (blade), inside a hazardous volume is monitored in real-time by at least one electronic camera. That allows a processor that is examining a series of images taken from the cameras to take action if required.

[0018] This document relates to imaging improvements that tend to render all views of a glove within the hazardous volume along a skinner blade as relatively similar, from an image- processing point of view, so that “glove sensitivity” is relatively constant over the entire hazardous volume.

[0019] There is a need to provide a large effective distance between the hazardous volume and the or each camera so that images of nearest points and farthest points have similar attributes such as size, while maintaining reasonable focus and a spreading the image over the CMOS sensor surface. Preferably, a colour of any one pixel from a coloured glove stays above the recognition threshold and is not mingled with colours from nearby background pixels.

[0020] The particular problem to be solved is seen as making “glove sensitivity” more constant along the hazardous volume while reducing the size of the physical tunnels described in the Applicant’s past publications. The present application improves on the Applicant’s past inventions.

[0021] Summary of Invention

[0022] In a first broad aspect, the invention provides a skinner machine having operator protection for hands approaching a defined, elongated hazardous volume along the blade of the skinner, said hazardous volume ending near an adjacent window of a sealed tunnel, each window admitting a bundle of image-forming rays of light (herein ‘image light’) derived from within the hazardous volume and having an optical axis to a first mirror that, when in use, deflects the image light to a lens of at least one camera placed further along the tunnel, wherein a second or return mirror is located at the far end of the tunnel and the lens of at least one machine-vision camera receives the image light over an extended path.

[0023] In a first aspect, the return mirror is a plane mirror disposed perpendicular to the optical axis and the camera is located between the diverting mirror and the return mirror and along the optical axis.

[0024] In a second aspect, the return mirror is a plane mirror tilted away from the optical axis so that, when in use, the return mirror directs the image light to a camera located to one side of, and not obstructing light from the diverting mirror.

[0025] In a third aspect, the return mirror is comprised of an array of at least two plane facets tilted at an angle to the optical axis so that, when in use, at least two separate points of view into the hazardous volume are focused as side by side areas upon a sensor within the or each camera. In a fourth aspect, the or each camera is located above a partially reflective diverting mirror; preferably having a coating having an about 50% reflectivity; the camera is located on the axis of the light and receives the light from the return mirror after passage through the diverting mirror and through the partially reflective diverting mirror for a second time.

[0026] Notices:

[0027] In this specification, references to one tunnel assembly should be interpreted as embracing a mirror image of the assembly that is normally provided on the other side of the skinner machine.

[0028] In this specification, reference numerals, including reference numerals appearing in the Summary of Invention, are provided for clarification only and are not intended to restrict the scope of the invention to the particular embodiments of the components in conjunction with which the reference numerals are used.

[0029] Throughout this specification unless the text requires otherwise, the word "comprise" and variations such as "comprising" or "comprises" will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps. Each document, reference, patent application or patent cited in this text is expressly incorporated herein in their entirety by reference. Reference to cited material or information cited in the text should not be understood as a concession that the material or information was part of the common general knowledge or was known in New Zealand or in any other country.

[0030] List of figures

[0031] Fig 1 : is an elevation sectional diagram of the prior-art arrangement for optical surveillance of a hazardous volume within a skinner.

[0032] Fig 2: illustrates a tunnel having a plane return mirror at the far end of the tunnel from the window, extending the effective viewing distance.

[0033] Fig 3: illustrates a tilted plane return mirror at the end of the tunnel, avoiding obstruction of the image light.

[0034] Fig 4: illustrates a faceted return mirror at the end of the tunnel, simulating two viewpoints with a single electronic camera.

[0035] Fig 5: is a surface view of mirror 601 showing two facets. Fig 6: illustrates a plane return mirror and a partially reflective diverting mirror with the electronic camera and its lens located above the mirror, increasing the effective viewing distance.

[0036] Preferred embodiments:

[0037] Attempts to apply optical identification of a hazard within a skinner machine, useful for obtaining advance warning of imminent injury so that the machine can be stopped in time, require the generation and utilization of image light emerging from within a hazardous volume located along the length of the fixed knife, by an electronic camera. Image light arises from objects within or near the hazardous volume as a result of scatter, reflection or fluorescence of light introduced into the hazardous volume. It includes a background of the fleshy material being skinned, and may include parts of coloured gloves worn by the operator. The hazardous volume is accessible in a horizontal direction from one or both ends. A view from above, as for a bandsaw, is not feasible for a skinner or a derinder since the item being skinned is likely to cover the hazardous area along the knife edge.

[0038] Please refer to Fig 1 which is a diagram of the prior-art symmetrical arrangement used in the Applicant’s earlier skinners. It is an elevation view of a section across the elongated hazardous volume 101, showing the left and right bent tunnels 102L and 102R. Note the symmetrical placement of the tunnels; one at each end of the elongated hazardous volume, where operator protection is intensively applied. An inherent risk of optical protection for a skinner is that part of an object being skinned, within the hazardous volume, may block image light from reaching the electronic cameras, so the second view from the opposite side is useful at least as a backup. That elongated volume lies along the length of the blade (not shown; typically 500 mm long) and preceding (in terms of item and hand movement) the blade. 106 represents a work surface top. At the left and at the right of that hazardous volume are windows 107 providing a transparent seal for each tunnel. A 45 degree slanted diverting mirror 103 near the window of each tunnel diverts the image light downward along an optical axis 108, bent to better fit the optical apparatus within the profile of the skinner.

[0039] In these examples, image light derived from the hazardous volume 101 is brought to a focus by a lens 104 on the image sensor chip of the or each electronic camera 105 and subsequent software within processor 112 is configured in order receive video signals 110 and 111 from the cameras. The processor is instructed to report if a gloved finger having a contrasting hue or brightness to that of the background has been detected within the predetermined hazardous volume. The report will stop the motor 114 of the skinner via connecting line 113. Display device 116 may be used to warn the operator following detection of a hand approaching the hazardous volume. Four cameras may be used.

[0040] This document relates to imaging improvements as compared to the Applicant’s prior art intended to render all views of a glove appearing within the hazardous volume at any position along a skinner blade as relatively similar in size, from an image-processing point of view. Thid could be called “proximity distortion”. The rationale for the developments of the present invention is that to locate an electronic camera close to an end of the hazardous volume, just inside window 107 would cause close objects inside the hazardous volume to be imaged as much larger than far objects.

[0041] The elongated volume is always viewed from its ends. Close objects subtend a larger solid angle to the or each imaging lens. That would make the sensing process more sensitive to close objects. Suppose the lens was 50 mm from the nearest end. Then, given that the hazardous volume is the same length (500 mm) as the feeder roller, the ratio of far end / near end distance is 550 / 50 or 11:1 which is detracts from capturing and recognising a clear image of a transgression anywhere along the hazardous volume. Trends in the following Table show that a distance of at least 250 mm is desirable, for a 500 mm knife.

[0042] Table 1.

[0043] A. 50 mm from lens to nearest end of hazardous volume: 11:1 ratio

[0044] B. 250 mm: 3:1

[0045] C. 500 mm: 2:1

[0046] As the distance 108 from the object to the image increases, an increasingly long focal length imaging lens 104 is needed to enlarge the image on to a typical electronic camera’s imaging sensor. It is known that long lenses have a relatively restricted depth of field within which an object is focused relatively sharply. Depth of field increases for a stopped-down lens aperture. Selected cameras 105 may be chosen for a physically small imaging surface or CMOS sensor chip. This document relates to imaging improvements that tend to render all views of a glove within the hazardous volume along a skinner blade as sufficiently free from proximity distortion, from an image-processing point of view.

[0047] Example 1. Fig 2 shows use of an added plane return mirror 201, oriented normal to an optical axis of the image light 108 at the far end of the tunnel from the diverting mirror 103 so that the image light is reflected back again up the tunnel to the camera 105, now rotated by 180 degrees so that its lens 104 is directed at the return mirror. For this option the selected camera preferably is relatively small in cross-sectional area (which is, for the preferred examples, 29 x 29 mm) so that it does not cause undue obstruction to the image light. However, a stopped-down lens tends to accept only a central cone of image light which cannot pass through the mass of the electronic camera. Examples 2 and 4 relate to that problem.

[0048] Example 2.

[0049] Fig 3 is an elevation diagram as viewed from a side of the skinner machine. The plane return mirror 301 is now tilted relative to an optical axis of the incoming image light. The camera 105 and lens 104 are mounted inside the tunnel and near the diverting mirror, again with the lens 104 below the camera 105. The lens is directed at a (2 * tilt) angle toward the tilted return mirror 301 to receive image light along axis 3O8R. Now the camera and lens are completely clear of the image light 308 and can not obstruct a cone of image light accepted by a stopped-down lens passing between the diverting mirror and the return mirror. As a result the lens 104 may be stopped down to a practical minimum as required to provide a desired depth of field. The tunnel 300 may be made shorter than for the tunnels described in ‘399, or if remaining at a previous external length, the tunnel is inherently more effective for minimising proximity distortion.

[0050] Rotation of the diverting mirror 303 (about 5 degrees anticlockwise in Fig 3) can be used to optimise the volume required for the tunnel 300.

[0051] It is evident that the total path length from an end of the hazardous volume to the camera lens is increased, as compared with a path to the approximate position of the prior-art camera as in Fig 1. Further plane mirrors (not shown) could be included so that the light follows a zigzag path having many reflections before reaching the camera. Since practical mirrors exhibit a 5 to 10% of loss of brightness at each reflection, assuming front-silvered and protectively coated mirrors, while, as in the numerical examples above, diminishing returns apply to ever- longer camera to hazardous volume distances, only a few repeat reflections would be useful. Nevertheless the multiple internal reflection version of Fig 3 may result in further shortening of the length of the tunnels.

[0052] Example 3. Fig 4 shows a variant of Fig 2. (Note that the diagram Fig 4 is rotated to lie horizontally and fit the page). It has previously been found useful to surveil the interior of the hazardous volume by installing more than one camera, giving more than viewpoint at each end, as described in the Applicant’s earlier patent. In order to reduce the number of cameras required, the plane return mirror 401 is comprised, as in this drawing, of two facets 501, 502 as shown in Fig 5. More facets may be useful. Each facet is tilted sufficiently to cause the camera lens to focus two slightly separated fields of view on to the sensor chip of the camera; preferably without overlap. Having two or more viewpoints reduces a risk of obscuration of the gloved finger by, for example, nearer fragments of the workpiece being skinned, within the hazardous volume 101. The concave tilt actually shown in Fig 4 is excessive. A “convex” tilt (where the included angle of the surfaces of the two mirrors exceeds 180 degrees) would allow use of two separate, side-by-side electronic cameras (not shown) to capture two images having slightly different viewpoints.

[0053] The video feed (or digital equivalent) 110,111 from each electronic camera to the processor is handled in the same way as in the Applicant’s previous patent documents by looking for any instance of pixels having the characteristic glove colour within a certain proximity to the blade.

[0054] In an alternative, the diverting mirror 103 at the commencement of the path through each tunnel may be comprised of two facets as shown in Fig 5. That can provide two side-by-side viewpoints into the hazardous volume. If the faceted diverting mirror is concave, a single lens may be used to bring two focused images side by side on the camera CCD. If the faceted diverting mirror is slightly convex, then a pair of electronic cameras, one each side of the midline, will capture the two images. If the facets themselves carry colour filters, the processor may differentiate one image from the other on the basis of mean colour content

[0055] The processor 112 will be programmed, as in the Applicant’s previous patent documents, to halt the skinner motor if a patch of colour is detected in any incoming image light.

[0056] Example 4.

[0057] Fig 6 illustrates a plane return mirror 201 and a selected diverting mirror 603; typically one having a 50% reflective coating. The optical axis of the incoming image light entering window 107 is shown with an arrow. The camera 105 and lens 104 are now located above mirror 603 and in an upward (with reference to the drawing) extension 602a of tunnel 602. Note that in practice the entire assembly shown may be rotated about the optical axis of the incoming image light so that tunnel 602 / 602a lies in a horizontal axis to each side of the operator’s work table. The camera 105 receives some of the incoming image light reflected back again from mirror 201 through the tunnel 602 from the non-tilted (though optionally faceted as per Fig 5) return mirror 101 through the imperfect diverting mirror. Typically, the 50% of incoming image light that is passed through mirror 603 without reflection is absorbed by a light-absorbent layer or baffle 604 inside the tunnel so that it is not reflected up to the camera as stray light. For a similar reason, the window 107 is preferably mounted with a slight tilt away from the axis of the incoming image light so that reflections off the window glass are shifted off the axis. The camera body 105 and lens 104 are now above the axis of the image light coming out of the hazardous volume, providing an increased length between the camera and the near edge of the hazardous volume. One limit to growth of the extension of the tunnel 402a is whether or not it is acceptable as part of construction of the skinner. An advantage of this configuration is that components located inside the tunnel, except the diverting mirror 603, are mounted in an in-line configuration normal to an optical axis. A dichroic mirror (one having a coating that reflects a specified band of colours) may be used if the LEDs are selected to emit light in a particular colour range and fluorescent gloves are used.

[0058] Advantages

[0059] Improvements to the optical system as described will provide more reliable machine recognition of a hazardous situation in which a blue-gloved hand or part thereof intrudes into the hazardous volume. The intention is to render a glove appearance anywhere along the hazardous volume more similar by reducing proximity distortion. That may be expressed as “making ‘glove sensitivity’ more constant”.

[0060] The physical configuration of the skinner machine may benefit from having shorter tunnels.

[0061] Finally it will be understood that the scope of this invention as described and / or illustrated herein is not limited to the specified embodiments. Those of skill will appreciate that various modifications, additions, known equivalents, and substitutions are possible without departing from the scope of the invention as set forth in the following claims.

Claims

WE CLAIM.

1. A skinner machine provided with operator protection for hands approaching a defined, elongated hazardous volume along the blade of the skinner, said hazardous volume ending near an adjacent window of a sealed tunnel, each window admitting a bundle of image-forming rays of light (herein ‘image light’) derived from within the hazardous volume and having an optical axis to a first mirror that, when in use, deflects the image light to a lens of at least one camera placed further along the tunnel, characterised in that a second or return mirror is located at the far end of the tunnel and the lens of at least one machine- vision camera receives the image light over an extended path and thereby reducing an effect of proximity distortion wherein hands closer to the window appear larger to the camera. .

2. The skinner machine as claimed in claim 1, characterised in that the return mirror is a plane mirror disposed perpendicular to the optical axis and the camera is located between the diverting mirror and the return mirror and along the optical axis.

3. The skinner machine as claimed in claim 1, characterised in that the return mirror is a plane mirror tilted away from the optical axis so that, when in use, the return mirror directs the image light to a camera located to one side of the diverting mirror.

4. The skinner machine as claimed in claim 2, characterised in that the return mirror is comprised of an array of at least two plane facets tilted at an angle to the optical axis so that, when in use, at least two separate points of view into the hazardous volume are focused as side by side areas upon a sensor within the or each camera.

5. The skinner machine as claimed in claim 1, characterised in that the or each camera is located above a partially reflective diverting mirror; preferably having a coating having an about 50% reflectivity; the camera is located on the axis of the light and receives the light from the return mirror after passage through the diverting mirror and through the partially reflective diverting mirror for a second time.

Images