DE102008031240B4 - Method for determining a distortion by a line scan camera - Google Patents

Abstract

Method for determining a distortion in object recordings by a line camera (ZK), in particular when used in an automatic placement machine, wherein an object (Ob) to be recorded and the line camera (ZK) are moved relative to one another during recordings, with a line camera (ZK) recording a structure with a known geometry is created (2), target positions for individual points of the structure in the recording are determined (3), then actual positions of the individual points of the structure in the recording are measured (4), then the determined target Positions are compared with the recorded actual positions of the points of the structure and the distortion factor (shx) is estimated and determined on the basis of a deviation between the determined target positions and the recorded actual positions of the points of the structure (5, 6, 7) through which the distortion in the object image due to a non-orthogonality between the line camera and the direction of movement of the light compensating object, and for a movement of the object to be recorded (Ob) or the line scan camera (ZK) a traversing system is used, characterized in that a compensation of the distortion by appropriate control of the traversing system on the basis of the distortion factor (shx) when recording objects (Whether) is carried out.

Description

Technical area

The present invention relates to a method for determining a distortion by a line scan camera, in particular when used in a placement machine, wherein an object to be picked up and the line scan camera are moved relative to each other during recording by means of a positioning system.

State of the art

For tasks of quality assurance, sorting operations and especially in an automatic assembly of printed circuit boards with components in so-called placement machines line scan cameras are used for image capturing objects in image processing systems. As a line camera while a digital camera type is referred to, which has no two-dimensional sensors, but at least a single sensor line. However, there are also technical versions of line scan cameras with more than one sensor line.

If pictures of objects are to be taken with the line camera, then an object (eg electronic component) or a sequence of objects performs a movement relative to the line camera. It is characteristic of a line scan camera that for a complete image of an object to be recorded or a sequence of objects several partial images are always made (ie, the object or objects are "scanned"), with a relative movement of the object or objects taking place between the partial images , However, it is also possible to move the line scan camera relative to the object, for example, for a shot. By this movement, it is possible to obtain two-dimensional images, although the line scan camera only has at least one one-dimensional sensor line for the image recordings.

When assembling printed circuit boards with electronic components in placement machines, for example, an object (eg electronic component) is transported via the sensor line of the camera in a traversing direction perpendicular to the line direction of the line camera for recording with a positioning system (eg traversing system of the placement head) until the sensor line of the line scan camera has completely passed.

Usually, however, the direction of movement of the object or, when the line camera moves, the direction of movement is not exactly perpendicular to the line direction of the line scan camera. This means that the direction of travel of the object or direction of movement of the camera and line direction of the camera are not aligned exactly orthogonal to each other or the angle between the direction of travel or direction of movement and line direction of the camera does not exactly 90 °. Due to this non-orthogonality, images of the line scan camera are distorted. A recorded rectangle is displayed for example as a parallelogram in the recording. Will now z. As a position determination of an object, a sorting of objects, etc. performed on the basis of such a distorted recording, so may result due to the distortion position or sorting error.

In placement machines, electronic components must be precisely measured in order to be placed with high accuracy on printed circuit boards. For the measurement, for example, line scan cameras are used. Therefore, due to the non-orthogonality of travel and line direction when using line scan cameras, distortions of the images may, for example, lead to positioning errors and, as a consequence, to malfunctioning and faulty circuits.

To avoid such errors in the assembly of printed circuit boards is from the Scriptures US 5,864,944 a method for placement machines, in which by means of several calibration and compensation operations, the distortions of the line camera images are compensated. In this case, a possible rotation of the object to be recorded is first determined on the basis of a recording by the line camera, wherein the object to be recorded is attached to a movement system and is moved by the latter relative to the camera. This rotation is corrected by a corresponding orientation of the object or the positioning system (eg mechanical rotation of the positioning system). Based on a new shot of the same object is measured whether there is a non-orthogonality or a deviation of 90 ° between the direction of movement of the object and the line direction of the camera. In the case of non-orthogonality, the line camera is then aligned by means of a mechanical device (eg motor) correspondingly orthogonal to the direction of travel in order to compensate for the non-orthogonality. By a further image of the object is then determined whether or not the distortions in the recording have been corrected by the mechanical orientation of the object and the line scan camera. If further distortions are detected in the recording, the process steps are repeated.

A problem with the font US 5,864,944 disclosed method is that a multi-stage of the process on the one hand and by possibly repeating the process several times, a relatively large amount of time for the determination and compensation of the distortions arises. In addition, for the implementation of the method additional mechanical components - such. B. a separate mechanical device for aligning the line scan necessary, resulting in higher manufacturing costs and a greater effort for production and maintenance of placement arise.

JP 6-253141 A discloses an image reading apparatus which can take an image without distortion. First, distortions of the optical system of the image reading device are detected. This is done using a two-dimensional reference structure which has an array of equidistantly spaced bars and which is detected by the optical system. The acquired image data are compared with theoretical image data of the reference structure and used to determine the distortions of the optical system. Furthermore, a virtual read-in position is stored for each pixel. Later, when taking a real image, the distortions of the optical system are compensated by performing arithmetic interpolation based on the virtual read positions, resulting in a distortion-free image density signal as a result.

Presentation of the invention

The invention has for its object to provide a method by which a simple, inexpensive and time-saving way, a distortion in object recordings can be determined and compensated.

The solution of this object is achieved by a method of the type described above, wherein a recording of a structure with known geometry is created by a line scan camera and then target positions for individual points of the structure in the recording are determined. In the same step, actual positions of the individual points of the structure in the image are then measured. Then, the determined target positions of the points of the structure are compared with the recorded actual positions of the points of the structure, and based on a deviation between the determined target positions and the recorded actual positions of the points of the structure, a distortion factor is estimated and determined then used for a compensation of the distortions in the following shots of the line scan camera.

The main aspect of the proposed method is that for a determination of the distortion factor, which can be used in a simple manner for a compensation of the distortions in the images of the line scan, z. B. for a placement process only a recording of the structure with known geometry (eg., Dot grid, grid with circles, etc.) or only one calibration is necessary. Thus, the inventive method is advantageously associated with a relatively small amount of time. In addition, for example, when using the method according to the invention in an automatic placement no additional device to z. B. mechanically align the line scan camera to compensate for distortions. As a result, no additional expenses and costs for the compensation of distortions arise during production and maintenance of the placement machine or any other device (eg sorting system, etc.).

According to the invention, a movement system is used for a movement of the object to be picked up or the line scan camera. In this case, the object (eg electronic component) is usually moved by means of a positioning system. In a placement automatically expediently a placement is used to move the object for a recording relative to the line scan camera.

According to the invention, compensation of the distortion is also carried out-in particular in the case of placement machines-by appropriate control of the positioning system. The basis for the control is supplied by the distortion factor. For compensation, for example, the traversing axes (eg x-direction, y-direction) of the positioning system (eg traversing system of the placement head) are controlled in such a way that the object to be recorded is moved orthogonally to the sensor line of the camera during image acquisition. Ie. The object is moved by the positioning system in the traversing direction (eg y-direction) over the sensor line and a corresponding, slight correction movement is carried out in a direction normal to the travel direction (eg x-direction). This provides the camera with a low-cost and easy way to capture images without distortion due to non-orthogonality.

beschrieben wird. Dabei werden durch den Vektor [x_I, y_I] Koordinaten der Ist-Position eines Punkts der Struktur und durch den Vektor [x_S, y_S] Koordinaten der Soll-Position eines Punkts der Struktur dargestellt. Vom Vektor [t_x, t_y] wird eine translatorische Verschiebung beschrieben. Durch den einen Winkel φ wird eine Objektrotation (z. B. Verdrehung des Objekts am Verfahrsystem) dargestellt. Von der Variablen s wird ein Skalierungsfaktor der Zeilenkamera dargestellt, welche z. B. durch Differenzen zu einem bekannten Abbildungsmaßstab der Zeilenkamera bedingt wird. Von der Variablen s_hx wird der Verzerrungsfaktor beschrieben. Aus dieser Formel kann damit auf einfache Weise mittels einer numerischen Optimierung der Verzerrungsfaktor s_hx abgeleitet werden und es können andere Einflussfaktoren auf die Aufnahme (z. B. translatorische Verschiebung, Skalierung der Kamera, Objektrotation, etc.) ebenfalls berücksichtigt werden.It is advantageous if a relationship between a desired position of a point of the structure with known geometry and an actual position of this structure in the recording by means of the formula

is described. In this case, coordinates of the actual position of a point of the structure are represented by the vector [x _I , y _I ] and coordinates of the desired position of a point of the structure by the vector [x _S , y _S ]. The vector [t _x , t _y ] describes a translational shift. An object rotation (eg, rotation of the object on the positioning system) is represented by the one angle φ. From the variable s, a scaling factor of the line scan camera is shown, which z. B. is due to differences to a known magnification of the line scan camera. The variable s _hx describes the distortion _factor . From this formula, the distortion _factor s _hx can thus be derived in a simple manner by means of a numerical optimization, and other influencing factors on the recording (for example translational displacement, scaling of the camera, object rotation, etc.) can also be taken into account.

ermittelt wird, durch welches die Abweichung zwischen Soll- und Ist-Positionen der Punkte der Struktur mit bekannter Geometrie beschrieben wird. Dabei werden durch die Variablen x_n,I und y_n,I die Ist-Position eines n-ten Punkts der Struktur und durch die Variablen x_n,S und y_n,S die Soll-Position eines n-ten Punkts der Struktur dargestellt. Die Variable v steht für einen Vektor, in welchem translatorische Verschiebung, Objektrotation, Skalierungsfaktor der Zeilenkamera und auch den Verzerrungsfaktor s_hx beinhaltet sind. Von der Laufvariablen n wird ein Bereich von 1 bis N durchlaufen, wobei die Variable N eine Anzahl an Gitterpunkten der aufgenommenen Struktur bezeichnet. Durch eine Minimierung des angegebenen Fehlerfunktionals kann auf einfache Weise aus dem Vektor v der Verzerrungsfaktor s_hx ermittelt werden.In a preferred embodiment of the invention it is provided that the distortion factor by minimizing the Fehlerfunktionals

is determined by which the deviation between the target and actual positions of the points of the structure is known geometry described. The variables x _{n, I} and y _{n, I represent} the actual position of an n-th point of the structure and the variables x _{n, S} and y _{n, S} the desired position of an n-th point of the structure , The variable v stands for a vector in which translational displacement, object rotation, scaling factor of the line camera and also the distortion _factor s _{hx are} included. The run variable n traverses a range from 1 to N, where the variable N denotes a number of grid points of the recorded structure. By minimizing the specified error function, the distortion _factor s _hx can be determined in a simple manner from the vector v .

As a structure of known geometry, ideally a dot grid or a grid with a collection of circles regularly arranged in the grid can be used. Such structures have the advantage that a sufficiently large number of actual positions of points can be measured by them, which can be compared in a simple manner with desired positions of these points. The desired positions of the points or circles, which are regularly arranged in the grid, are determined in a simple manner by means of image evaluation algorithms, the actual positions of the points or circles in the recording are measured. Furthermore, it is conceivable that the structure with known geometry consists of crosses, which are arranged irregularly, but in a known form.

Brief description of the drawing

The invention will now be described by way of example with reference to the accompanying drawings. Show it:

1 schematically recording an object with a line scan camera

2 a schematic sequence of the method according to the invention for determining a distortion in object images by a line scan camera

Embodiment of the invention

In 1 In a schematic and exemplary manner, the recording of an image from an object Ob (eg electronic component) with a line scan camera ZK is illustrated, for example, by means of an insert in a placement machine. In 1 In the alternative, two axes x and y are entered for better understanding, which are orthogonal to one another. The sensor line of the line camera ZK is aligned parallel to the axis x.

In placement machines, for example, objects such. B. electronic components for precise positioning optically recorded and measured by means of a line camera ZK. The objects Ob or components are used for a recording of a in 1 not shown in a travel direction VR z. B. over the line camera ZK moves. As a positioning system z. B. that traversing system used on which a placement is mounted. From the placement head, the component is picked up and Ob example, moved before placing on a circuit board for recording or for measuring via the line camera ZK. This movement system consists for example of two axes, which can be moved independently.

The object Ob can be received in any direction from the positioning system and therefore has a so-called object rotation, which in 1 is represented by an angle φ. The angle φ or the object rotation φ on the positioning system is described by a deviation between the object Ob and the axis x.

Usually, the travel direction VR, in which the object Ob is moved by the movement system via the camera ZK, is not exactly orthogonal to the orientation of the sensor line of the camera ZK. The direction of travel VR therefore differs by an angle α from a direction which is parallel to the axis y or normal to the axis x. By this deviation α or non-orthogonality between line camera ZK and travel direction VR arise in the images of the line camera ZK distortions, which for z. B. exact positioning operations of electronic components must be compensated.

2 shows the schematic sequence of the method according to the invention for determining a distortion in object images by a line camera ZK. An object Ob (eg electronic component) is moved by a positioning system (eg traversing system of the placement head) in the travel direction VR, which is not exactly orthogonal to the alignment of the sensor line of the camera ZK, via the line scan camera ZK.

The method according to the invention starts with a starting step 1 began. In a second process step 2 is moved by a positioning system a structure with known geometry - such as a point grid with a known number and known positions of the grid points on the line camera ZK. In this case, the two-dimensional camera ZK produces a two-dimensional image of this structure, for example the dot grid. In the further course of the method according to the invention will be described by way of example with a point grid as a structure of known geometry. However, the method according to the invention can be carried out in the same way with a grid in which accumulations of circles are regularly arranged. It is also possible to use a structure which has irregular, but arranged in a known form crosses. The essential requirement of the structure with known geometry is that it has a sufficiently large number of points whose desired and actual positions in the recording of the line scan camera are determined in a simple manner (eg image evaluation algorithm, etc.) can.

In a third process step 3 are determined due to the known geometry of the dot grid target positions of the individual points of the dot grid in the recording. In a fourth process step 4 are then measured actual positions for that point of the dot grid in the recording of the line camera ZK.

beschrieben werden, wobei diese Formel in gleicher Weise für andere Strukturen mit bekannter Geometrie (z. B. Gitter mit regelmäßig angeordneten Kreisen, Gitter mit unregelmäßig angeordneten Kreuzen, etc.) gilt. Dabei werden von den Variablen x_I, y_I Koordinaten für eine Ist-Position eines Punktes in der verzerrten Aufnahme der Zeilenkamera beschrieben. Von den Variablen x_S, y_S werden Koordinaten für eine Soll-Position eines Punktes in der Aufnahme angegeben. Vom Vektor [t_x, t_y) wird eine eventuell vorliegende translatorische Verschiebung dargestellt. Mit der Variablen φ wird die Objektrotation φ berücksichtigt. Von der Variablen s wird ein Skalierungsfaktor der Zeilenkamera ZK angegeben, welche durch eine Differenz zu einem bekannten Abbildungsmaßstab der Zeilenkamera bedingt wird. Die Variable s steht für einen Verzerrungsfaktor s_hx, durch welchen die Verzerrung aufgrund der Nicht-Orthogonalität beschreiben wird.In a fifth process step 5 Target positions and the actual positions of the respective points of the dot grid in the recording are compared and a relationship or deviation between the target positions and the actual positions of the respective points of the dot grid formed in the recording. This connection can be made by means of the formula

This formula applies equally to other structures of known geometry (eg, grids with regularly arranged circles, grids with irregularly arranged crosses, etc.). Here are described by the variables x _I , y _I coordinates for an actual position of a point in the distorted recording of the line scan camera. The variables x _S , y _S indicate coordinates for a desired position of a point in the recording. From the vector [t _x , t _y ) a possibly present translational shift is shown. The variable φ takes into account the object rotation φ. The variable s indicates a scaling factor of the line camera ZK, which is caused by a difference to a known image scale of the line scan camera. The variable s stands for a distortion _factor s _hx by which the distortion due to the non-orthogonality is described.

In a simplified form, the relationship between the desired positions and the actual positions of the respective points of the dot grid or any other structure used with known geometry in the recording of the line camera ZK can also be represented as follows:

In this representation of the relationship, the terms s · (cosφ + shx · sinφ); s · (-sinφ + s _hx · cosφ); s · sinφ and s · cosφ are represented as coefficient matrix a ₁₁ to a ₂₂ .

beschrieben wird. Im Vektor v sind dabei die Koeffizientenmatrix a₁₁ bis a₂₂ sowie die translatorische Verschiebung [t_x, t] zusammengefasst. Von den Variablen x_n,I und y_n,I wird die Ist-Position eines n-ten Punkts des Punktgitters und von den Variablen x_n,S und y_n,S die Soll-Position eines n-ten Punkts des Punktgitters dargestellt. Die Variable n ist eine Laufvariable, die eine Wertebereich von 1 bis N annehmen kann, wobei von N eine Anzahl an Punkten des Punktgitters angegeben wird.From the relationship between actual and desired positions of the points in the recording with the line camera ZK is then in a sixth method step 6 derived an error functional f ( v ), which by the formula

is described. In the vector v , the coefficient matrix a ₁₁ to a ₂₂ and the translational displacement [t _x , t] are summarized. Of the variables x _{n, I} and y _{n, I} , the actual position of an n-th point of the dot grid and of the variables x _{n, S} and y _{n, S} the desired position of an n-th point of the dot grid is shown. The variable n is a run variable that can take a range of values from 1 to N, where N indicates a number of points in the dot grid.

By minimizing the error function f ( v ), for example by means of numerical optimization, it is now possible in the sixth method step 6 the coefficients a ₁₁ to a ₂₂ as well as the translational displacement (t _x , t _y ) are determined from the known coefficients a ₁₁ to a ₂₂ is then in a seventh process step 7 the distortion factor s is estimated and thus determined, which can then be used as a consequence for a compensation of the distortions caused by the non-orthogonality in the recordings of the line scan camera ZK.

The determination of the distortion s _hx is thereby in a single calibration process, which the process _steps 1 to 7 includes and can be used for all other recordings with the same requirements to compensate for the distortions. For example, in a placement machine in a calibration process, the distortion _factor s _{hx is determined} for a present configuration of the placement machine (eg placement head, certain components to be assembled and printed circuit boards) and can then be used for all placement operations with this configuration.

The compensation of the distortions can take place on the one hand in the line scan camera ZK itself. In this case, a recorded image is equalized in the line _scan camera, wherein a corresponding equalization _{process is} derived from the determined distortion _factor s _hx . The equalization process, for example, from a camera electronics such. B. a microprocessor or a so-called FPGA module are performed.

On the other hand, it is easy to compensate for the distortions due to non-orthogonality with the aid of the displacement system, by means of which either the object Ob or the camera ZK to be photographed is moved during recording. The traversing axes of the positioning system (eg placement head in a placement machine) are controlled in such a way that the non-orthogonality is compensated. The object Ob is thus moved orthogonally to the line scan camera ZK during image acquisition. The corresponding control is derived from the distortion factor s.

In the case of image acquisition in a placement machine, the movement system is usually moved only along a first travel axis, while no movement takes place in the direction of a second travel axis. This results in a travel direction VR of the object Ob to be recorded which is not oriented orthogonally to the sensor line of the camera ZK. By the appropriate control of the positioning system on the basis of the distortion _factor s _hx not only a movement in the direction of the first _{travel axis} , but also a slight movement in the direction of the second _{travel axis} is executed by the positioning system, by which the non-orthogonality and thus the distortions in the recording be compensated. A speed ratio of the respective movements along the first or second _{travel axis} corresponds to the distortion _factor s _hx . However, the method according to the invention and the possibilities of compensation can not only be used for placement processes in placement machines, but also for other processes in which line scan cameras ZK are used for taking pictures of objects, such as eg. As tasks of quality assurance, sorting operations, etc. are used to compensate for distortions in line camera recordings.

Similarly, the described method of detecting and compensating for distortions due to non-orthogonality may be used even if, for example, the line camera ZK is moved for shooting instead of the object Ob.

Claims (4)

Method for determining a distortion in object recordings by a line scan camera (ZK), in particular when used in a placement machine, wherein an object to be recorded (Ob) and the line scan camera (ZK) are moved relative to each other during recording, wherein the line scan camera (ZK) is a recording a structure with known geometry is created ( 2 ), Target positions for individual points of the structure in the recording are determined ( 3 ), then actual positions of the individual points of the structure in the recording are measured ( 4 ), then the determined target positions are compared with the recorded actual positions of the points of the structure and estimated and determined based on a deviation between the determined target positions and the recorded actual positions of the points of the structure of the distortion _factor (s _hx ) ( 5 . 6 . 7 ), by means of which the distortion in the object photograph is compensated due to a non-orthogonality between the line camera and the direction of travel of the object to be recorded, and a movement system is used for a movement of the object to be photographed (Ob) or the line camera (ZK), characterized in that a compensation of the distortion by appropriate control of the positioning system on the basis of the distortion _factor (s _hx ) when taking pictures of objects (Ob) is performed. A method according to claim 1, characterized in that a relationship between a desired position of a point of the structure and an actual position of a point of the structure in the recording means

is described ( 5 ), where from the vector [x _I , y _I ] coordinates of the actual position of a point of the structure, from the vector [x _S , y _S ] coordinates of the desired position of a point of the structure, from the vector [t _x , t _y ] a translational shift, the angle φ an object rotation (φ), the variables s a scaling factor of the line _scan camera (ZK) and of the variable s _hx the distortion _factor (s _hx ) are described. Method according to one of claims 1 to 2, characterized in that the distortion _factor (s _hx ) by minimizing the error function as (f ( v ))

estimated and determined ( 6 . 7 ), by which the deviation between nominal and actual positions of the points of the structure is described, of the variables x _{n, I} and y _{n, I being} the actual position of an n-th point of the structure, of the variables x _{n , S} and Y is _{N, S is} the nominal position of an n-th point of the structure, a vector of the variable v in which translational displacement, object rotation (φ) scale factor of the line camera (ZK) and the distortion factor (s _hx) contain and variable n describes a variable with a value range of 1 to N, where variable N denotes a number of points of the structure. Method according to one of claims 1 to 3, characterized in that as a structure of known geometry a point grid or a grid with a collection of circles, which are regularly arranged in the grid, is used.

Images