DE102018114222A1 - Procedure for examining matching test objects - Google Patents

Abstract

In order to create a method for examining at least one test criterion of matching test objects with any position in a test room, which allows the complete examination of all test objects moving through the test room, in particular also the examination of position-dependent test criteria, a test is proposed in a test room that is in the detection area several digital image recording devices and is free of built-in components. The test objects are moved individually and without guidance through the test room and recorded simultaneously by the digital image recording devices. In a data processing unit, a three-dimensional reconstruction of an object image from the recorded images takes place in a three-dimensional space. In addition, a three-dimensional reference image is provided in the three-dimensional space in the data processing unit, which has the target values for the test criteria to be examined. The image registration to be carried out as part of the method according to the invention for preparing the evaluation finally serves the purpose of superimposing the object image and the reference image, i.e. to bring them into the best possible agreement (transformation). The superimposed images are evaluated by comparing the image information of each object image representing the test criterion with the image information of the reference image representing the test criterion.

Description

The invention relates to a method for examining at least one test criterion of matching test objects in a test room.

The test objects are, for example, rotationally symmetrical hollow bodies produced by plastic injection molding with an end opening, a bottom surface opposite the opening and a lateral surface extending between the bottom surface and the opening, such as containers, preforms for producing containers or container closures.

For quality assurance of mass-produced test objects, it is known from the prior art to continuously examine at least one test criterion of the test objects produced using optical inspection systems. The test criteria are selected in such a way that typical properties and / or errors of the test object can be determined during the examination. Typical defects of injection molded hollow bodies are dimensional defects, material defects and color defects. Dimensional errors can relate to the length, shape and diameter of the hollow body. Material defects can result, for example, from an incorrect length of the injection point, incompletely melted material, bubbles, oil splashes and contamination. Color errors include color and intensity deviations as well as incorrect amounts of UV blockers.

From the EP 2 112 502 A1 A method is known in which preforms are checked for mechanical and optical errors in a continuous process using an optical inspection system and defective preforms are automatically eliminated. The device for carrying out the method comprises three digital cameras which are offset at an angle of 120 degrees in the image plane and which produce an almost complete image of the wall of each preform. The captured images from the cameras are transferred to an electronic processing unit and compared with stored reference images. Depending on the result of the comparison, the processing unit issues a command to discard the unaccepted preforms in the event of non-tolerable deviations. The preforms are transported and guided while the images are being taken by overhead transport with the mouth facing upwards. For overhead transportation, a horizontally running vacuum belt is provided on the back, on which the preforms are held at their mouths exclusively by the vacuum.

From the DE 10 2013 102 653 A1 A device and a method for camera-based examination of matching test objects is known, in which the test objects are guided during the recording by two belts, each of which rotates endlessly around two drive wheels. Defective test objects are sorted out by changing the trajectory of ejected test objects.

For the aforementioned camera-based examination methods, it is necessary that the test objects have to be aligned in a specific position with respect to each digital camera.

From the WO 2014/147176 A1 a camera-based examination method is already known exclusively for examining the color properties of preforms, which does not require a specific position of the preforms in relation to the digital camera. The preforms, in particular in free fall after leaving a transport device, are disorderly transferred into a receptacle, the images of the preforms being taken by means of the digital camera between leaving the transport device and transferring them into the receptacle. Several preforms in free fall can be identified in the image. Since the preforms fall out of order, the image of some preforms may be incomplete, since they are completely or partially covered by other preforms in the image. In addition, some preforms are imaged at an angle to the image plane, which means that only a part of their surface can be seen in the captured image. In extreme cases, a preform with its longitudinal axis is in free fall exactly in the viewing direction of the digital image recording device. In this case you could only see the head or bottom of the preform in the picture. To examine the color properties, one or more preforms are identified in each image, which were in free fall and when the image was taken at such an angle with respect to the image plane that its entire length can be seen as completely as possible in the image. Only these preforms can then be compared with a reference image of a correct preform in order to check the color quality of the preform to be checked.

That from the WO 2014/147176 A1 Known methods do not allow all preforms to be checked. In addition, no location-dependent test criteria can be examined. In the extreme case, it is conceivable in the known method that not a single image of a preform in a recorded image is oriented in such a way that a meaningful evaluation by the processing device is possible.

Based on this prior art, the object of the invention is to provide a method for examining at least one test criterion of matching test objects with any position in a test room, which allows the complete examination of all test objects moving through the test room, in particular also the examination of position-dependent test criteria ,

This object is achieved by a method with the features of claim 1. Advantageous refinements of the method result from the subclaims.

The test area is located in the detection area, hereinafter also referred to as the field of vision, of several digital image recording devices and is free of built-in components. The test area is the overlap area of the intersecting fields of view of the image recording devices or a defined partial area thereof. The subarea can be determined by means of software in a data processing unit for the captured images. The occasional, unguided movement of the test objects through the test room ensures that there is always only one test object in the test room during the recording. The images recorded simultaneously by the digital image recording devices always only depict one and the same test object from different directions. Covers by other test objects in the figure are excluded.

The images of each test object recorded simultaneously by the digital image recording devices are transmitted to the data processing unit. A three-dimensional reconstruction of an object image from the recorded images takes place in a three-dimensional space in the data processing unit. The three-dimensional space is the reference system in which the position and location of the reconstructed object image is known. The reference system is, for example, an origin coordinate system (world coordinate system). The position of the object image in the coordinate system can be specified by specifying the coordinates and the position by specifying the orientation and the position angle in the coordinate system.

A high computing effort is required for the three-dimensional reconstruction of the object image. The computational effort can be reduced by only three-dimensionally reconstructing the areas relevant to each test criterion from the recorded images of each test object. If the test object is, for example, a container closure and the test criterion is the diameter of the front opening of the container closure, it is not necessary to carry out a three-dimensional reconstruction of the entire container closure. Rather, it is sufficient for the investigation of the diameter if the image of the mouth area of the container closure is reconstructed from the recorded images in the three-dimensional space.

In addition, a three-dimensional reference image in the three-dimensional space, i.e. provided in the same reference system, the reference image having the target values for the test criteria to be examined. The reference image can completely depict a "correct" test object. In order to reduce the computing power in the context of the subsequent image registration, however, it can be expedient that the reference image only contains the image information of the test object that represents each test criterion. For a diameter determination of the container closure, it is sufficient, for example, to provide only the image information representing the correct diameter in the three-dimensional space as a reference image.

The image registration to be carried out as part of the method according to the invention for preparing the evaluation serves the purpose of superimposing the object image and the reference image, i.e. to bring them into the best possible agreement (transformation). During the transformation, the object image can be overlaid with the reference image or the reference image with the object image.

1. (1) Das Bestimmen von Merkmalen kann durch automatische Extraktion einer Anzahl von Merkmalen aus dem Objektbild und dem Referenzbild erfolgen. Als Merkmale kommen Regionen, Linien und Punkte in Betracht. Als Regionen eignen sich Flächen, die sich von den umgebenden Flächen deutlich abheben. Linien können als Konturen oder als Begrenzungen von Regionen oder als Linien selbst auf den Bildern vorhanden sein. Punkte in den Bildern können als Schnittpunkte von Linien oder als Ecken von Konturen gegeben sein. Vorzugsweise findet bei dem erfindungsgemäßen Verfahren die Bestimmung der Merkmale über die vorgenannten Merkmalsgruppen statt, um den Rechenaufwand bei der Merkmalsanpassung zu reduzieren. Bei der flächenbasierten Merkmalsbestimmung werden die Intensitätswerte der einzelnen Bildpunkte als Merkmal herangezogen.
2. (2) Das Ziel der Merkmalsanpassung besteht darin, eine paarweise Korrespondenz der zuvor bestimmten Merkmale des Objektbildes mit denen des Referenzbildes herzustellen. Die Herstellung einer Korrespondenz zwischen den Merkmalen in dem Objektbild und dem Referenzbild erfolgt durch Korrelations-Verfahren, z.B. eine Kreuzkorrelationsfunktion.
3. (3) Die anschließende Transformationsberechnung umfasst die Auswahl von Transformationsfunktionen und die Berechnung von Transformationsparametern, um das Objektbild mit dem Referenzbild oder das Referenzbild mit dem Objektbild zu überlagern. Durch drei Transformationsfunktionen lässt sich jeder Punkt des Referenzbildes auf den korrespondierenden Punkt des Objektbildes oder umgekehrt abbilden. Bei den Transformationsparametern handelt es sich um den Rotationswinkel, den Skalierungsfaktor und den Translationsparameter.
4. (4) In einem letzten Schritt werden die Transformationsfunktionen und die berechneten Transformationsparameter genutzt, um das Referenzbild und das Objektbild zu überlagern.

The image registration comprises the four main steps known per se, namely the feature determination, the feature adjustment, the transformation calculation and the transformation:

1. (1) Features can be determined by automatically extracting a number of features from the object image and the reference image. Regions, lines and points come into consideration as characteristics. Areas are suitable as regions that stand out clearly from the surrounding areas. Lines can be present on the images as contours or as borders of regions or as lines themselves. Points in the images can be given as intersections of lines or as corners of contours. In the method according to the invention, the features are preferably determined via the aforementioned feature groups, in order to reduce the computational effort involved in adjusting the features. In the area-based feature determination, the intensity values of the individual pixels are used as a feature.
2. (2) The goal of feature matching is to pair the previous correspondence to produce certain features of the object image with those of the reference image. Correspondence between the features in the object image and the reference image is established by means of correlation methods, for example a cross-correlation function.
3. (3) The subsequent transformation calculation includes the selection of transformation functions and the calculation of transformation parameters in order to overlay the object image with the reference image or the reference image with the object image. With three transformation functions, each point of the reference image can be mapped onto the corresponding point of the object image or vice versa. The transformation parameters are the rotation angle, the scaling factor and the translation parameter.
4. (4) In a last step, the transformation functions and the calculated transformation parameters are used to overlay the reference image and the object image.

Following the image registration, the superimposed images are evaluated by comparing the image information of each object image representing each test criterion with the image information of the reference image representing the respective test criterion.

The reference image is preferably overlaid with the object image. The image information of the reference image representing each test criterion then represents the target values and the image information of each object image representing each test criterion represents the actual values.

The complete absence of guides for the test objects in the test room in connection with the occasional moving of the test objects through the test room has the consequence that the digital image recording devices can record each test object without covers from different directions. In particular, with the method according to the invention it is possible for the first time that the images recorded simultaneously depict each test object on all sides. The test objects move through the test room, for example in free fall or along a trajectory to which each test object is brought before entering the test room.

The all-round representation of the test object makes it possible to examine almost all test criteria that are customary for plastic injection molded parts while each test object is moving through the test space. In particular, the method according to the invention can be used to examine not only area-based but also position-dependent test criteria, for example compliance with certain dimensions of the test objects. Changes in the position of the test objects moved through the test space one after the other are harmless for the examination of the test criteria. Due to the simultaneous recording of the images of each test object in the test room, the method according to the invention also tolerates changes in the position of the respective test object as it travels through the test room.

The method according to the invention primarily serves to examine rotationally symmetrical test objects mass-produced in the injection molding process or of test objects which have a rotationally symmetrical part. Such test objects can be described well by quadrics. In an advantageous embodiment of the invention, a method for 3D modeling of quadrics is used to reconstruct the object image. Due to the known basic shape of the parts, information is incorporated into the reconstruction that reduces the computational effort.

Methods for 3D modeling of quadrics are known per se (see Geoffrey Cross, Quadric reconstruction from Dual-Space Geometry, Robotics Research Group, Oxford University, 1998).

For a three-dimensional reconstruction of the object image, depth information is recorded for each image of the test object in a further advantageous embodiment of the method according to the invention. The available depth information allows the reconstruction of the test object as a point cloud and derived from it as a wireframe model. The depth information can be captured with 3D cameras, which allow the visual representation of distances of an entire scene. For example, known stereo cameras, triangulation methods, light field cameras or TOF cameras can be used to record the depth information (cf. 3D scene interpretations, Dipl.-Ing. Andreas Winter, TZI, University of Bremen, October 17, 2000).

To provide the three-dimensional reference image, images of a reference test object are first recorded in one embodiment of the invention. With regard to the test criteria, the reference test object corresponds to the target values for the test objects to be tested. The images of the reference test object can be taken under comparable conditions as the images of the test objects, ie during the unguided movement through the test room. This is followed by a three-dimensional reconstruction of the reference image from the recorded images of the reference test object. The 3D reconstruction can be a point cloud or another description of the surface of the reference test object in the three-dimensional space, so that the position and location in the three-dimensional space are assigned can be. The reference image can then be transferred in the three-dimensional space into a normal position, for example into the coordinate origin of the world coordinate system. However, due to the known position and location of the reference image in the three-dimensional space, this transformation is not absolutely necessary. Each reference test object and the test objects have at least one feature that enables the detection of position and location in space, such as a label.

The test objects are illuminated as homogeneously as possible in the test room, at least while the images of each test object are being taken. The most homogeneous lighting possible is of particular importance for examining area-related test criteria, such as the color of the test objects. The spatial distribution of the colors over the surface of the test object is reproduced in the object image, the position and location of which are known in the three-dimensional space. Consequently, the color values and their distribution over the surface of the object image can be compared with the color values and their distribution over the surface of the reference image. The method according to the invention therefore also allows the examination of a printed image.

For example, a hollow sphere construction concentrically surrounding the test space can be considered as a homogeneous light source. The homogeneous illumination can take place by means of the transmitted light process through a light-scattering material of the hollow sphere or through a surface which is diffusely reflecting on the inside of the hollow sphere.

If a targeted illumination of the test objects is advantageous for the examination of certain test criteria, it is proposed that the position of each test object in the test room is determined immediately before the pictures are taken and the lighting of each test object is controlled depending on the previously determined position, at least while the pictures are being taken becomes. If the test criterion is, for example, the detection of thin spots in a flat surface of the test object, depending on the recognized position of the test object, those light sources are activated before the pictures are taken whose radiation essentially strikes the flat surface to be tested. If the test criterion is, for example, the detection of unevenness in a surface, those light sources are activated whose radiation strikes the surface of the test object to be tested at the most acute angle possible.

The targeted illumination of the test object can take place, for example, via a multiplicity of individually activatable light sources which are arranged around the test space in such a way that the radiation from the individual light sources strikes the test object from different directions. The light sources can be attached to a lattice-like structure that surrounds the test space. All light sources are preferably attached to the lattice structure at a uniform distance from one another. The spaces between the struts of the lattice structure allow the test objects to freely enter and exit the test space. The digital image recording devices can either also be attached to the lattice structure or to separate holders.

The position of the test objects for controlling the lighting is preferably carried out using digital image recording devices. The same image recording devices that are also used to record the images for the examination of the test criteria can be used. Alternatively, an additional set of digital cameras can be provided for the location determination. The time between determining the position and taking the pictures of the test objects is so short that the position of the test object practically does not change.

The computing power and thus the time required for the three-dimensional reconstruction of the object image and the image registration can be reduced if the positional variation of the test objects in the test room is restricted. In one embodiment of the invention, the variance of the position of the test objects moved through the test space one after the other is limited by guide means arranged outside the test space. The guide means lead every test object with limited or no position variance to the test room. The remaining distance in which the test objects are moved without guidance until the image recording devices record the images of each test object in the test space is so short that only slight changes in position occur between the test objects moved successively through the test space without guidance. The guide means can end above the test room, for example, so that the test objects then move freely through the test room. Alternatively, the test objects are ejected from a guide arranged transversely to the test room, the parabola extending through the test room.

• 1a Einen Prüfraum sowie mehrere um den Prüfraum herum angeordnete digitale Stehbild-Kameras,
• 1b eine Gitterstruktur zur Halterung der digitalen Stehbild-Kameras nach 1a,
• 1c die Gitterstruktur nach 1b aus einem anderen Blickwinkel,
• 2a das Bereitstellen eines dreidimensionalen Referenzbildes in einem dreidimensionalen Raum,
• 2b die dreidimensionale Rekonstruktion eines Objektbildes in dem dreidimensionalen Raum,
• 2c das Auswerten des mit dem Referenzbild überlagerten Objektbildes.
• 2d ein Verfahren bei dem als Prüfkriterium der Durchmesser einer stirnseitigen Öffnung des Prüfobjektes erfasst werden soll.

The method according to the invention is explained in more detail below with the aid of schematic representations. Show it

• 1a A test room and several digital still cameras arranged around the test room,
• 1b a grid structure for holding the digital still cameras 1a .
• 1c the lattice structure after 1b from a different perspective,
• 2a the provision of a three-dimensional reference image in a three-dimensional space,
• 2 B the three-dimensional reconstruction of an object image in the three-dimensional space,
• 2c the evaluation of the object image superimposed with the reference image.
• 2d a method in which the diameter of a front opening of the test object is to be recorded as the test criterion.

1b shows a representation of a test arrangement ( 1 ) to carry out the method according to the invention. The test arrangement comprises a lattice structure ( 2 ), on whose struts ( 2.1 ) a variety of digital imaging devices ( 3 ) is attached. With the image acquisition devices ( 3 ), it is digital still cameras in the illustrated embodiment. All cameras are on the grid structure ( 2 ) attached so that their lenses lie on a common spherical surface within the lattice structure ( 2 ). The distance between all the lenses on the spherical surface preferably coincides. The spherical surface surrounds a test room ( 4 ) in the detection area ( 3.1 ) all image acquisition devices ( 3 ) lies (cf. 1a) ,

A tour ( 5 ) extends through an opening ( 2.2 ) in the lattice structure ( 2 ) through. The leadership ( 5 ) ends at a short distance from the test room ( 4 ). About the tour ( 5 ) are successively test objects ( 6 ) fed. For the test objects ( 6 ) it is in the illustrated embodiment, caps ( 6.1 ). The caps ( 6.1 ) with their floor area ( 6.2 ) along the management level in the direction of the test room ( 4 ) conveyed at a relatively high speed so that after leaving the tour ( 5 ) along a throwing parabola ( 7 ) guided through the test room ( 4 ) move. After leaving the test room ( 4 ) kick the caps ( 6.1 ) from the lattice structure ( 2 ) and can be sent for further processing.

How especially out 1a recognizable, the caps ( 6.1 ) sporadically through the test room ( 4 ) moved through so that the image recording devices ( 3 ) At the same time, pictures only of the one in the test room ( 4 ) record the test object.

The individual image recording devices ( 3 ) are connected to a data processing device, not shown. The captured images are transmitted to the digital data processing device and processed there. Based on 2 a - 2 c the processing of the images in the data processing unit is explained in more detail:

First, a three-dimensional reference image ( 8th ) of a reference test object ( 6 ) provided in a three-dimensional space. The three-dimensional space is defined by the origin coordinate system ( 9 ) with the coordinates x . y . z in 2a described (world coordinate system). The three-dimensional reference image ( 8th ) of the reference test object is provided, for example, in that the reference test object by means of the 1 Test arrangement described under the same conditions as the test objects to be subsequently examined in bulk ( 6 ) moved through and by the image recording devices ( 3 ) is recorded on all sides. From the recorded images of the reference test object, this becomes 2a shown three-dimensional reference image ( 8th ) reconstructed. The reference image ( 8th ) is in one with the original coordinate system ( 9 ) linked first relative coordinate system ( 9.1 ) with the coordinates x ' . y ' . z ' , The reference test object and thus also the reference image ( 8th ) points just below an opening at the front ( 6.3 ) a texture on the inner wall ( 6.4 ) on.

After providing the reference image ( 8th ) the regular inspection operation begins. Test objects matching the reference test object ( 6 ) are isolated by the test room ( 4 ) passed through. The image acquisition devices ( 3 ) take several pictures of the test object at the same time ( 6 ) from different directions ( 3.2 ) on. The test criterion is the correct placement and rendering of the texture ( 6.4 ) on each test object ( 6 ).

From the images of each test object taken at the same time ( 6 ) will that in 2 B displayed three-dimensional object image ( 10 ) reconstructed. The object picture ( 10 ) is in one with the original coordinate system ( 9 ) linked second relative coordinate system ( 9.2 ) with the coordinates x " . y " . z " ,

The object image ( 10 ) of each test object ( 6 ) in the second relative coordinate system ( 9.2 ) with the reference image ( 8th ) in the first relative coordinate system ( 9.1 ) superimposed. Suitable transformation functions for this are included in the image registration selected and the transformation parameters for the rotation and translation determined.

2c finally illustrates the evaluation of the superimposed images by comparing the test criterion (texture 6.4 ) representing image information of each object image ( 10 ) with which image information of the reference image representing the test criterion (texture) ( 8th ). How out 2c recognizable, the texture ( 6.4 ) of the object image ( 10 ) to the extent of the texture of the reference image ( 8th ) that the underlining under the lettering " x y z Inc. ”is missing. As part of the comparison of the images superimposed by transformation after image registration, the data processing unit recognizes the deviations in terms of texture.

The data processing unit can be configured in such a way that when it detects such a deviation as part of the examination of the test criterion, it activates a rejection unit that automatically rejects the test object identified as faulty from the flow of test objects following the examination.

2d shows a method in which the diameter at the front opening as a test criterion ( 6.3 ) of the test object ( 6 ) should be recorded. In this case, it is sufficient if the reference image ( 8th ) contains as image information a circle representing the test criterion, its position and location in the original coordinate system ( 9 ) is known. By overlaying the reference image (circle) ( 8th ) with the object image ( 10 ) the diameter of the circle with the actual diameter of the object image ( 10 ) of the test object ( 6 ) are compared. If there are deviations between the actual diameter and the diameter of the circle of the reference image that lie outside of a tolerance range, the object image ( 10 ) assigned test object ( 6 ) are automatically discarded. In the event of deviations lying within a tolerance range, the data processing unit can change parameters of the manufacturing process of the test objects to be examined in order to counteract the deviation. No. description 1. test set Second lattice structure 2.1. strut 2.2 opening Third Image recording device 3.1 detection range 3.2 directions 4th testing room 5th guide 6th UUT 6.1 cap 6.2 floor area 6.3 Front opening 6.4 texture 7th bombtrajectory 8th. reference image 9th Origin coordinate system 9.1 First relative coordinate system 9.2 Second relative coordinate system 10th object image

QUOTES INCLUDE IN THE DESCRIPTION

This list of documents listed by the applicant has been generated automatically and is only included for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent literature cited

• EP 2112502 A1 [0004]
• DE 102013102653 A1 [0005]
• WO 2014/147176 A1 [0007, 0008]

Claims (13)

Method for examining at least one test criterion of matching test objects (6) in a test room (4) comprising the steps: 1.1 occasional, unguided movement of the test objects (6) through the test room (4), the position of the test objects (6) moved successively through the test room (4) in the test room (4) being able to vary, 1.2 Aligning a plurality of digital image recording devices (3) to the test room (4) such that the images recorded simultaneously by the digital image recording devices (3) depict each test object (6) from different directions (3.2) as it passes through the test room (4) is moved, the images containing at least the image information of the test object (6) representing each test criterion, 1.3 three-dimensional reconstruction of an object image (10) from the recorded images, each object image (10) containing at least the image information of the test object (6) representing each test criterion and position and position of the object image (10) in a three-dimensional space (9) are known, 1.4 Providing a reference image (8) in the three-dimensional space (9), each reference image (8) containing at least the image information of the test object (6) representing each test criterion and the position and position of the reference image (8) in the three-dimensional space (9) known are, 1.5 performing an image registration for each object image (10) and the reference image (8) in order to overlay each object image (10) with the reference image (8) or the reference image (8) with each object image (10), 1.6 Evaluating the superimposed images (8, 10) by comparing the image information of each object image (10) representing each test criterion with the image information of the reference image (8) representing the respective test criterion. Procedure according to Claim 1 , characterized in that the digital image recording devices (3) are aligned with the test space (4) in such a way that the simultaneously recorded images depict each test object (6) on all sides. Procedure according to Claim 1 or 2 , characterized in that the test objects (6) have at least one rotationally symmetrical part and a method for 3D modeling of quadrics is used to reconstruct the object image (10). Procedure according to Claim 1 or 2 , characterized in that depth information is recorded for each recorded image of a test object (6) and the reconstruction of the object image is carried out using the depth information. Procedure according to one of the Claims 1 to 4 , characterized in that to provide the reference image (8) images of a reference test object are recorded and the images taken are reconstructed in the three-dimensional space (9). Procedure according to one of the Claims 1 to 5 , characterized in that performing the image registration for each object image (10) and the reference image (8) comprises the following steps 6.1 determining features in the object image (10) and the reference image (8), 6.2 establishing correspondence between the features in the object image (10) and the reference image (8), 6.3 the selection of transformation functions and calculation of transformation parameters in order to overlay the object image with the reference image or the reference image with the object image and 6.4 performing the transformation using the transformation functions and the transformation parameters, to overlay the object image (10) with the reference image (8) or the reference image (8) with the object image (10). Procedure according to one of the Claims 1 to 6 , characterized in that the reference image (8) is superimposed on the object image (10). Procedure according to one of the Claims 1 to 7 , characterized in that each test object (6) in the test room (4) is illuminated homogeneously at least while taking the images of the test object (6). Procedure according to one of the Claims 1 to 8th , characterized in that the position of each test object (6) in the test room (4) is determined before the pictures are taken and the illumination of each test object (6) is controlled at least during the taking of the pictures depending on the previously determined position. Procedure according to one of the Claims 1 to 9 , characterized in that the position is determined with the aid of digital image recording devices (3) aligned with the test space. Procedure according to one of the Claims 1 to 10 , characterized in that the variance of the position of the test objects (6) moved successively through the test space (4) is limited by guide means (5) arranged outside the test space. Procedure according to one of the Claims 1 to 11 , characterized in that test objects (6) for which the Evaluation of at least one test criterion has resulted in values outside a tolerance range. Procedure according to one of the Claims 1 to 12 , characterized in that at least one control parameter of a production machine for the test objects is automatically influenced, provided that the evaluation of a test criterion yields values that deviate from the target values, but are within a tolerance range.

Images