DE2544233A1 - AUTOMATIC MACHINE TOOL - Google Patents

Description

PART OF PODDIG

Paionla.fsejsor

Applicant: UV Philips'Gloeilomnenfabrtiltii
File No .: f> Hß 21 Y £?
Registration from »O * a, q -y <r

2 5 4 A 2 3 3

PHB. 32467 September 26, 1975

Automatic machine tool,

In general, this invention relates to a control device for production machines with visual input Out of his devotion, In the present Pail can be the visual Input via a suitably prepared television camera. This visual scene will therefore generally be two-dimensional. In particular, the invention therefore relates to machine tools, work on objects in two dimensions. A machine that punches a two-dimensional surface of a solid object is a typical example. A circuit board is a typical object with a two-dimensional surface.

A series of holes usually have to be drilled in such circuit boards, which make the connection wires more electronic

609815/045 «

- 2 -. PHB.32467

Pick up components. The circuit boards are made from a sheet of insulating material made, which is covered with copper foil. This film was partially etched away, so that conductor tracks were created are. A conductor track results in a solder connection point for the lead wire of a component, and in a preferred example a small central field is etched away in the track, so that a spot marks the desired position of the hole. If a hole is needed that is not intended for a soldering point, it can be through a point or another predetermined one Icon to be highlighted. Numerical controls for machine tools are well known. The manufacture of the ■ controlling memory, e.g. a punch tape memory, costs a lot of effort for small series of products. The invention has the task of producing the perforated tape or other The invention provides a machine tool with visual input that identify a drilling location marked in this way, which is then positioned under the drill to drill the hole without human intervention needs and without the position of the hole as with numeric controlled machines must be specified. ) That's why you r'.st die The object of the invention is a fully automatic and self-regulating. To create machine tool.

With prolonged use, such a visually controlled Drill Feheier of adjustment and translation between television camera, drill and movements of a carriage on which the PCB is attached, accumulate. As is usual with machine tools, this carriage is referred to below as a coordinate

6098 15/0456

- 3 - PHB.32467

tent table or simply referred to as a table. These errors can accumulate for a variety of mechanical and electronic reasons. It may be desirable to have this To measure and correct errors periodically, and it is also a purpose of the present invention to do the same therefor to use a visual input system that is also used to control the drilling process.

Although the problem and the individual parts of the

Invention are described in their application to drilling operations, the invention is also to other forms of machining and other tools applicable. It is applicable to a chainsaw, to an automatic welding or bonding machine and the like.

The invention now provides an automatic machine tool with a two-dimensional coordinate table, a tool holder which has a work center for the two dimensions mentioned, first means for separate measurement of the first positions reached by the table in each of the two dimensions mentioned, second means, those for receiving of information is set up to display secondIbsitions what positions the table in each of the _; should achieve the dimensions mentioned. Drive means for generating a relative displacement of said tool holder in at least one of said dimensions in response to a discrepancy signal emitted by each of said first and second means, characterized in that said automatic tool holder

609815/0456

- 4 - PHB.32467.

254A233 ^26-9 · ¹⁹⁷⁵

machine optical means for viewing part of a The working plane of the Tsiches contains third means for inputting an image of a cross thread means onto the working plane of the Table so that the center of the said image is correctly aligned with the work center of the tool holder, where said automatic machine tool includes a visual one Input device which is oriented in such a way that it forms part of the above-mentioned working plane including the above-mentioned raster image sees and includes a pattern recognition device

a) to identify a previously defined mark on a flat surface of an object that is so on said table is such that the flat surface lies in said working plane;

b) identifying said raster image and

c). to measure the displacement of said rake from

Center of said raster image, while an output signal of said pattern recognition device controls said drive means.

The grid can be any chosen two-dimensional pattern be that the | finding of its center point .by the visual Input device facilitated. A preferred form of the grid can be the general shape of a Maltese cross with high contrast between the bars of the cross and the back ground. The tool holder can accommodate any machining device be set up, which has a well-defined work center, which is orthogonal to the known table advanced onto the object A typical tool can be a drill, a cutting

609815/0456

- 5 - ■ PHB.32467

26.9 1975

tool or any other tool that has a cutting, Has a grinding, erosion or electrolytic effect. The working level des tsiches is a plane parallel to that of the two named dimensions defined plane, e.g. after a rectangular or rotationally symmetrical system. The working plane can at a fixed distance from the bed of the machine tool to facilitate the input of the raster image into the Working level. Mechanical stops or other registration devices can be used be provided that bring the flat surface of the object into the working plane.

The visual input device includes a camera capable of two-dimensional electrical mapping of the two-dimensional visual scene - a conventional one Television camera with a photocathode that is scanned by an electron beam can be used for this purpose. A camera with an arrangement of individual sensing elements one after the other be scanned electrically, but can also be used.

The visual input device may also be an apparatus for converting the output signals of this camera into a form that is suitable for storage in a digital computer or equivalent device that is suitable for the Recording of the converted camera output signals is set up and outputs output values for controlling the table displacement.

Such an automatic machine tool has more

Information about its environment as a conventional machine tool. This has two main consequences. First is the controls

809815/0456

- 6 - PHB.32467

situations that would otherwise be difficult to control were let. Second, it is when the visual information is fed into a coupling loop that leads to the separation of Mark and raster image center addresses, no longer necessary to use the precise mechanical position measuring devices, which are used for numerically controlled machines without feedback of the position of the object on the machine table.

Vox returns, for example, to an automatic printing machine or circuit board drilling machine : a television camera is used to record the circuit board, and a small computer processes the television image. The computer program determines the position of the drilling points on the circuit board. Once a drilling point has been determined, the computer gives an XY table the command to bring the circuit board into the required position, where the drilling is then carried out.

£ $ The whole system works in a closed loop; This means that any errors in the movement of the camera or circuit board can be automatically compensated for. The grid is mechanically fixed to the position of the drill, and an image from the R&S fcer is projected into the optical system of the television camera and appears coplanar with the circuit board.

The task of the system is to create a. Place on the circuit board where a hole is to be drilled, like this position so that it appears to coincide with the center of the raster image.

The camera can monitor its performance for this task if necessary, corrections can be made. It

609815/0456

- 7 - PHB / 32467 '.

oc / / ο ο ^ 9/26/1975

It must be emphasized that the control principle outlined above is not limited to an automatic drilling machine, but is generally applicable to control using visual feedback. A drill of this type could be in a research or in a development laboratory Find application where short series of printed circuit boards of many different types are required.

If the marks are invariant under rotation in certain applications, the angular position can be variable e.g. in each of the four 90 directions.

The control principle outlined above can be used, for example, with machines that locate individual parts correctly and align them so that they can be picked up and fed into an automatic assembly machine. at For some objects this is not easily achievable by mechanical means. One possible solution is to use the.; single or to visually grasp the object and determine its position and orientation. The tool holder is then held by a Replaced a gripper mechanism that rotates under the control of the computer. The visual input device then allows one Determine the relative orientations of the part and the gripper and the rotation of the gripper and the movement of the Partly so that it is aligned with the gripper.

The visual input device can be arranged to have a previously processed by the tool Body identified that they have a relative dislocation between

609815/0456

- 8 - PHB.32467 '

this point and the said raster image center and that it uses this offset in such a way that the mark is perfectly aligned with the tool in a subsequent machining operation is, in the example of the drill can 'the visual input device be able to find the position of the grid center point with regard to a reference point on the light-sensitive surface of the camera. This reference point can a certain position on a certain cable of the television set if the camera is a conventional television camera as described above. The reference point can be a be a specific detector element in the array if a camera with stationary array of elements is used. The geometry of the mark on the object to designate the drilling point is chosen so that they can be easily felt and its position in relation to the reference point of the camera can be easily measured. A drilled hole also creates a clear output signal. So can the position of a drilled hole with respect to the reference point of the camera to be determined. A comparison can then be made between the position assumed by the drill in relation to the grid and the reference point of the camera be employed. A correction offset can then be derived if necessary and when drilling the next hole in use a different part of the circuit board.

If the first shift of the circuit board is made so that a hole position coincides with the center of the raster image, this shift can be a large part of the camera's field of view

609815/0456

- 9 - .- FHB. 32467

September 26, 1975

subsequent corrective shifts are reduced if this first shift is exact in size and direction. Therefore, distances in the two coordinate directions should be as determined by the camera, if possible be exactly the same as the required table movements. That Ratio between the distances, like that of the camera and when table movements were measured, can also can be measured and, if necessary, appropriate corrections can be made ■ be determined. A particular brand, e.g. a drilled hole or a Bobiss-Jungstelle can be attached to one on the table Disk can be identified and its position determined with respect to the camera. Then the table can be underneath Use the displacement measurement tools to move a known amount and move to the new position of the same

Mark in relation to the camera. The shift of the mark in relation to the camera can now be determined and with the known displacement of the table can be compared. The ratio of displacements depends, among other things, on the magnification of the image of the plate on the light-sensitive The area of the camera used or the size of the haste on the television camera. That so determined Ratio of shifts can be stored in a memory and used to issue correct shift commands for the following (Fish movements based on Karaera measurements can be used.

This invention also relates to apparatus for recognizing marks on a flat surface. The level

609815/0458

- 10 - FHB.32467

Surface can be the original surface on which the camera is located or the light-sensitive surface of the electronic Camera. The mark can be applied to the workpiece, as an aid in automatic treatment or processing of the workpiece, e.g. to align the brands under the Work center of a tool holder. Such a brand can

be regarded as a limited device or patch that is much darker or lighter than the surrounding area. The present The invention relates to apparatus for detecting the presence of such a mark or such a spot. the automatically operating machine tool has' a preferred embodiment, wherein said pattern recognition device means for applying an operator to said predetermined mark and said operator an operational one Contains central element that can coincide with a given picture element and has peripheral elements that coincide with others Image elements coincide, as well as means for generation a detection signal when at least a predetermined part of the said peripheral elements fulfills the condition, that it coincides with an image learner whose intensity varies from the de ^ figurative element, which deals with the work center minest a predetermined threshold with the same sign differs.

The planar distribution of picture elements can be as above are derived from an electron beam television camera. In this case, the lines are in regular intervals Video signals recorded at intervals on the lines of the television

609815/0456

- 11 - PHB.32467

rasters and the columns are derived from corresponding signals in successive and adjacent rows of the raster. The intensity of each picture element then results from the amplitude of the corresponding part of the video output signal. However, the video output signal, which corresponds to the strength of each pixel being scanned, can be digitized so that the maximum range of expected video output signals is divided into a series of regularly spaced intensity levels, e.g. A given intensity is assigned to the fourth of the level to which :, it is closest. This level value can

be encoded as binary code, e.g. as 5 ~ 1> it binary code at 32 Levels. Rows of operator elements can be on the lines or on

the columns or on diagonal lines that line the rows and columns, or on a combination thereof. An output signal arises when the work center is surrounded by a certain set of picture elements, of which a previously defined subset has exclusively higher or only lower intensity values, the picture element in the work center. The operator element can lie on radial lines. Each radial line can only contain one operator element that is at least © in picture element away from the work center. In this way, the operator can recognize spots of various shapes. One can say that the operator in this respect has an elastic quality, since the most diverse stretchings and distortions of the leak can be processed. The radial lines may not have any operator elements for a predetermined number of eggs next to the work center and can be used as

609816/0456

- 12 - P HB.32467

254 A 23 3 ^2ί5 · ⁹ · ¹⁹⁷⁵

Broken line to be described. In this case, if

the contrast field surrounding the spot is so small that the radial lines of the operator are not reached, the operator is not satisfied and such a spot is rejected. It may be desirable that such ring-like features be rejected. On the other hand, the operator condition in the case of such broken lines, be such that an operator element of each line must differ from the brightness in the work center by at least a threshold amount.

Part of the invention will now be described with reference to the figures.

Figure 1 is a schematic drawing of the PCB drilling machine,

Figure 2 is a schematic drawing of a practical implementation of the machine of Figure 1,

Figure 3 shows the determination of the movement scale factor,

Figure 4 shows the determination of the hole for the grid verse tongue:

Figure 5 shows a typical digitized circuit board image as it is stored in the computer.

Figure 6 shows a typical binarized circuit board image, and the resulting possible drilling sites (candidates).

Figure 7 shows an operator with only one operator element on eight radial lines each.

Figure 8 shows a dark field that .not from the operator from picture 7 is recognized. '

609815 / 04.56

- 13 - .. FHB.32467- '

., -, ". $ 2. j. 1975 '

25A423 3

Figure 9 shows a dark field created by the operator of BiIa 7 is recognized.

Image IO shows a dark field that is too close to another dark field to be. from the operator of image 7 to be recognized.

Figure 11 shows a general form of expansion of the operator.

Figure 12 shows the problem of Figure 10, solved by an extension of the operator.

Figure 13 shows an operator version that rejects a spot that is surrounded by too little contrast area.

Figure 14 shows another version of the operator that rejects a spot, at least its surroundings is too thin at one point.

Fig. 15 shows an apparatus for the application of the Operator on a digitized image.

M Re picture Ii A printing circuit 1, the surface of which is perpendicular to the plane of the picture, is connected to a motor-driven two-dimensional coordinate table 2.

Table 2 can move linearly without rotating in two directions at right angles to each other in one plane Run parallel to the printed circuit board 1 and the stationary support surface 3. The drill 4 (in a tool holder not shown) is located under area 3 and can be replaced by a looh in area 3 so that it drills a hole in printed circuit board 1 when the latter is stationary and clamped on surface 3 is. A flat, semi-mirrored mirror is centrally above the

609 815/0456

- 14 - PHB.32467

Drill 4 arranged and lets television camera 6 perpendicular to the Look at the drilling surface. The optical axis of the camera lens 7 runs horizontally. An Easter 8 is in the axis of Objektif 7 in one Distance from mirror 5 attached., Which corresponds to the distance between circuit board 1 and mirror 5. So becomes a picture of Easter 8 entered into the working level of the table and onto this The camera is in focus when it is focused on circuit board 1. Camera and computer 9 make one Visual input device. Computer 9 receives and processes the television camera image and issues movement commands to the motor-driven table 2. Stepper motors are used used that respond to commands that consist of pulse trains. Computer 9 also controls the lighting of the window and moves the drill 4 in the, forward direction if this is the case a drilling sequence is required.

To picture 2: This picture is a detailed schematic representation of the mechanics of the drilling machine incl. Drill holder 4A. So that access to the machine table is more convenient, If the camera sees the Easter in a plane mirror 11 at an angle of 45 to the lens axis, mirror 11 can also be seen funds not shown can also be set. Its angular position can be about two axes at right angles to each other can be changed so that the virtual image of the easter 8 coincides with the virtual image of the one seen in mirror 5 Drill is brought. The lighting system is illuminated with the elongated light source φ2 and a cylindrical eeflector 13 the circuit board is reflective and results in a first light source

609815 / 0Λ 56

- 15 ~ PHB.32467 -

area. An inclined concave reflector 14 has an ellipsoidal shape Configuration that creates a converging light beam, in the middle of which the axis of lens 7 runs, Mach fieflektion arrives at the flat, semi-mirrored mirror 5 the bundle on circuit board 1. Those parts of the bundle which are reflected in a specular manner from circuit board 1 converge Via the mirror 5 again on the diaphragm plane of the objective 7. Thus, the mirror 14 forms the source 12 on the diaphragm from lens 7.

The operation of the drilling machine system described with reference to Figures 1 and 2 will now be described.

The image from the television camera is processed by the computer in two phases. First, the search for potential ones takes place Drilling locations, which are referred to below as candidates. Most of the candidates found are in indeed at drilling sites. In certain environments no further search is necessary, in other environments some Candidates wrong and no drilling sites. An operator for finding candidates looking for a dark area that is surrounded by a lighter area is described below. It is entirely possible that a false candidate is found that fulfills the conditions of this operator. This case is preferable because a careful The search for objects that are clearly drilling points would proceed extremely slowly over the entire area. It is better to search the area very quickly for places that could be drilling sites, i.e. for candidates. then

609815/0 4 56

- 16 - PHB.32467

these areas are searched in detail with an operator /, who follows the edges to check whether they are indeed candidates. If a candidate is found, it will be its coordinates are stored in a memory. If a list of all candidates has been compiled within a sub-area, which is determined by the field of view of the television camera, the candidates are successively in the center of the field of view brought and scanned there again, and then an operator who follows the edges is used to scan the edge of the dark area in the pattern consisting of copper strips to determine if the candidate is a real well or a false well. This depends ofr the edge-scanning operator returns to its starting point within a fixed series of stages. In this case it is a "real" candidate. On the other hand, the pattern of the copper tracks can be such that none dark area exists, and then it could be the outside are scanned. If only a single conductor track leads there, the edge would be scanned within a fixed row of steps almost lead to a return (i.e. to the other side of the individual conductor track mentioned).

If now a correct candidate, i.e. a real The center of the dark area is found at the drilling site. This becomes the center Aligned over the drill, and the hole is drilled. If all Candidates within a sub-area have undergone this treatment, the camera's field of view is reduced to one

609815 / 045Θ

- 17 - P HB.32467

Relocated other sub-surfaces to which the same procedure was applied The in-French patent specification 2150150

The edge scanning operators described are suitable for the Scanning the edge of the dark area when the stored image of that area has been converted to binary form.

If the individual candidates are in the center of the The field of view are moved, the centering takes place on the image of the basters, which is used as a position replacement for the drill serves. This is done by scanning around the candidate to determine the center point.,:,

If you want to check the positions of the box again, the grid lighting is now switched on and regulated to a sufficient brightness, so that the image of the circuit board in the television camera is hidden. In a preferred fairy game, automatic gain reduce: un £; s ·? control circuits in the camera control the entire video gain so, that the raster image is in the dynamic range of the camera and the details of the circuit board image in practice get lost. A separate program in the computer is set up to determine the position of the grid center point -in bezuf the camera grid to an operator. The again determined position of the center of the drilling point and the position of the The center of the grid is then compared in the computer and the correct one is sent to the Pösifcioniermotoren on the circuit board Move command when the difference between these two positions larger than the previously specified permissible error. Fa'chdem if a correction has been made, this will be

609815/0458

- 18 - FHB.32467

9/26/1915- ■

Hole drilled, or the coordinates of the hole location are punched into a tape, or both. A storage of the coordinates would be advantageous in the case of a series of identical objects that need to be edited. One benefit of using stepper motors to drive the table is that the computer get a signed running total of all Can store displacement pulses that have been passed on as commands to the motors. This sum can be used at the beginning the hole of a circuit board can be set to zero when a position reference point on the circuit board is on the grid is centered. In this way, the coordinates of all drilled holes or drilling locations can be found in relation to this zero point on the circuit board be registered. These coordinates are mechanical measurements and do not depend on the coordinates of the camera grid away.

Before a circuit board or a set of circuit boards are drilled on the machine, the parameters of the so-called "World Model" of the computer will be updated. This world model consists first of all of a set of basic assumptions that are based on the Not far from the drilling job. These assumptions are based on the Machine construction, such as the orthogonality of the table movements or the parallelism of the camera grid lines to one of the table movements addicted. Further assumptions concern the shape of the drilling locations and the grid geometry. ' These assumptions depend on the choice of the operator used to determine it. Further assumptions concern the possibilities of the machine behavior, those in response to the facts determined by the television camera

6098 15/0456

- 19 - Pim. 32467 · '

the environment are permissible, after all, there is an assumption in the Choice of method for processing the camera video output built into binary signals.

Second, the world model consists of the set of parameters mentioned in the assumptions, which only have a limited life to have. These are: (a) The position of the circuit board in relation to the origin of the coordinates on the table. (B) Pas relationship between a given stage displacement and the value of that displacement as measured by the camera. This Ratio can be referred to as the camera-Ti'sch scale. (c) The position of the grid center point as viewed by the camera is seen, (d) the position of the drill in relation to the grid center point and (e) a histogram of the gray levels representing the Camera perceives each and which are used to determine the binary threshold.

This is where the holes on a circuit board begin the computer an introductory process for measuring updated Values the first four of these five parameters. So the jump (a), a scale sector (b), the grid position (c) and the Grid offset (d) is determined when: a circuit board is clamped on the table. The parameters (a), (b) and (c) depend from the drift of the camera grid. The displacement of the drill to the grid (d) remains constant much longer because it is only dependent on the theoretically rigid machine structure depends. The gray level histogram (e) must be constantly updated while a circuit board is being drilled, as the reflection values of the circuit board vary from. Can distinguish place to place and the lighting through the

609815/0456

■ - 20 - · PHN.32467

26.91975

Light source can change over time.

The determination of the moving channel factor (b) is shown in Figure 3. The fixed table 3 is shown schematically as a rectangular surface with a camera 6 looking up vertically. The camera view display 15 is shown schematically. A isolie ^ feature it is shown in an initial position 16 on the table 3 and in a Endstellungl7 by known shifts in ~ £ and £. This feature, which can be, for example, a drilling position or a special symbol etched into the circuit board, is recognized by the computer after the camera image has been processed. The starting position (l6 ') and the end position (l7 ^f ) of the image of the feature appear on screen 15. The ratio of the real displacement Δχ of the feature x to the displacement of the feature measured by Δχ' is then calculated in the computer and used as a correction factor that must be applied to each subsequent camera measurement of displacements. The ratio of Ay to Ay ^{1 is also} calculated and stored for the correction factor of jr. Figure 4 shows a schematic of the measurement of the offset between the grid and the drill. For the sake of clarity, the image 19 of the grid is shown on the table when it is related to the table 3. The drill 18 is shown schematically as if it had borrowed a hole. When the grid is illuminated, the detent identification operator can determine the position of the grid center point on the screen 15 in the computer. If the grid lighting is switched off and the borehole is illuminated, the edge scanning operator in the computer can determine the effective position of the drill on the screen 15. The offsets Xo and Yo of the drilling center point to the grid center point can then be designated and used as correction

609815/0 4-56

- 21 - ΓΗΒ.32467

September 26, 1975 '.

Offsets are stored that must be applied at the hole location to coordinates that are related to the Easter have been determined by the camera.

The picture element speed of a normal 625-line television camera is around 10 million per second too high for convenient storage in the computer. To reduce this, the interface circuits between the camera and the computer memory scan only one picture element of each line of the picture. Although the video amplitude for a picture element is only about 0.1 μβ long, this amplitude can be scanned and held by a suitable circuit during the full 64 μβ ^ -βτ line period. In this 64 \ X8 , an analog-digital converter can split the sampled video signal into five: binary digits, ie into 32 gray levels. Then there is still enough time to transfer nasty five digits to a selected memory location. The next line is the single element, the

is vertically below the last scanned picture element, i.e. the picture element at the same address on the next line, scanned, digitized and to one. -other. Transfer memory location. Thus, within one raster time of the camera - 40 ms - a single vertical column of Image elements saved. In the next raster line, the Address changed on cbn lines by one picture element width and the next adjacent vertical column of picture elements scanned, digitized and stored. By continuing this process in successive raster lines, each -Previously defined area in the field of view of the camera is digitized and transferred to the memory,

609815/045 »

25U233

If desired, an image can be stored with a lower resolution by changing the line address by more than one pixel width when scanning at the beginning of this time raster. The stored picture element columns are then spaced apart by a predetermined number of picture elements. Instead of scanning every line of the camera image, a predetermined number of lines between scanned lines can also be omitted. The stored rows of picture elements are then also spaced apart by a predetermined number of picture elements. Figure 5 shows in diagram form a typical scene that could be the input material for the computer from a circuit board. The gray levels shown here are on a scale from 0 to 31. To determine the drilling points, the candidates must first be determined. The candidates have dark points. 1 ⁸ , in other words, high gray level values, surrounded by a number of points that have dark gray level values. Picture element 50Jin picture 5 with a gray value 26 is a candidate. Picture element 51 with a gray value 'of 18 is located sio' a near the center of an area that may be a candidate. There can be more than one candidate per I-hole position. Once a candidate has been found, a higher resolution image is obtained that is centered on the candidate. This image is binarized and an edge-sampling operator is applied to the candidate to determine if there is a closed loop of edge points around the candidate exists around. The edge-scanning operator used in this case is only suitable for application to a binarized image. Thus, in the digitized image of picture ^

609815/0456

- 23 - PHB.32467

September 26, 1975

a threshold gray value can be applied to each picture element so that it can be decided whether it should be represented in the binarized image by a "1" which appears as a black point or a ¹¹ O '"which appears as a white point.

The threshold appropriate for an area of the image may not be appropriate for another area as the gray levels of all image elements in one area are different from those in one different area. Further, a well site may be an area with low gray level values, that of is surrounded by an area with high gray levels. For each Image, a histogram of the local gray values is saved so that there is a threshold that the drilling locations clearly separates from the copper surfaces.

Such a binarized image is shown in image 6 in which at (a) for the sake of convenience the areas 60, 61, 62 in the drawing, which consist exclusively of copper, are surrounded by a dashed line. The pictured Denotes minimum level size. a picture element. The surfaces 63, 64, 65 are areas of white points in the black Areas, and white areas contain candidates, at (b) in Figure 6, the candidates who pass the edge-scan operator's test for closed loops become the binarized image shown in solid lines. Rules can now be used to select only one candidate from a group of contiguous candidates Candidates are placed, e.g. the three next to each other Candidates 66. These rules prevent any attempt

609815/0456

- ZH " PEB. 32467 '

2 * .9.1975 '.

the machine to two or more intersecting holes In such attempts, the drill would probably drill abort. ■ ■

To picture 7: Me rows 110 and the columns 111 of the picture elements 112 are shown at right angles to one another and define square picture elements. The intensity value of a Picture element is the average intensity of the area of the square that defines this picture element. The work center ■ has 8 lines of operator elements that radiate around it. Each line has only one operator element, for example Element 102flas some element in the vertically rising line and element 103 the one element in a diagonal line. With this operator, the elements in a line are not adjacent, but are at a distance of one picture element from each other - so the whole operator is in a 'square, which has one side of five picture elements.

To picture 8: An L-shaped dark area 113, the three

Occupied picture element is displayed in such a way that it falls within the range of the operator of Fig. 7. Operator element 106 coincides with a dark picture element- »Work center 109 is lighter than element 106 by the threshold amount, but just as bright as operator elements 101, 105 ₅ 106 and 108. As a result, the operator condition is not fulfilled and the operator does not emit a recognition signal. In Figure 9, an L-shaped dark area 114 is shown in such a way that it falls more centrally into the area of the operator. The work center 109 is now the only element of the operator that coincides with a dark picture element and

60 9 815/04 5

- "£ 2 - ΡΠΒ.32467

September 26, 1975

is darker than all line elements 101 to IO8, also around the Threshold amount. The operator condition is met and the operator emits a signal that a mark has been recognized.

The operator shown in Figure 7 (defined on a square grid) is correlated with the image to be analyzed and scanned on the same square grid. The operator is designed in such a way that only the center point and eight points distributed on the edge of the square of (odd number) are taken into account. Let P (i) be the density of a picture element seen through the template ± _ . Let T be a threshold set.

If the condition:

P (9)> P (i) + T for χ = (1,8) (l)

is fulfilled, then there is a spot in that of the external elements 101 to IO8, and one of its elements is element IO9 (see Figure 3).

The actual size of the stain is not measured, but that is not the job of the operator. A situation where a dark line or edge is the operator's domain crossed out is not sufficient to satisfy condition (l). However, if another object is so close that it is of an or multiple boundary elements of the operator is seen, then condition (l) is not fulfilled, although one that appears as a spot Object available (see picture IO).

A general form of an extension of the operator that can cope with the situation in Fig. 10 is shown in Fig. 11. At. This operator sees the spot as a dark object, 609815/0456

-25 - PHB..32 /] 6T

which is surrounded by a closed ring of picture elements that are lighter than the object itself. The shape of this ring. is not defined, but its internal dimension must be smaller than dimension D. D is the size of the operator who is responsible for this situation should recognize. D is an odd number and in Figure 5 D = 9. This operator is a rectangular template made up of eight linear arrays 116 radiating from the center 115 go out.

So that the operator condition can be expressed more succinctly, let P (i, r) be the density of a picture element which is represented by the element r of point x. ^ - ^{it is} seen operators. ' and let P (θ) be the density that. seen through the center of the operator.

Condition (l) now becomes:

P (0)> Pmin (i) + T fHr i = (1.8) ■ (2);

where "Pmin (i) is the minimum value of T at each r from 1 to t to (D-1) for each line i.

In other words; on each branch there must be an element condition (l) fulfill. This operator can detect an object that is close to another. An operator of this type with D = 5 is shown in Figure 12. The neighboring dark area 117 covers an operator element 118, but element 119 in the same line as element 118 is not covered. Consequently the operator condition is fulfilled.

Figure 13 shows one form of the operator with two of the innermost Operator elements of every radial line are missing. The dark area 20 surrounding the patch or pith 21 does not extend as far as

60981 5/0456

- 2? -. PHB.32467

one of the inner ends of the cadial lines. In this case becomes the condition of the opera-fcors of none of the radial lines fulfilled and the Fleete is rejected because he is of too little contrasting surface is surrounded.

Figure 14 shows a fora of the operator in which only one innermost operator element is missing in every radial line and for which the operator condition requires that at least two Operator elements in each radial line differ in intensity - from the work center 23 by at least one threshold value »An example is shown for a dark area 24, does not cover the two operator elements on the radial line 25. This stain is rejected because it is from a dark area which is too thin in one place.

• Figure 15 shows an apparatus for the use of the operator - from image 7 to a digitized image. It will be in the following Description of Figure 15, assume that the image was scanned in 64 vertical columns of 64 picture elements each and that each Picture element was digitized into a 5-tit binary code, which 32 gray levels defined. The image information for the apparatus of Fig. 15 is available as a sequence of 5-kit codes, which each describe a picture element, starting at the top left in scanned field, downwards in successive columns to the right continuously. The 5-bit codes should be connected to five conductors in parallel If the 5-bit code for the first element is available on At the top of the left column of the scanned area becomes available, the five digits of the code are sent in parallel to the inputs from ffinf; 64-bit shift registers 31. These five entrances 30

609815/0456

- 28 - PHB.32467

are shown with their binary ratings Wl or ¥ 2, ¥ 4, ¥ 5 and Vi 16. A shift pulse is then applied to all stages of the registers 31 by means not shown. The first 5-bit code is thereby shifted one step down in the registers 31, and the inputs of the shift registers 31 are then free to receive the next 5-bit code of the second picture element down in the left-hand column of the scanned area when it becomes available This process continues until all 64 pixels of the left column of the scanned area have been entered sequentially into registers 31 and these registers are completely filled are shown as a single conductor are removed from the registers 31 at the bottom and applied to the corresponding inputs at the top of a second set 33 of five 64rbit shift registers. All stages of the registers 33 are fed with shift pulses from the same source as the registers 31 as well all other levels of the more distant sets of five 64-bit shift registers r 34 ₅ 35 and 36 wide column from the left side of the scanned area, entered into register 31 at the top and the digits of the first column transferred to register 33 one after the other. This is because the number of stages in each register has been made equal to the number of picture elements in a column. When the second column has been scanned, the digitized image information of these two first columns can be found in this order in registers 31 and 33, respectively. Put the picture elements on one line

609815/0456

' PHB. 32467

in the scanned area are located on the same Addresses downwards in the registers and therefore appear as in Fig. 15, horizontally next to each other in registers 31 and 33. In registers 33 and 35, the individual No stages and in registers 34 and 36 only in skeletal form shown at the points where initial infoxination decreased

From the left side of the scanned area, column by column, the field of five register columns 3I ₅ 33> 34, 35 and 36 is entered, with the corresponding connections 32, 37j 38 and 39 from the lower end of one register set to the upper end of the next register set to the right ensure that the image information runs through each shift register set. When five complete columns have been scanned, the five lowest levels of the five register sets contain at this point in time the image information of a square field of picture elements, the side length of which in the upper left corner of the picture is five elements.

Re Figure 7i The operator shown here is in one Contain a square whose side contains five elements. Thus, at the above point in time, the gray values of a sentence of picture elements which fit into the operator of picture 7, taken in parallel from the relevant stages of the shift register and the operator feed condition can be applied to them. The work center IO9 is with the third level from the lower Connected to the end of the register 34. The operator elements 101 to 1C8

609815/0458

- 30 - PHB.32467- ·

26.9 ^r . 1975 ■

are connected to levels 41 to 48. The welding value T is applied as a binary number to a digital subtractor 49, just like the binary number of level 40. The variable A is used by. Subtractor 49 as the gray value of element 9 minus threshold value T issued. Eight digital subtractors 50A, 51A, 52 ... 57 contain the value A, and each receives the gray value of the relevant Operator element. For example, element / ol is replaced by ^ -den Lowest stages 41 of register 31 are supplied and subtractor supplies a "1" output if the result of subtracting the gray value of element 1 from A = 0 or a positive number is. The registers 33 and 35 provide the required memory space, since the operator of Fig. 7 alternates between Columns of picture elements is applied. The eight input MD gate 58 thus receives eight binary input values, all of them have the value "1" if the operator condition is fulfilled and in this case outputs an output value "1" to terminal 59.

When the sixth column is scanned from the left and the registers 31 receive the digitized gray values of the picture elements in this column, the content of all registers 31, 33, 34, 35 and 36 shifts downwards. This moves the operator down on the left side of the image. The operator condition is reapplied with each shift pulse and the output value appearing at terminal 59 indicates whether a spot is immediately at the point of the work center. The scanning of further columns in the image causes ,, that the operator aud the successive columns in the ·. Image is applied. The information is from level 43 at the bottom of the registers

609815/0 4 56

- 31 - ΡΏΓ. 32467 ■

September 26, 1975

emptied. In a period near the end of each column the operator receives information partly from picture elements at the bottom of the column and partly from picture elements at the top of the column the next column. The IMI) gate 58 will not go through here The means shown are depleted, but the output is controlled by the sample cycle during this transition period.

In the foregoing had the operators described Operator element rows that lay on straight lines that ran .radially from the work center. On the other hand you can these lines also have a tangential direction, that of said Eadialrichtung is superimposed. It is not always necessary that these lines are straight lines. In other cases, multiple Lines can be applied or other combinations can be made. of light and dark operator elements prove useful.

609815/0456

Claims (1)

- 3Γ2- FHB.324 : 67- PA TEIJTAT-TSP RÜCHE; - . 1. An automatic machine tool with a two-dimensional Coordinate table, a tool holder that has a work center for the two dimensions mentioned, first Means for separately measuring the first positions reached by the table in each of the two dimensions mentioned, the second Means which are set up for receiving information for displaying second positions, which positions the table should reach in each of the dimensions mentioned, drive means for generating a relative displacement of the named tool holder in at least one of the named prostitutes in response to a discrepancy signal emitted by said first and second means, characterized in that that said automatic machine tool optical means - for viewing a 'part of a working plane of the table third means for inputting an image of a cross thread means onto the working plane of the table so that the center of the said image is correctly aligned with the work center of the tool holder, the said automatic Machine tool includes a visual input device that ' is aligned in such a way that it sees part of the said working plane including the said raster image and a pattern recognition device contains a) to identify a predetermined brand a flat surface of an object that is located on said table in such a way that the flat surface is in said Working plane lies; b) identifying said raster image and c) to measure the displacement of said mark from the center of said raster image, while an output signal of the '609815 / 04S6 - 33 -. , PHB. 3.24.67 · ' named "pattern recognition device" the named drive means controls. 2.. An automatic machine tool according to patent claim 1, in which the visual input device is set up in such a way that it identifies a location previously machined by the tool can that they measure a possible relative displacement between the position of this point and the said raster image center and that it uses the said offset to align the mark with the tool in a subsequent operation. 3. An automatic machine tool according to claims 1 or 2, in which the visual input device for measuring the ratio between a given table displacement and the value of said given displacement measured by the visual input device has been set up, as well as for the storage of the mentioned ratio and for its application as a correction factor following table displacements measured by the visual input device. ^ f; An automatic machine tool according to the preceding claims, in which the visual input device, the pattern recognition device on an image of the Object during the alignment of the mark to the named center point of the raster image, which has a higher resolution than that for identifying the previously established brand serving. ■ $ 1 An automatic machine tool according to the preceding patent claims, in which the visual input device contains a camera which is capable of capturing part of said work plane 609815/0456 - 34- ■ PHB.32467 to represent two-dimensionally with electric (j.hen means, where the representation of regularly spaced lines and splitting of picture elements, means for applying a threshold level to the gray level of the camera output for each picture element to generate a binarized form of the table surface considered and means for compiling a histogram of the gray levels of the picture elements of the observed Area, so that said histogram provides a poptimal value for said threshold level. 6. An automatic machine tool according to claim 5, wherein the camera is a television camera and the rows and columns of picture elements are derived from the Easter of the television camera. 7. An automatic machine tool according to claim 5 or 6, in which the said pattern recognition device Comprises means for applying an operator to said predetermined mark and said operator an operative ^ Contains central element associated with a given picture element can coincide and has peripheral elements that coincide with further Image elements can coincide, as well as means of generation of a recognition signal, if at least a previously determined one Part of said peripheral elements fulfills the condition that it coincides with a picture element whose intensity is identical from that of the picture element that deals with the working center distinguishes at least one predetermined threshold value with the same sign. 8. An automatic machine tool according to that Claim, according to which the said peripheral "elements are located on lines that extend from the territory of the manned operational • 609815/04 * 5 6 -3-5- PBB. 32467 26.9.1-975 Central element extend outwards and after at least one peripheral element per line must meet the stated condition in order for the stated recognition signal to arise. 9. An automatic machine tool according to claim 8, in which the lines mentioned extend radially from the central operative element mentioned. 10, An automatic machine tool according to claim 95 where a gap the width of at least one picture element between the operative central element and the first peripheral element of each line. H, An automatic machine tool according to the claims 8, 9 or 10, in which each of the lines mentioned comprises the same number of peripheral elements. 12. An automatic machine tool according to claims 8, 9 or 10, wherein each of said lines is a Contains a single peripheral element. 13. 'An automatic machine tool according to the claims 8, 9j 10 or 11, where for the generation of the named recognition signal per line at least two peripheral Elements must meet the stated condition. 14. An automatic machine tool according to the claims .7 "to 13j with the largest expected scale of the pixel intensities into a plurality of levels with at regular intervals and at which the level value a pixel intensity to which it comes closest is assigned. 609815/0456