DE102021210375A1 - calibration procedure - Google Patents

Abstract

The invention is based on a calibration method (10) for an image capture device (20), wherein in at least one method step (14) a surface (36) to be captured by a camera of the image capture device (20) is marked with physical reference markers (50), wherein a digital image (34) of the surface (36) to be recorded is created by means of the camera, with the reference markers (50) being used by the image recording device (20) to generate a virtual grid (52) which is oriented to the reference markers (50). , wherein pixels of the created image (34) grid elements (54) of the grid (52) are assigned, the grid elements (54) all having the same actual format.It is proposed that the reference markers (50) be used to perform a perspective correction in the Image (34) existing distortion of an image (56) of the surface to be detected (36) is performed, the distortion caused by an angle between the Oberfläc he (36) and a recording plane of the camera, the grid (52) being divided into grid units (58), all of which have the same actual format and in each of which the same number of grid elements (54) are arranged, with one actual height and width of the grid units (58) are specified by the reference markers (50).

Description

State of the art

A calibration method for an image capturing device has already been proposed, in which in at least one method step a surface to be captured by a camera of the image capturing device is marked with physical reference markers, with the camera creating a digital image of the surface to be captured, with the reference markers being used to the image capturing device generates a virtual grid which is oriented to the reference markers, with pixels of the image created being assigned to grid elements of the grid, with the grid elements all having the same actual format.

Furthermore, a calibration method for calibrating an image capturing device has already been proposed, with an exposure duration or an exposure intensity of the image capturing device being determined in at least one method step depending on a calibration object captured by the image capturing device.

Furthermore, a calibration method has already been proposed in which, in at least one method step, a position of a hand-held device, which is intended in particular for a defined application of paint to a surface, is determined in real time, with the device using an inertial measuring unit of the device and using a camera an image capturing device is monitored, optical position data, which are determined by the image capturing device, and movement data, which are determined by the inertial measuring unit, being evaluated in combination to predict a position of the device.

Disclosure of Invention

The invention is based on a calibration method for an image capturing device, in which a surface to be captured by a camera of the image capturing device is marked with physical reference markers in at least one method step, with the camera creating a digital image of the surface to be captured, with the reference markers being used to the image capturing device generates a virtual grid which is oriented to the reference markers, with pixels of the image created being assigned to grid elements of the grid, with the grid elements all having the same actual format.

It is proposed that a perspective correction of a distortion of an image of the surface to be recorded that is present in the image is carried out using the reference markers, the distortion being caused by an angle between the surface and a recording plane of the camera, the grid being divided into grid units, which all have the same actual format and in which the same number of grid elements is arranged in each case, with an actual height and width of the grid units being specified by the reference markers. The configuration of the calibration method according to the invention advantageously enables a perspective correction to be carried out. Due to the configuration according to the invention, an advantageous transfer of the data of the digital image to a real position on the surface to be recorded can be created, particularly if the image of the image of the surface to be recorded is distorted.

The image capturing device is preferably designed as a mobile, personal electronic device. The image capturing device is particularly preferably designed as a smartphone or as a tablet PC. The image capturing device preferably has a camera. The image acquisition device is preferably provided for image processing, in particular dynamic image processing. The image capturing device is preferably provided for capturing a digital image of a surface to be captured, in particular a motif original on the surface to be captured. The image capturing device is preferably provided to perform a perspective correction of a distortion of an image of the surface to be captured, in particular of the original motif, present in the recorded image, in order to calibrate the image capturing device with respect to a real position of the original motif on the surface. The image capturing device is preferably provided to capture a position of a device, in particular a hand-held device, in particular relative to the surface and/or the motif template, in order to calibrate the device with respect to a real position of the device. The image capturing device is preferably calibrated using a calibration object before a digital image of the surface to be captured is recorded. The image capturing device is preferably calibrated in order to be able to capture the surface to be captured by the image capturing device, in particular the original motif on the surface to be captured, with correct exposure. An exposure setting of the camera, in particular the exposure duration and/or the exposure intensity, is preferably adjusted by means of the calibration method in order to match the surface to be recorded, in particular the original motif on the surface to be recorded Surface to be captured by the camera and to be able to carry out image processing.

Preferably, a physical position of each pixel of the digital image on the surface is determined from the reference markers and from the virtual grid. As a result, an advantageous simple perspective correction can be carried out. The surface to be detected is preferably arranged in one plane. An object is preferably imaged on the surface to be detected. The object is preferably in the form of a motif template, with the motif template being able to be applied to the surface, in particular permanently, onto the surface and/or introduced into it after it has been captured by the image acquisition device by means of a medium application, in particular an application of paint.

The reference markers are preferably arranged on the surface, in particular in an adhesive manner, but can be detached without being destroyed. The reference markers are preferably arranged in a band. The band preferably forms a frame which is arranged in particular around the object imaged on the surface to be detected. Preferably, the frame formed by the band is rectangular. Preferably, the fiducial markers each have a pattern arranged on a surface-facing side of the fiducial markers. The reference markers preferably have a square shape. All reference markers are particularly preferably identical. Immediately adjacent reference markers are preferably spaced apart from one another by a distance that corresponds to a format of the reference markers. Preferably, each reference marker has an actual format of 100mm x 100mm. The reference markers are preferably identified by the image acquisition device, in particular by means of the pattern.

In order to create the digital image, light waves which emanate from the surface to be recorded are preferably converted into digital signals by means of an image sensor of the camera. The digital image is preferably broken down into a large number of pixels, each of which has color information. The digital image is preferably digitally processed by the image acquisition device. In principle, the digital image can also be displayed on a display of the image acquisition device.

An “image of the surface to be recorded” should preferably be understood to mean a view of the surface to be recorded in the digital image, in particular taking distortion into account. A “distortion” should preferably be understood as meaning a perspective distortion, which consists in particular of a recording plane of the image sensor of the camera that is not aligned parallel to the surface to be recorded. In principle, it would also be conceivable that distortion, which is caused in particular by the optics of the camera, is taken into account in the perspective correction.

The virtual grid is preferably generated digitally, in particular by means of the image acquisition device. The virtual grid preferably forms a regular structure of grid lines. The virtual grid is preferably based on the reference markers. The virtual grid preferably forms a matrix to which pixels of the digital image are assigned. The virtual grid is preferably inserted into the digital image and/or stored as an independent data set as a reference for assigning the pixels. The virtual grid in the digital image is preferably distorted depending on an arrangement of the reference markers in the image of the surface to be recorded.

The virtual lattice preferably comprises a multiplicity of lattice units. The grid units preferably have the same format as the reference markers. Preferably each grid unit has an actual format of 100mm x 100mm. An "actual format" should preferably be understood to mean a format which is related to the real surface to be recorded. An “actual height and width of the grating units” should preferably be understood to mean a height and width which are related to the real surface to be detected. Directly adjacent grating units are preferably arranged without any spacing from one another. Preferably, each lattice unit comprises a plurality of the lattice elements. Preferably, the actual format of the lattice elements is square in each case. Preferably, each grid element has an actual format of 1mm x 1mm. Directly adjacent lattice elements are preferably arranged without a spacing from one another.

Furthermore, it is proposed that within a virtual frame, which is directly surrounded by the reference markers, horizontal and vertical pairs of reference points are defined, each of which is arranged on a grid line of the grid. With this configuration, positions of the grid lines can advantageously be determined in a simple manner. As a result, the virtual grid can advantageously be generated simply using the reference markers. The virtual frame preferably borders directly on the reference markers, in particular on the band. Reference points of the pairs of reference points are preferably each at a corner point of a reference marker on the virtual frame arranged. "Horizontal and vertical pairs of reference points" should preferably be understood to mean pairs of reference points that are exactly opposite one another in the band in which the reference markers are arranged, in particular running orthogonally to the virtual frame, with an alignment of a corresponding grid line at least substantially horizontal or runs at least substantially vertically.

Furthermore, it is proposed that gradients of grid lines are determined for the calculation of corner coordinates of the grid units, the gradients being caused by the distortion, with the grid lines of the grid units in each case resting on reference markers that are closest to the grid units. With this configuration, orientations of the grid lines can advantageously be determined in a simple manner. Furthermore, a position of the lattice units within the virtual lattice can advantageously be determined easily with this configuration. The grid units are preferably each defined by four grid lines of the virtual grid. The gradients of the grid lines are preferably determined using the horizontal reference point pairs or the vertical reference point pairs. The corner coordinates of a grid unit are preferably formed as intersection points of the grid lines of the grid unit. The corner coordinates of the grid unit are preferably determined using the slopes of the grid lines of the grid unit and using the horizontal and vertical pairs of reference points.

In addition, it is proposed that at least one distance between corner coordinates of a respective grid unit is calculated in order to calculate an assignment of pixels to grid elements. This configuration allows at least one reference coordinate to be defined for the grid elements arranged in the grid unit. A particularly advantageous perspective correction can be carried out as a result.

Furthermore, it is proposed that a distance between corner points of each grid element within a grid unit is determined as a function of a number of horizontally and/or vertically adjacent grid elements in the grid unit and of a distance between corner coordinates of a respective grid unit. An advantageously uniform size of the lattice elements within a lattice unit can be achieved with this configuration. A particularly advantageous perspective correction can be carried out as a result. Furthermore, as a result, the computing effort can advantageously be kept low.

Furthermore, it is proposed that a position of the lattice elements is calculated in floating-point numbers with double precision. This refinement advantageously reduces deviations to a calculation accuracy in decimal places. Furthermore, deviations can advantageously be localized for each point without the possibility of error propagation.

In addition, it is proposed that a lookup table be created in a parsing process, with a size of the lookup table corresponding to a resolution of the camera and/or a resolution of a display of the image capturing device. With this configuration, the data determined in the perspective correction can advantageously be further processed. Furthermore, pixels can advantageously be easily assigned to the positions as a result of this refinement. Preferably, the look-up table has a number of data fields corresponding to the resolution of the camera and/or the resolution of the display. Preferably, the look-up table contains color information corresponding to the pixels.

Furthermore, it is proposed that the look-up table is transmitted from the image acquisition device to a device, in particular a hand-held device. With this configuration, the data determined in the perspective correction can be advantageously used and/or further processed. Preferably, the device accesses the look-up table. Preferably, the device uses the look-up table to determine color information at locations on the surface.

It is also proposed that the look-up table be used depending on a position of a device, in particular a hand-held device, for a defined application of paint to the surface by means of the device. With this configuration, an advantageously precise reproduction of the original motif on the surface can be achieved, in particular if the image of the surface to be recorded is distorted. The device is preferably designed as a media application device, in particular a paint application device. The media application device preferably applies at least one medium, in particular colors, to the surface in a defined manner in accordance with the look-up table.

In addition, it is proposed that at least one sensor means of the image capturing device is used to determine a position of a device, in particular a hand-held device, with the position of the device being used for navigating in the look-up table. With this configuration, an advantageously precise reproduction of the original motif on the surface can be achieved, in particular if the image of the surface to be recorded is distorted. preferred wise, the image capture device captures the device relative to the surface. The image capturing device preferably determines a position of the device by means of a position marking of the device, which is recognized by the image capturing device, in particular by the sensor means. The image capturing device preferably transmits the position of the device to the device.

The invention is also based on a calibration method for calibrating an image capturing device, wherein an exposure duration and/or an exposure intensity of the image capturing device is/are determined in at least one method step depending on a calibration object captured by the image capturing device.

It is proposed that the calibration object has an illuminant which is intended to assume different color states, with the image capturing device detecting the calibration object in different calibration sequences in the different color states of the calibration object to determine the exposure duration and/or the exposure intensity. Due to the configuration according to the invention, the image acquisition device can be calibrated in a particularly advantageous manner. The configuration according to the invention enables a particularly advantageous exposure setting of the image capturing device to be determined for capturing a digital image. As a result, image processing, in particular further processing of a digital image that is created by means of the image acquisition device, can advantageously take place simply and reliably.

The light source preferably has an LED. The lighting means preferably emits colored light corresponding to a respective color state of the different color states.

The camera preferably has an optoelectronic image sensor. The image sensor is preferably exposed permanently, in particular without a mechanical shutter. The image sensor is preferably exposed to light for recording a digital image in a defined period of time, in particular the exposure time. Alternatively, the exposure time could be designed as a shutter speed of the camera. The exposure strength is preferably designed as a light sensitivity of the image sensor, in particular as an ISO value.

The calibration object is preferably positioned in a field of view of the camera. The calibration object is preferably recognized by the image capturing device. The calibration object is preferably captured by the image capturing device for executing the calibration sequences in a capturing phase of the calibration method. The image capturing device preferably captures the calibration object in each calibration sequence of the different calibration sequences in one of the different color states of the calibration object. In each calibration sequence of the different calibration sequences, a value range for an exposure duration and/or an exposure intensity is preferably determined, in which an image of the calibration object fulfills at least one defined brightness condition. The calibration object, in particular an outer contour of the calibration object, can preferably be distinguished from a background and/or an environment of the calibration object under the at least one defined brightness condition. It is preferably determined on the basis of at least one predetermined satisfaction factor at which brightness values the calibration object lies in the value range.

The acquisition phase is preferably followed by a calculation phase in which an optimal value for the exposure duration and/or the exposure intensity of the image acquisition device is determined. Preferably, the optimal value for the exposure time and/or the exposure intensity of the image capturing device is calculated from the determined value ranges for the exposure time and/or the exposure intensity from all calibration sequences.

The image capturing device is preferably set to the determined exposure duration and/or the determined exposure intensity. The image capturing device is preferably set to the calculated optimal value for the exposure duration and/or the exposure intensity. The determined exposure duration and/or the determined exposure intensity, in particular the optimum value, is preferably preset in the camera. Preferably, the image capturing device is calibrated again whenever there is a change in ambient brightness, with the image capturing device recapturing the calibration object in the various calibration sequences in the different color states of the calibration object to determine the exposure duration and/or the exposure intensity.

Furthermore, it is proposed that the image capturing device recognizes the calibration object from a shape of the calibration object. As a result of this configuration, the calibration object can advantageously be recognized quickly and reliably by the image acquisition device. As a result, an advantageously efficient calibration method can be provided. Preferably the Shape of the calibration object is stored in the image capturing device and/or the image capturing device has learned the shape of the calibration object. The calibration object preferably has at least a circular shape. The illuminant is preferably circular or spherical in shape. Alternatively, the calibration object can have at least one other form that appears suitable to a person skilled in the art, by means of which the calibration object can be recognized easily and reliably.

Furthermore, it is proposed that the calibration object runs through more than three, in particular six, different color states during the detection phase, in which the calibration object assumes a color in each case. As a result of this configuration, the exposure duration and/or the exposure intensity can advantageously be determined reliably. As a result, an advantageously reliable imaging of motif templates of different colors can be achieved. The color that the calibration object assumes in the different color states is preferably blue, cyan, green, yellow, red or magenta.

In addition, it is proposed that a color state of the calibration object is changed automatically during the detection phase. With this configuration, an advantageously automated and rapid calibration of the image acquisition device can take place. The calibration sequences are preferably carried out automatically in succession. The color states are preferably run through in a fixed sequence with a defined period of time. The sequence of the color states and the duration of the individual color states are preferably known to the image acquisition device. In principle, it would also be conceivable for a color state of the calibration object to be changed manually, for example using a toggle button on the calibration object.

Furthermore, it is proposed that the image capturing device captures the calibration object multiple times during each color state. With this configuration, the value range for the exposure duration and/or the exposure intensity can advantageously be precisely determined, in particular within a calibration sequence.

It is also proposed that the image capturing device varies the exposure time and/or the exposure intensity during each color state of the calibration object over essentially the entire intended value range of the exposure time and/or the exposure intensity as long as the image capturing device recognizes the calibration object. As a result of this configuration, the exposure duration and/or the exposure intensity can advantageously be determined reliably. Furthermore, an advantageously automated and rapid calibration of the image capturing device can take place as a result of this refinement. The range of values is preferably defined by minimum and maximum values for the exposure duration and/or the exposure intensity that can be set on the image capturing device. In principle, it is conceivable that the image capturing device no longer recognizes the calibration object, at least in part, at minimum and/or maximum values for the exposure duration and/or the exposure intensity, in particular if, depending on the ambient brightness, contours, in particular outer contours, of the calibration object no longer distinguish themselves from a background and /or an environment of the calibration object are distinguishable. The calibration object is preferably operated autonomously by the image acquisition device.

In addition, it is proposed that the calibration object assumes primary colors based on at least two different color models in the color states. With this configuration, different colors can advantageously be used for the calibration method. As a result, the exposure duration and/or the exposure intensity can advantageously be determined reliably for motif originals of different colors.

Furthermore, it is suggested that one of the color models is the CMY color model and another of the color models is the RGB color model. With this configuration, advantageously defined color models can be used.

It is also proposed that the color states be run through in a fixed order, with at least one primary color of a first color model of the at least two color models being followed by a primary color of a second color model of the at least two color models, which has the smallest spectral overlap with the primary color of the first color model. This configuration allows the sequence to be run through with color states that advantageously differ optically from one another. This allows an advantageously precise calibration of the image capturing device. Preferably, the first color model of the at least two color models is the CMY color model or the RGB color model, the second color model of the at least two color models being the CMY color model or RGB color model, which is not the first color model.

In addition, it is proposed that at least one primary color of the first color model of the at least two color models is a complementary color to a primary color of the second color model of the at least two color models. Through this issue design, color states that differ optically from one another can advantageously be achieved. This allows an advantageously precise calibration of the image capturing device.

In addition, the invention is based on a calibration method in which, in at least one method step, a position of a hand-held device, which is intended in particular for a defined application of paint to a surface, is determined in real time, the device being measured using an inertial measuring unit of the device and using a Camera of an image capturing device is monitored, with a prediction of a position of the device optical position data, which are determined by the image capturing device, and movement data, which are determined by the inertial measuring unit, are evaluated in combination.

It is proposed that the device has at least one light source, which emits a defined sequence of alternating color states for a predefined period of time, with the image capturing device detecting and evaluating the color states emitted by the light source. The configuration according to the invention enables an advantageously precise position determination to be achieved in real time. Furthermore, an advantageously simple synchronization of the image capturing device and the hand-held device can be achieved by the configuration according to the invention. This allows the defined application of paint to the surface to be particularly precise. As a result, an advantageously high quality of the paint application on the surface is achieved.

The device is preferably designed as a media application device, in particular a paint application device. The device preferably prints the surface depending on print data. The device preferably has a paint spray unit which is intended to apply paint in a defined manner to the surface. When moving over the surface, the device is preferably at an at least essentially constant distance from the surface. The inertial measuring unit preferably has at least one acceleration sensor and at least one gyroscope. The light source is preferably designed as an LED. Alternatively, the device can also be embodied as a detection device which is intended to detect at least one property of the surface and/or at least one property of a structure arranged below the surface. A property of the surface and/or a property of a structure arranged below the surface is preferably, for example, a surface type, a surface quality, a surface color or a position of a component, for example a cable or a water pipe.

The defined sequence of alternating color states of a predefined time duration preferably encodes a time stamp of the device. A time stamp of the device is preferably transmitted optically by means of the defined sequence of alternating color states of a predefined time duration. The time period is preferably 200 ms. In principle, however, the period of time can also be a different time that appears suitable to a person skilled in the art. The duration is preferably the same for all color states. Alternatively, it would be conceivable that the defined time periods in which the color states are emitted differ from one another for the different color states.

Preferably, the image capturing device is positioned stationary, in particular relative to the surface. The image capturing device can preferably be positioned on a stand unit, in particular a tripod, in a fixed position relative to the surface.

The image capturing device preferably captures the current position of the device. The current position of the device is preferably determined by means of at least one sensor means of the image capturing device, in particular the camera of the image capturing device. Preferably, the position of the device is determined relative to a virtual grid. The virtual grid is preferably generated by the image acquisition device. The virtual grid preferably reproduces an imaginary grid on the surface. The optical position data are preferably based on the virtual grid. Preferably, the optical position data indicates the position of the device relative to the virtual grid. Preferably, the optical position data indicates the position of the device on the surface.

Preferably, the image capture device captures a color of the color states and the length of time for which the color states are emitted. Preferably, the image capture device decodes the emitted sequence of alternating color states and determines the timestamp of the device.

Furthermore, it is proposed that before a position of the device is predicted, the time stamp of the device and a time stamp of the image capturing device are synchronized. With this configuration, the image acquisition device and the device can advantageously be synchronized quickly and easily. As a result, the position of the device can advantageously be predicted precisely. The time stamp of the device is preferably designed as a primary time stamp. Preferably will the time stamp of the image acquisition device is adapted to the time stamp of the device, in particular overwritten by it.

It is also proposed that a time stamp of the device is determined as a function of an image frequency of the camera of the image capturing device. As a result of this configuration, the defined sequence of alternating color states of a predefined time duration can advantageously be reliably detected and evaluated. The defined sequence of alternating color states of a predefined time period is preferably recorded more precisely the higher the image frequency of the camera.

In addition, it is proposed that the prediction is determined using an extended Kalman filter, which uses the motion data and the optical position data. With this configuration, optical position data and movement data can be evaluated particularly advantageously in combination to predict the position of the device. As a result, the position of the device can advantageously be precisely predicted or determined. The movement data and the optical position data are preferably merged with one another by means of the extended Kalman filter.

Furthermore, it is proposed that optical position data be transmitted from the image acquisition device, in particular wirelessly, to the device. With this configuration, an advantageously simple transmission of the optical position data can be achieved. As a result, an advantageously user-friendly system made up of the device and the image acquisition device can be provided. The optical position data are preferably transmitted to the device by means of Bluetooth, in particular Bluetooth Low Energy.

It is also proposed that the optical position data be transmitted, in particular wirelessly, to the device every 20 ms, with a prediction of a position of the device being calculated every 2.5 ms. With this refinement, the position of the device can advantageously be precisely determined in real time. Preferably, an iteration to calculate the prediction is performed every 2.5 ms.

In addition, it is proposed that for each iteration for calculating the prediction it is checked whether current optical position data and current movement data are available. With this refinement, an advantageously reliable prediction of the position of the device can be achieved.

Furthermore, it is proposed that only current movement data be used for the prediction if there is no current optical position data in the device. With this refinement, the prediction can advantageously be carried out permanently. Furthermore, this refinement makes it possible for optical position data to be transmitted at a lower frequency than a frequency of the iterations. Preferably, only current movement data is used for an iteration to calculate the prediction if there is no current optical position data in the device.

It is further proposed that historical information be used to estimate device drift and to correct future device drift predictions. With this refinement, an advantageously precise prediction of the position of the device can be achieved. A “drift” can be understood to mean that the prediction of the position of the device deviates from an actual position of the device, in particular due to measurement inaccuracies and/or calculation inaccuracies. Optical position data and/or movement data that do not have a current time stamp are preferably in the form of historical information. Data that has already been processed, in particular optical position data and/or movement data, is preferably stored in the device and, in particular, not deleted.

In addition, it is proposed that the optical position data be used for a data update, with the optical position data being used with a time stamp that is closest to the movement data. With this refinement, an advantageously precise prediction of the position of the device can be achieved even if no current optical position data is available. Furthermore, drifting of the device can advantageously be corrected by this configuration. The optical position data are preferably used regularly, in particular every 20 ms, for the data update. This preferably results in an update of a position of the device predicted by means of the movement data every 2.5 ms.

The calibration method according to the invention should not be limited to the application and embodiment described above. In particular, the calibration method according to the invention can have a number of individual elements, components and units as well as method steps that differs from the number of individual elements, components and units as well as method steps specified herein in order to fulfill a function described herein. In addition, the value ranges specified in this disclosure should also lie within the stated limits Values are seen as revealed and can be used at will.

character list

Further advantages result from the following description of the drawings. In the drawings an embodiment of the invention is shown. The drawings, the description and the claims contain numerous features in combination. The person skilled in the art will expediently also consider the features individually and combine them into further meaningful combinations.

• 1 ein erfindungsgemäßes Kalibrierungsverfahren in einer schematischen Darstellung,
• 2 eine Bilderfassungsvorrichtung und ein Kalibrierungsobjekt zur Durchführung des erfindungsgemäßen Verfahrens in einer schematischen Darstellung,
• 3 ein Ablaufschema einer Erfassungsphase eines ersten Verfahrensschritts des erfindungsgemäßen Kalibrierungsverfahrens in einer schematischen Darstellung,
• 4 eine von der Bilderfassungsvorrichtung zu erfassende Oberfläche mit Referenzmarkern in einer schematischen Darstellung,
• 5 ein digitales Bild mit einem verzerrten Abbild der zu erfassenden Oberfläche und mit einem virtuellen Gitter in einer schematischen Darstellung,
• 6 eine Gittereinheit des virtuellen Gitters in einer schematischen Darstellung,
• 7 eine Oberfläche mit einer auf die Oberfläche aufgebrachten oder aufprojizierten Motivvorlage, die von der Bilderfassungsvorrichtung erfasst wird, in einer schematischen Darstellung,
• 8 eine Anzeige der Bilderfassungsvorrichtung mit einem Abbild eines digitalen Bildes der erfassten Oberfläche in einer schematischen Darstellung,
• 9 eine Ansicht der zu erfassenden Oberfläche, über die ein handgehaltenes Gerät geführt wird, und der Bilderfassungsvorrichtung sowie ein Ablauf eines Verfahrensschritts in einer schematischen Darstellung und
• 10 ein Ablaufschema einer Vorhersageberechnungsphase in einer schematischen Darstellung.

Show it:

• 1 a calibration method according to the invention in a schematic representation,
• 2 an image acquisition device and a calibration object for carrying out the method according to the invention in a schematic representation,
• 3 a flow chart of a detection phase of a first method step of the calibration method according to the invention in a schematic representation,
• 4 a surface to be captured by the image capturing device with reference markers in a schematic representation,
• 5 a digital image with a distorted image of the surface to be recorded and with a virtual grid in a schematic representation,
• 6 a lattice unit of the virtual lattice in a schematic representation,
• 7 a surface with a motif template applied or projected onto the surface, which is captured by the image capturing device, in a schematic representation,
• 8th a display of the image acquisition device with an image of a digital image of the acquired surface in a schematic representation,
• 9 a view of the surface to be recorded, over which a hand-held device is guided, and the image recording device and a sequence of a method step in a schematic representation and
• 10 a flow chart of a prediction calculation phase in a schematic representation.

Description of the embodiment

In the 1 a calibration method 10 is shown. The calibration method 10 includes a first method step 12. The calibration method 10 includes a second method step 14. The calibration method 10 includes a third method step 16. In principle, the calibration method 10 can have additional method steps. The calibration method 10 is provided for the preparation of an operation and/or for the operation of a device 18, in particular a hand-held device. In the present case, the device 18 is designed as a media application device, in particular a paint application device.

In the 2 an image capture device 20 and a calibration object 22 are shown. The calibration method 10 is provided for calibrating the image capturing device 20 . The calibration method 10 is provided for calibrating a system composed of the device 18 and the image acquisition device 20 . In the first method step 12, an exposure duration and/or an exposure intensity of the image acquisition device 20, in particular an optimal value for the exposure duration and/or the exposure intensity of the image acquisition device 20, is/are determined as a function of the calibration object 22 acquired by the image acquisition device 20. The first method step 12 includes a detection phase 24, a calculation phase 26 and a calibration phase 28 (cf. 1 ).

In the present case, the image capturing device 20 is embodied as a mobile, personal electronic device. The image capturing device 20 is in the form of a smartphone or a tablet PC. In the present case, the image capturing device 20 is stationarily positioned on a stand unit 30 which is designed as a tripod. The image capturing device 20 has a camera. The image capturing device 20 has a display 32 which is embodied as a screen. The image acquisition device 20 is provided for image processing, in particular dynamic image processing. The image capturing device 20 is provided for capturing a digital image 34 of a surface 36 to be captured, in particular a motif template 60 on the surface 36 to be captured. In the present case, the surface 36 is designed as a wall surface of a room in a building. The image capturing device 20 is calibrated using the calibration object 22 before a digital image 34 of the surface 36 to be captured is recorded. The image capturing device 20 is calibrated in order to correctly expose the surface 36 to be captured by the image capturing device 20, in particular the original motif 60 on the surface 36 to be captured tet to be able to capture. An exposure setting of the camera, in particular the exposure duration and/or the exposure intensity, is adjusted using the calibration method 10 in order to be able to capture the surface 36 to be captured, in particular the original motif 60 on the surface 36 to be captured, using the camera and to be able to carry out image processing .

The camera has an optoelectronic image sensor. The image sensor is permanently exposed, in particular without a mechanical shutter. In order to record the digital image 34, the image sensor is exposed for a defined period of time, in particular the duration of the exposure. Alternatively, the exposure time could be designed as a shutter speed of the camera. The exposure strength is designed as a light sensitivity of the image sensor, in particular as an ISO value.

The exposure setting of the camera must be adjusted so that the camera can detect objects during image processing. If an exposure is too high, more light reaches the image sensor and consequently an image of the calibration object 22 becomes too white. On the other hand, if too little light reaches the image sensor, an image of the calibration object 22 becomes too dark. In both cases, the image processing might not be as reliable as required. This problem occurs very often when the ambient light, especially room light, is either too bright (high lux value) or too dark (very low lux value).

The calibration object 22 is positioned in a field of view 40 of the camera, in particular before the start of the acquisition phase 24 . The calibration object 22 is recognized by the image capturing device 20 . The image capturing device 20 recognizes the calibration object 22 from a shape of the calibration object 22. The shape of the calibration object 22 is stored in the image capturing device 20 and/or the image capturing device 20 has learned the shape of the calibration object 22.

The calibration object 22 has an illuminant 42 which is intended to assume different color states 44 . The illuminant 42 has at least one LED. The illuminant 42 emits colored light according to a respective color state of the different color states 44.

The image capturing device 20 captures the calibration object 22 in different calibration sequences in the different color states 44 of the calibration object 22 to determine the exposure duration and/or the exposure intensity. A flow chart 46 of the detection phase 24 is in FIG 3 shown. The calibration object 22 is captured by the image capture device 20 to perform the calibration sequences in the capture phase 24 . The image capture device 20 captures the calibration object 22 in each calibration sequence of the different calibration sequences in one of the different color states 44 of the calibration object 22. In each calibration sequence of the different calibration sequences, a value range for an exposure duration and/or an exposure intensity is determined, in which the image of the calibration object 22 meets at least one defined brightness condition. The calibration object 22, in particular an outer contour of the calibration object 22, can be distinguished from a background and/or an environment of the calibration object 22 under the at least one defined brightness condition.

The image capturing device 20 captures the calibration object 22 multiple times during each color state. During each color state of the calibration object 22 , the image capturing device 20 varies the exposure time and/or the exposure level over essentially the entire intended value range of the exposure time and/or the exposure level as long as the image capturing device 20 recognizes the calibration object 22 . The range of values is defined by minimum and maximum values for the exposure duration and/or the exposure intensity that can be set on the image capturing device 20 . In principle, it is conceivable that the image capturing device 20 no longer recognizes the calibration object 22 at least in part at minimum and/or maximum values for the exposure duration and/or the exposure intensity, in particular if, depending on the ambient brightness, contours, in particular outer contours, of the calibration object 22 are no longer of a background and/or an environment of the calibration object 22 are distinguishable. At least one predefined satisfaction factor is used to determine the brightness values at which the calibration object 22 lies in the value range. The at least one predetermined satisfaction factor is predetermined. For each color state, the full value range of exposure duration and/or exposure intensity that can be implemented by the camera, from absolutely dark to very bright, is used. When the image of the calibration object 22 becomes brighter, the calibration object 22 can be recognized. The calibration object 22 is considered to be detected by means of the at least one predetermined satisfaction factor until the image of the calibration object 22 becomes very bright and the calibration object 22 can no longer be recognized. The range of values for the exposure time and/or the exposure intensity defines a range in which the at least one specified satisfaction factor applies. The range of values for the exposure duration and/or the exposure intensity defines a range in which the calibration object 22 can be recognized with a high degree of satisfaction. The value range determined for the exposure duration and/or the exposure intensity is stored in the image capture device 20 for each calibration sequence.

In the present case, the calibration object 22 assumes primary colors in the color states 44 based on two different color models. One of the color models is the CMY color model and another of the color models is the RGB color model. The color states 44 are run through in a fixed sequence 48 in the detection phase 24 . A primary color of a first color model of the two color models is followed by a primary color of a second color model of the two color models, which has the smallest spectral overlap with the primary color of the first color model. In the present case, the first color model of the two color models is the RGB color model and the second color model of the two color models is the CMY color model. At least one primary color of the first color model of the two color models is a complementary color to a primary color of the second color model of the two color models.

In the present case, six calibration sequences are run through during the acquisition phase 24 . In the present case, the calibration object 22 runs through six different color states 44 during the detection phase 24, in which the calibration object 22 assumes a color in each case. In each calibration sequence, the calibration object 22 emits a different color state. The color that the calibration object 22 assumes in the different color states 44 is respectively blue, cyan, green, yellow, red or magenta. To ensure that a colored object, in particular the calibration object 22, is recognized under different ambient brightness conditions, in particular in a room, the detection of the object is calibrated using six primary colors of a color wheel. During the detection phase 24, a color state of the calibration object 22 is changed automatically. The calibration sequences are automatically executed sequentially. In the present case, the value range for the exposure duration and/or the exposure intensity is determined a total of six times. The range of values for the duration of exposure and/or the intensity of exposure is determined individually for each color state.

The calculation phase 26 follows the acquisition phase 24. In the calculation phase 26, the optimum value for the exposure duration and/or the exposure intensity of the image acquisition device 20 is determined. In the calculation phase 26, the optimal value for the exposure duration and/or the exposure intensity of the image capturing device 20 is specified for further phases in the image processing. The optimal value for the exposure time and/or the exposure intensity of the image capturing device 20 is calculated from the determined value ranges for the exposure time and/or the exposure intensity from all calibration sequences. The optimum value is designed as an average of the value ranges determined for the exposure duration and/or the exposure intensity from all calibration sequences. In principle, it would be conceivable for the calculation phase 26 to run parallel to the detection phase 24 in accordance with the calibration sequences.

In the calibration phase 28, the image capturing device 20 is set to the determined exposure duration and/or the determined exposure intensity. In the calibration phase 28, the image capturing device 20 is set to the calculated optimal value for the exposure duration and/or the exposure intensity. The exposure duration determined and/or the exposure intensity determined, in particular the optimum value, is preset in the camera. The image capturing device 20 is calibrated again whenever there is a significant change in the ambient brightness, with the image capturing device 20 re-capturing the calibration object 22 in the different calibration sequences in the different color states 44 of the calibration object 22 to determine the exposure duration and/or the exposure intensity.

In the second method step 14, the surface 36 to be captured by the camera of the image capturing device 20 is marked with physical reference markers 50, the camera creating the digital image 34 of the surface to be captured 36, with the reference marker 50 being used by the image capturing device 20 to create a virtual Grid 52 is generated, which is based on the reference markers 50, wherein pixels of the created image 34 grid elements 54 of the grid 52 are assigned, the grid elements 54 all having the same actual format. A perspective correction of a distortion of an image 56 of the surface 36 to be recorded in the image 34 is carried out using the reference marker 50, the distortion being caused by an angle between the surface 36 and a recording plane of the camera, the grid 52 being divided into grid units 58 which all have the same actual format and in which the same number of lattice elements 54 is arranged, with an actual height and width of the lattice units 58 being specified by the reference markers 50 . The angle between the surface 36 and the receiving plane of the Camera may be due to real world limitations and/or user inexperience, creating perspective of surface 36 . Calibration method 10 is for image capture device 20.

A physical location of each pixel of the digital image 34 is determined on the surface 36 using the reference markers 50 and the virtual grid 52 . The surface 36 to be detected is arranged in one plane. An object is imaged on the surface 36 to be detected. The object is designed as a template 60 motif. After being captured by the image capturing device 20, the original motif 60 can be applied to the surface 36, in particular permanently, by means of a medium application, in particular an application of paint, to the surface 36 and/or introduced into it.

The second method step 14 comprises a preparation phase 62, an acquisition phase 64, a division and initialization phase 66, a data transformation phase 68 and a data transfer phase 70 (cf. 1 ).

In the preparatory phase 62, the reference markers 50 are arranged on the surface 36, in particular in an adhesive manner, but can be detached without being destroyed. The reference markers 50 are arranged in a band. The band forms a frame which is arranged around the object imaged on the surface 36 to be detected, in particular the original motif 60 . The frame formed by the band is rectangular. The fiducial markers 50 each have a pattern disposed on a side of the fiducial markers 50 facing the surface 36 . The reference markers 50 have a square shape. All reference markers 50 are identical. Immediately adjacent reference markers 50 are spaced apart from one another by a distance that corresponds to a format of the reference markers 50 . Each fiducial marker 50 has an actual format of 100mm x 100mm. The reference markers 50 are identified by the image acquisition device 20, in particular by means of the pattern.

To create the digital image 34, light waves emanating from the surface 36 to be recorded are converted into digital signals in the recording phase 64 by means of the image sensor of the camera. The digital image 34 is broken down into a large number of pixels, each of which has color information. The digital image 34 is digitally processed by the image capture device 20 . In principle, the digital image 34 can also be displayed on the display 32 of the image acquisition device 20 .

In the division and initialization phase 66, a space within the band of reference markers 50 is divided into the grid units 58, in particular starting from individual reference markers 50. The virtual grid 52 is generated digitally by means of the image acquisition device 20. The virtual grid 52 forms a regular structure of grid lines 72 . The virtual grid 52 is based on the reference markers 50. The virtual grid 52 forms a matrix to which pixels of the digital image 34 are assigned. The virtual grid 52 is inserted into the digital image 34 and/or stored as a standalone data set as a reference for mapping the pixels. The virtual grid 52 is distorted in the digital image 34 depending on an arrangement of the reference markers 50 in the image 56 of the surface 36 to be detected.

The virtual grid 52 comprises a plurality of grid units 58. In the partitioning and initialization phase 66, the grid 52 is partitioned into the grid units 58. FIG. As a result, an accumulation of errors that would result from estimates of values can advantageously be completely eliminated. This allows faults within a grating unit 58 to be isolated. Consequently, errors are reset with each grating unit 58 and advantageously cannot propagate. The grid units 58 have the same format as the reference markers 50. Each grid unit 58 has an actual format of 100mm x 100mm. Directly adjacent grating units 58 are arranged without a spacing from one another. Each lattice unit 58 includes a plurality of lattice elements 54. The actual format of the lattice elements 54 is square in each case. Each grid element 54 has an actual format of 1mm x 1mm. Immediately adjacent lattice elements 54 are arranged without a spacing from one another.

A coordinate system 74 of the grid 52 starts from an upper left corner of a virtual frame which is immediately surrounded by the reference markers 50 . A (0, 0) point 76 of the grid 52, specifically an origin point of the coordinate system 74, is defined in the upper left corner of the frame. The coordinate system 74 has an x-axis 78 and a y-axis 80 . The coordinate system 74 extends horizontally along the x-axis 78 and vertically along the y-axis 80 in 1 mm increments. A physical meaning of x and y coordinates forming a (x,y) point of grid 52 represents a displacement in mm on x-axis 78 and y-axis 80, respectively, measured as a relative distance between the (x,y) point and the (0,0) point 76, the origin point.

The grid units 58 each have five properties, namely four corner coordinates A, B, C, D (bottom left, bottom right, top right, top left) related to pixels of the digital image 34 and absolute coordinates of a corner coordinate 82 top left in mm related to the coordinate system 74. The corner coordinate 82 corresponds to the corner coordinate D (cf . 5 and 6 ). Absolute coordinates are determined using reference markers 50 and/or using white spaces between reference markers 50 .

Horizontal and vertical pairs of reference points 84 , 86 , 88 , 90 are defined within a virtual frame, which is immediately surrounded by the reference markers 50 , and are each arranged on a grid line 72 of the grid 52 . The virtual frame is directly adjacent to the reference markers 50, in particular to the tape. Reference points of the pairs of reference points 84, 86, 88, 90 are each arranged at a corner point of a reference marker 50 on the virtual frame. A first pair of vertical reference points 84 has an upper point T1 and a lower point B1. A second pair of vertical reference points 86 has a top point T2 and a bottom point B2. A first horizontal reference point pair 88 has a left point L1 and a right point R1. A second horizontal reference point pair 90 has a left point L2 and a right point R2. Within a pair of reference points 84, 86, 88, 90, the points have the same absolute coordinates in mm (not in pixels) on the x-axis 78 for the vertical pairs of reference points 84, 86 and on the y-axis 80 for the horizontal pairs of reference points 88, 90 .

In order to calculate the corner coordinates A, B, C, D of grid units 58 in relation to pixels of digital image 34, gradients of grid lines 72 are determined, with the gradients being caused by the distortion, with grid lines 72 of grid units 58 each adjoining to grid units 58 nearest reference markers 50 abut. The grid units 58 are each defined by four grid lines 72 of the virtual grid 52 . The gradients of the grid lines 72 are determined using the horizontal pairs of reference points 88, 90 and the vertical pairs of reference points 84, 86, respectively. The corner coordinates A, B, C, D of a grid unit 58 are in the form of intersection points of the grid lines 72 of the grid unit 58 . The corner coordinates A, B, C, D of the grid unit 58 are determined, in particular calculated, using the slopes of the grid lines 72 of the grid unit 58 and using the horizontal and vertical pairs of reference points 84, 86, 88, 90.

A basic equation for calculating a slope between two points with known coordinates is given in formula (1). $$m=\frac{{\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">y</figure-callout>}}_1-y_2}{x_1-x_2}$$

A gradient m ₁ between the points L1 and R1 is calculated using formula (2). $${\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">m</figure-callout>}}_1=\frac{\mathrm{<figure-callout\; id="1"\; label="R"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">R</figure-callout>}1.y-\mathrm{<figure-callout\; id="1"\; label="L"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">L</figure-callout>}1.\mathrm{<figure-callout\; id="1"\; label="y\; R"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">y</figure-callout>}}{R}$$

A gradient m ₂ between the points T1 and B1 is calculated using formula (3). $${\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="DE102021210375A1\_0021.png"\; state="\{\{state\}\}">m</figure-callout>}}_2=\frac{\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">T</figure-callout>}1.y-\mathrm{<figure-callout\; id="1"\; label="B"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">B</figure-callout>}1.\mathrm{<figure-callout\; id="1"\; label="y\; T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">y</figure-callout>}}{T}$$

$$Z.y=T1.y+m_1\ast \left(Z.x-T1.x\right)$$

In relation to these points, the gradients m ₁ and m ₂ are known, as are the coordinates of points T1 and L1. An intersection point Z is calculated using formulas (4) and (5). $$Z.x=\frac{\mathrm{<figure-callout\; id="1"\; label="L"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">L</figure-callout>}1.y-\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">T</figure-callout>}1.y+{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">m</figure-callout>}}_1\ast \mathrm{<figure-callout\; id="1"\; label="T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">T</figure-callout>}1.x-{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="DE102021210375A1\_0021.png"\; state="\{\{state\}\}">m</figure-callout>}}_2\ast \mathrm{<figure-callout\; id="1"\; label="L"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">L</figure-callout>}1.x}{m_1-{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="DE102021210375A1\_0021.png"\; state="\{\{state\}\}">m</figure-callout>}}_2}$$

$$Z.y=\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">T</figure-callout>}1.y+{\mathrm{<figure-callout\; id="1"\; label="m"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">m</figure-callout>}}_1\ast \left(Z.x-\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">T</figure-callout>}1.x\right)$$

$$Z.y=T1.y+m_2\ast \left(Z.x-T1.x\right)$$

There is a mathematical singularity in formula (1) in the event that the first and/or second vertical reference point pairs 84, 86 are vertical lines with undefined slopes due to a division by zero. For this particular case, the formulas for calculating the Z intersection point are derived from two-line equations. For example, the intersection point Z is calculated using formulas (6) and (7) when the slope m ₁ is undefined. $$Z.x=\mathrm{<figure-callout\; id="1"\; label="L"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">L</figure-callout>}1.x$$

$$Z.y=\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">T</figure-callout>}1.y+{\mathrm{<figure-callout\; id="2"\; label="m"\; filenames="DE102021210375A1\_0021.png"\; state="\{\{state\}\}">m</figure-callout>}}_2\ast \left(Z.x-\mathrm{<figure-callout\; id="1"\; label="T"\; filenames="DE102021210375A1\_0020.png"\; state="\{\{state\}\}">T</figure-callout>}1.x\right)$$

The remaining corner coordinates C, A, B (top right, bottom left, bottom right) related to pixels of the digital image 34 are calculated in the same way.

In the present case, the result of the partitioning and initialization phase 66 is an array with 68 rectangles, which are successively processed in the following data transformation phase 68 .

In the data transformation phase 68, a mapping matrix for all points, particularly within the reference markers 50, is created. The mapping matrix is configured as a look-up table. The lookup table is created in a parsing process. The parsing process is performed in the data transformation phase 68 . The look-up table is implemented as a C matrix. A size of the lookup table corresponds to one Resolution of the camera and/or a resolution of a display 32 of the image capture device 20. As a result, a number of rows in the lookup table corresponds to a display width of the display 32 and a number of columns in the lookup table corresponds to a display height of the display 32. The lookup table has a number of data fields that corresponds to the resolution of the camera and/or the resolution of the display 32. The lookup table contains color information corresponding to the pixels of image 34. A size of the lookup table is advantageously minimized since it is initialized to integer type data at 16 bits (between -32768 and +32767).

Each location in the look-up table (row number and column number) contains an indication of the coordinates in mm of the pixels that have row numbers and column numbers indexed. When a position marker of the image capture device 20 is located at an (x,y) point, the real coordinate can be found by accessing the stored point in the look-up table. As a result, color information can be accessed particularly quickly.

When the lookup table is initialized, all stored values are points with coordinates (-1,-1). In the following, the real positions are stored in the look-up table. This is done only for the pixels related to the digital image 34 that are located within the reference markers 50.

In the data transformation phase 68, all real coordinates and values for each pixel of the digital image 34 that is within the band are calculated. Each grid unit 58 is used by means of an algorithm and a ratio of pixels to real coordinates for each pixel within the grid unit 58 is determined for each grid unit 58 . The grid units 58 are each split into 10,000 grid elements 54 of 1 mm×1 mm.

At least one distance between corner coordinates A, B, C, D of a respective grid unit 58 in relation to pixels of the digital image 34 is calculated in order to calculate an assignment of pixels to grid elements 54 . A distance between vertices of each grid element 54 within a grid unit 58 becomes dependent on a number of horizontally and/or vertically adjacent grid elements 54 in the grid unit 58 and on a distance between corner coordinates A, B, C, D of a respective grid unit 58 in terms of pixels of the digital image 34 is determined. Corner coordinates A, B, C, D of the grid units 58 are calculated independently of one another using different reference markers 50 .

Formula (8) is used to calculate a distance between two points with the coordinates (x ₁ ,y ₁ ) and (x ₂ ,y ₂ ). $$Dist=\sqrt{{\left({\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="DE102021210375A1\_0021.png"\; state="\{\{state\}\}">y</figure-callout>}}_2-y_1\right)}^2+{\left(x_2-x_1\right)}^2}$$

The formula (8) is used to calculate a distance Dist _DA and a distance Dist _DC . Since a size of the grid units 58 is 100mm x 100mm, the distances Dist _DA and Dist _DC are divided into 100 equal parts. A position of the grid elements 54 is calculated in double precision floating point numbers.

$$W.x=D.x-\frac{d\left(D.x-C.x\right)}{Dist}$$

$$W.y=D.y-\frac{d\left(D.y-C.y\right)}{Dist}$$

After calculating the distances between consecutive points within a grid unit 58, vertical and horizontal pairs are identified. For points of the vertical pairs, the upper left corner coordinate D of the grid unit 58 is taken as the pivot point. A next point W(x,y) can thus be determined using formulas (10) and (11), since a distance d, which in the present case is calculated using formula (9), the initial corner coordinate D and a final corner coordinate C are known. $$\mathrm{i.e}=\frac{Dis{\mathrm{<figure-callout\; id="100"\; label="t\; D\; C"\; filenames="DE102021210375A1\_0018.png,DE102021210375A1\_0026.png"\; state="\{\{state\}\}">t</figure-callout>}}_{DC}}{100}$$

$$W.x=D.x-\frac{\mathrm{i.e}\left(D.x-C.x\right)}{Dist}$$

$$W.y=D.y-\frac{\mathrm{i.e}\left(D.y-C.y\right)}{Dist}$$

Formulas (10) and (11) are applied to the line DC to determine the top 100 points of the vertical pairs. For the lower points, formulas (10) and (11) are applied to line AB. It should be noted that a distance between consecutive points is different. The same applies to the horizontal pairs on lines DA and CB. When the pairs are known, the intersections of each vertical pair with the horizontal pairs are calculated. This allows the points P(x,y) to be determined. Since the calculations are made with double precision, the coordinates are usually non-integer values.

The look-up table and media applicator operate on integer coordinates, so approximations to the nearest integer values must be made. Since values with double precision were processed up to this point in time, an error is due to a calculation precision reduced to decimal places. Furthermore, the error is localized for each point, with no possibility of error propagation.

Values calculated within the grid unit 58 using the algorithm previously described are the actual values measured in mm relative to the absolute top left corner coordinate, the (0,0) point 76, relative to the coordinate system 74. Absolute coordinates of the grid elements 54 can advantageously be calculated simply by adding the relative mm coordinates to the absolute upper left corner coordinate D. Absolute coordinates of the grid elements 54 are calculated by adding the coordinates of the corner coordinates D of the grid unit 58 .

Finally, the calculated absolute coordinates of each grid element 54 of the grid unit 58 are stored in the look-up table. A real mm coordinate value is stored according to the correct index, which represents its position in pixels.

The division and initialization phase 66 and the data transformation phase 68 form a multi-step algorithm in order to be able to calculate the real points from the digital image 34 .

In the data transfer phase 70, the lookup table is enabled by the image capture device 20 for access to the lookup table. The look-up table is transmitted from the image capture device 20 to the device 18, which is in particular hand-held. Device 18 accesses the look-up table. Device 18 uses the look-up table to determine color information at locations on surface 36.

The look-up table is used for a defined application of paint to the surface 36 by means of the device 18 depending on a position of the device 18, in particular a hand-held device, in particular in a further phase or a further method step or a further method. The media application device applies at least one medium, in particular colors, to the surface 36 in a defined manner in accordance with the look-up table.

A position of the device 18, in particular a hand-held device, is determined by means of at least one sensor means of the image capturing device 20, the position of the device 18 being used for navigating in the look-up table. The image capturing device 20 captures the device 18 relative to the surface 36. The image capturing device 20 determines a position of the device 18 by means of a position marking of the device 18, which is recognized by the image capturing device 20, in particular by the sensor means. In the present case, the position marker is designed as a light source 92 of the device 18 . The image capturing device 20 transmits the position of the device 18 to the device 18. During a media order, a position of the position marker, in particular the light source 92 of the device 18, with respect to the pixels of the digital image 34 is detected. The real coordinates of the detected position marker are given by the look-up table, which has pre-stored millimeter values for all pixels in the digital image 34 or on the display 32.

In principle, it would also be conceivable for storing and indexing a virtual position map to use a hash table, a vectorial objective C implementation or other options for mapping between two objects that appear suitable to a person skilled in the art. The disadvantage of pre-implemented structures is that they require a lot of memory to store values and it takes a lot of calculation time to locate a specific object. Time-effective object localization is required because information about a specific point must be accessible immediately during operation. Otherwise, delays and backlogs could occur in an augmented reality view, which could lead to high inaccuracy in a media job and also to CPU overuse.

In the third method step 16 of the calibration method 10, a position of the hand-held device 18, which is intended in particular for a defined application of paint to the surface 36, is determined in real time, with the device 18 using an inertial measuring unit of the device 18 and using the camera of the Image capturing device 20 is monitored, with a prediction of a position of the device 18 being evaluated in combination with optical position data that are determined by the image capturing device 20 and movement data that are determined by the inertial measuring unit. The device 18 has the illuminant 92 which emits a defined sequence of alternating color states of a predefined time duration, with the image capturing device 20 detecting and evaluating the color states emitted by the illuminant 92 . The illuminant 92 of the device 18 is in the form of an LED.

The device 18 is freely movable on the surface 36 along the x-axis 78 and the y-axis 80 within defined limits. When moving across the surface 36, the device 18 is at an at least substantially constant distance from the surface 36. At certain points on surface 36, device 18 is designed to perform a task. For example, the device 18 is intended to apply a dot to the surface 36 at a location on the surface 36 when the device 18 is placed at that location (cf. 8th ). However, the device 18 is blind, whereby the device 18 itself does not know the position of the device 18 relative to the surface 36 at the current time. Therefore, in order to operate the device 18 in real time, it is necessary to know exactly where the device 18 is located. Since the optical position data on the current position of the device 18 wirelessly reaches the device 18, the device 18 has already moved and the optical position data is already out of date due to a latency time in wireless transmission and an image processing time. Therefore, in order for the device 18 to perform real-time tasks, a current position of the device 18 must be predicted at all times.

The third method step 16 comprises a detection phase 94, a transmission phase 96, a prediction calculation phase 98 and an execution phase 100 (cf. 1 and 9 ).

In the 7 the surface 36 is shown with a motif template 60'. In the 9 the surface 36 is shown with a motif template 60". The surface 36 is captured by the image capturing device 20 in the capturing phase 94. In FIG 8th An image 56 of the surface 36 to be captured is shown on the display 32 of the image capturing device 20 . Image capture device 20 is positioned stationary relative to surface 36 . In the present case, the image capturing device 20 is positioned on the stand unit 30 embodied as a tripod, stationary relative to the surface 36 .

In the acquisition phase 94, a time stamp of the device 18 and a time stamp of the imaging device 20 are synchronized and the current position of the device 18 is determined optically. Before predicting a position of the device 18, the timestamp of the device 18 and a timestamp of the imaging device 20 are synchronized. The time stamp of the device 18 is in the form of a primary time stamp. The time stamp of the image acquisition device 20 is adapted to the time stamp of the device 18, in particular overwritten by it. The device 18 controls transmission of the defined sequence of alternating color states for a predefined period of time. The defined sequence of alternating color states of a predefined time duration encodes the time stamp of the device 18. The time stamp of the device 18 is transmitted optically by means of the defined sequence of alternating color states of a predefined time duration. In the present case, the duration is 200 ms. In principle, however, the period of time can also be a different time that appears suitable to a person skilled in the art. In the present case, the duration is the same for all color states. The image capture device 20 captures a color of the color states and the length of time for which the color states are emitted. The image capture device 20 decodes the emitted sequence of alternating color states and determines the time stamp of the device 18. The time stamp of the device 18 is determined as a function of an image frequency of the camera of the image capture device 20. The defined sequence of alternating color states of a predefined duration is captured more precisely the higher the frame rate of the camera. The time stamp of the device 18 begins when the device 18 emits a first color state of the defined sequence of alternating color states of a predefined time duration. Capture of the device 18 timestamp begins when the image capture device 20 identifies images 34 of the camera where the first color state is detected. Based on the defined sequence of alternating color states of predefined duration emitted by the device 18, the exact time stamp of the device 18 is calculated in the image capture device 20. The optical position data, which are determined by the image acquisition device 20 , are synchronized with the time stamp of the device 18 by means of the time stamp of the device 18 .

The image capturing device 20 captures the current position of the device 18. The current position of the device 18 is determined by means of at least one sensor means of the image capturing device 20, in particular the camera of the image capturing device 20. The position of the device 18 relative to the virtual grid 52 is determined. The virtual mesh 52 is generated by the image capture device 20 . The virtual grid 52 reproduces an imaginary grid on the surface 36 . The optical position data is based on the virtual grid 52. The optical position data indicates the position of the device 18 relative to the virtual grid 52. FIG. The optical position data indicates the position of device 18 on surface 36 .

In principle, it is conceivable that the calibration object 22 is part of the hand-held device 18 or is designed as the hand-held device 18 . In principle, the first method step 12 could also be integrated into the third method step 16 , in particular into the detection phase 94 of the third method step 16 .

In the transmission phase 96, optical position data are transmitted from the image acquisition device 20 to the device 18, in particular wirelessly. In the present case, the optical position data is transmitted to the device 18 by means of Bluetooth, in particular Bluetooth Low Energy. A wireless transmission of the optical position data to the device 18 takes place every 20 ms.

In the prediction calculation phase 98, the optical position data transmitted to the device 18 is processed to predict the position of the device 18. In the prediction calculation phase 98 the optical position data determined by the image capturing device 20 and the movement data determined by the inertial measuring unit are evaluated in combination by the device 18 . The inertial measurement unit has at least one acceleration sensor and at least one gyroscope.

The prediction is determined using an extended Kalman filter using the motion data and the optical position data. The movement data and the optical position data are merged with one another using the extended Kalman filter.

$$yk+1=Hkxk+rk,rk~N\left(\mathrm{0,}Rk\right)$$

To predict the position of the device 18, a status xk = [xk; yk; ek; ak; ẍk; ÿk φk; ωk] analyzed. k stands for a time step, xk and yk are different positions, ẋk and yk are different velocities, ẍk and ÿk are different accelerations, φ is a rotation angle and ω is an angular velocity. The extended Kalman filter is based on a motion model (12) and a measurement model (13). $$xk+1=Akxk+Bk\mathrm{and}k+qk,qk~N\left(\mathrm{0,}Qk\right)$$

$$yk+1=Hkxk+\mathrm{right}k,\mathrm{right}k~N\left(\mathrm{0,}Rk\right)$$

The extended Kalman filter is used for data fusion between the inertial measurement unit of the device 18 and the camera of the imaging device 20 . The extended Kalman filter contains a status vector uk = [ẍ _IMU ;ÿ _IMU ;φ _IMU ]. ẍ _IMU and ÿ _IMU describe an acceleration of the device 18 based on the at least one acceleration sensor, and φ _IMU describes a roll angle of the device 18, with φ _IMU being calculated using formula (14). $${\phi }_{IMu}={\phi }_{Acc}+{\omega }_{Gy\mathrm{right}O}T$$

The extended Kalman filter contains a measurement vector yk = [x _CAM ; y _CAM ;ω _IMU ]. X _CAM and y _CAM describe the position of the device 18 from the camera and transmitted via Bluetooth Low Energy, and ω _IMU describes a roll speed of the device 18 calculated using formula (15) from the gyroscope. $${\omega }_{IMu}={\omega }_{Gy\mathrm{right}O}$$

A prediction of a position of the device 18 is calculated every 2.5 ms. An iteration to calculate the prediction is performed every 2.5 ms. In the 10 A flowchart 102 of the extended Kalman filter is shown. In a first step 104 the extended Kalman filter is started. In a second step 106, movement data, in particular new movement data, are read in and stored in a buffer of the device 18. For each iteration for calculating the prediction, a third step 108 checks whether current optical position data and current movement data are present.

$$\dot{x}k+1=\dot{x}k+\ddot{x}kT$$

For a case 110 in which no, in particular current, optical position data is present, current movement data is a single input into the extended Kalman filter. In this case, in an alternative step 112, equations of motion (16) and (17) are used to predict the position of the device 18. $$xk+1=xk+\dot{x}kT+\frac{1}{2}\ddot{x}\mathrm{<figure-callout\; id="2"\; label="k\; T"\; filenames="DE102021210375A1\_0021.png"\; state="\{\{state\}\}">k</figure-callout>}T2$$

$$\dot{x}k+1=\dot{x}k+\ddot{x}kT$$

Only current movement data is used for the prediction if there is no current optical position data in the device 18 . For an iteration to calculate the prediction, only current movement data is used if there is no current optical position data in the device 18 .

For a case 114 in which current optical position data is present, data fusion is carried out. As new optical position data becomes available, more data is available to predict the device 18 position. The extended Kalman filter is used for data fusion, with movement data being provided for the prediction and optical position data for a data update. In this case, the synchronization of the time stamps of the device 18 and the image acquisition device 20 must be ensured.

In a fourth step 116, a search is made for movement data that has a closest time stamp to the optical position data. The data fusion of movement data for the prediction and of optical position data for the data update is then carried out in a fifth step 118 . A correction is made by the data update. The optical position data is used for the data update, with the optical position data being used with a time stamp that is closest to the movement data. The optical position data are regularly frequently, in particular every 20 ms, for data updates. This results in an update of a position of the device 18 predicted every 2.5 ms by means of the movement data. After the data fusion, however, a status vector from the past is estimated. Therefore, a current timestamp must be reverted to. Historical information is used to estimate device 18 drift and to correct future predictions regarding device 18 drift. Optical position data and/or movement data that do not have a current time stamp are in the form of historical information. Data that has already been processed, in particular optical position data and/or movement data, is stored in the device 18 and, in particular, is not deleted. In a sixth step 120, an average of the movement data between the time stamp of the optical position data and the current time stamp is calculated. The equations of motion (16) and (17) between these time stamps are used in a seventh step 122 to predict a new status vector. After completion of the alternative step 112 or the seventh step 122, an end 124 of the individual iteration occurs. The new status vector represents the current position of device 18.

In the execution phase 100, a media order, in particular an ink order, takes place after determining the current position of the device 18 and by accessing the look-up table. The device 18 prints the surface 36 depending on print data. The device 18 has a paint spray unit which is provided for applying paint to the surface 36 in a defined manner.

Claims (10)

Calibration method (10) for an image capturing device (20), wherein in at least one method step (14) a surface (36) to be captured by a camera of the image capturing device (20) is marked with physical reference markers (50), a digital image being recorded by means of the camera (34) of the surface (36) to be recorded is created, with the reference markers (50) being used by the image recording device (20) to generate a virtual grid (52) which is oriented to the reference markers (50), with pixels of the created image (34) Grid elements (54) of the grid (52) are assigned, the grid elements (54) all having the same actual format, characterized in that the reference marker (50) is used to perform a perspective correction of a distortion present in the image (34) of a Image (56) of the surface to be detected (36) is performed, the distortion by an angle between the surface (36) and a recording plane e of the camera, the grid (52) being divided into grid units (58), all of which have the same actual format and in which an equal number of grid elements (54) are arranged, with an actual height and width of the grid units (58) are specified by the reference markers (50). Calibration procedure (10) according to claim 1 , characterized in that horizontal and vertical pairs of reference points (84, 86, 88, 90) are defined within a virtual frame, which is directly surrounded by the reference markers (50), each of which is on a grid line (72) of the grid (52) are arranged. Calibration procedure (10) according to claim 1 or 2 , characterized in that gradients of grid lines (72) are determined for the calculation of corner coordinates (A, B, C, D) of the grid units (58), the gradients being caused by the distortion, the grid lines (72) of the grid units ( 58) rest against the reference markers (50) closest to the grid units (58). Calibration method (10) according to one of the preceding claims, characterized in that at least one distance between corner coordinates (A, B, C, D) of a respective grid unit (58) is calculated to calculate an assignment of pixels to grid elements (54). Calibration method (10) according to one of the preceding claims, characterized in that a distance between corner points of each grid element (54) within a grid unit (58) depends on a number of horizontally and/or vertically adjacent grid elements (54) in the grid unit (58 ) and a distance between corner coordinates (A, B, C, D) of a respective grid unit (58). Calibration method (10) according to one of the preceding claims, characterized in that a position of the grid elements (54) is calculated in floating-point numbers with double precision. Calibration method (10) according to one of the preceding claims, characterized in that a lookup table is created in a parsing process, a size of the lookup table corresponding to a resolution of the camera and/or a resolution of a display (32) of the image acquisition device (20). Calibration procedure (10) according to claim 7 , characterized in that the look-up table from the image acquisition device (20). a device (18), in particular a hand-held device, is transmitted. Calibration procedure (10) according to claim 7 or 8th , characterized in that the look-up table is used depending on a position of a device (18), in particular a hand-held device, for a defined application of paint to the surface (36) by means of the device (18). Calibration method (10) according to one of Claims 7 until 9 , characterized in that a position of a device (18), in particular a hand-held device, is determined by means of at least one sensor means of the image capturing device (20), the position of the device (18) being used for navigating in the look-up table.

Images