WO2021176386A1

WO2021176386A1 - Method and systems for the detection and classification of defects on surfaces

Info

Publication number: WO2021176386A1
Application number: PCT/IB2021/051798
Authority: WO
Inventors: Daniel RASPONE; Marco Piloni
Original assignee: Geico Spa
Priority date: 2020-03-06
Filing date: 2021-03-04
Publication date: 2021-09-10
Also published as: IT202000004783A1

Abstract

A method for the automatic detection and classification of defects on the surface (15) of an object (11) comprises a first step and a second step. The first step involves detection of the defects and during this step a first scanning head (13a) for taking images at a first resolution is moved along scanning paths on the surface of the object, defects on the surfaces are located in the images taken by the first scanning head (13a) and the position of the defects is identified. The second step involves classification of the defects and during this step a second scanning head (13a) for taking images at a second resolution, higher than the first resolution, is moved to the positions of the defects detected in the first step and the defects are classified on the basis of the images taken by the second scanning head (13b).

Description

"Method and systems for the detection and classification of defects on surfaces"

DESCRIPTION

The present invention relates to a method and to systems for the detection and classification (or cataloguing) of defects on complex surfaces, such as motor vehicle bodies. In particular, the defects to be detected may be for example appearance defects on painted surfaces.

The defects present on the painted surfaces often have a three-dimensional character, i.e. are not simply local variations in colour, but are bumps, missing material or in any case irregularities on the surface.

These defects are referred in the technical jargon as "appearance defects" since the user may visually notice them. Generally they have dimensions of at least 10-20 microns.

The spatial surfaces to be scanned are defined as being "complex" since they may be a combination of concave surfaces and convex surfaces, both also with variable radii of curvature and with the presence of cusps and curvilinear lines connecting lines the different parts which form the said surface.

For example, a motor vehicle body may be regarded as being a complex surface since it has the aforementioned characteristics.

The localization of the defects on a complex surface is a fundamental step of the industrial process since it allows the tracing and any correction of the appearance defects of the product which may be noticed by the end user and which often are regarded as being an indication of the quality of the entire product.

In the prior art, automatic systems have been proposed where the surface to be examined is scanned so as to obtain electronic images of portions of the surface. The images thus obtained are then processed by suitable computer programs which are specially designed to detect defects on the surfaces. The programs may also divide up the defects depending on their nature (for example, colour irregularities, depressions, bubbles, scratches, etc.) so as to indicate also the type of defect and if necessary propose possible remedies for the human operating staff or automatic repair systems.

For correct definition and cataloguing of a defect it is necessary for the scanning head, which produces the images to be processed by the detection programs, to have a relatively high spatial resolution. Owing to the spatial resolution required, the area recorded by the head must be however necessarily very small and the head is moved along the surface in order to examine it completely. Again owing to the relatively high resolution, and therefore the size of image data which is produced by the head and must be processed, the image processing speed (and generally the speed of displacement of the head on the surface) is relatively low.

As a result of all this, with the known systems the operation of scanning, detecting the defects and cataloguing them requires a considerable amount of time, in particular in the case of large-area surfaces, such as the bodywork surfaces of motor vehicles.

The object of the present invention is to provide a method and systems for detection and classification of defects on surfaces, including large-area surfaces, such as motor vehicle bodies, which are fast and reliable.

In view of this object, the idea which has occurred is to provide, according to the invention, a method for the automatic detection and classification of defects on the surface of an object, comprising: a first defect detection step during which a first scanning head for taking images at a first resolution is moved along scanning paths on the surface of the object, defects on the surfaces are located in the images taken by the first scanning head and the position of the defects is detected; and a second defect classification step during which a second scanning head for taking images at a second resolution, higher than the first resolution, is moved to the positions of the defects detected in the first step and the defects are classified on the basis of the images taken by the second scanning head.

Still according to the invention, the idea which has occurred is to provide a system for the automatic detection and classification of defects on the surface of an object according to the above method, comprising the first scanning head for taking images at the first resolution; the second scanning head for taking images at the second resolution; at least one motorized positioner for moving and positioning the heads on the surfaces; at least one electronic control and processing unit connected to the heads and to the positioner for receiving the images taken by the heads and moving the heads on the surfaces of the object.

In order to illustrate more clearly the innovative principles of the present invention and its advantages compared to the prior art, examples of embodiment applying these principles will be described below with the aid of the accompanying drawings. In the drawings:

- Figure 1 shows a schematic view of a first defect detection system according to the invention;

- Figure 2 shows a schematic side elevation view of a possible embodiment of two scanning heads in a system according to the invention;

- Figure 3 shows a schematic plan view of the scanning heads according to Figure 2;

- Figure 4 shows a schematic perspective view of a second embodiment of a defect detection system according to the invention; - Figure 5 shows a schematic perspective view of a third embodiment of a defect detection system according to the invention.

With reference to the figures, Figure 1 shows a system, denoted overall by 10, provided in accordance with the invention for detecting surface defects on an object 11 , for example the painted body of a motor vehicle. The surface defects may for example comprise bubbles, dimples, scratches, possible missing paint, inclusions in the paint of foreign bodies (threads, powder, etc.), and so on.

The system 10 may also comprise a known transportation system 12 which conveys in sequence the objects 11 to be analysed.

The transportation system may be a conveyor for example. In the case where the objects 11 are vehicle bodies, the transportation system 12 may be a known conveyor of the skid type where the bodies are mounted on known skids.

The system 10 comprises a first scanning head 13a and a second scanning head 13b which are moved so as to follow suitable paths 14 (one of which is shown by way of example in Figure 1) and suitably travel along the surfaces 15 of the object 11 on which the possible defects are to be detected.

In the embodiment shown in Figure 1 , the two heads 13a and 13b are both moved by a suitable motorized positioner 16 which may be advantageously a robot suitably programmed to move automatically the head along the said paths above the surfaces of the object 11. The robot 16 may be advantageously a known anthropomorphic robot arm, for example with six axes controllable in an interpolated manner, with both heads 13a and 13b mounted on the wrist 17 of the robot.

According to the principles of the invention, the head 13a is an image recording head which has a first resolution, and the head 13b is an image recording head which has a second resolution greater than the first head. In particular, the first resolution may be chosen so as to allow detection of the presence of a defect in a certain position on the surface, although not with a high degree of precision, and without precise classification of the defect.

The second resolution instead will be chosen relatively high so as allow classification of the defects.

The resolution may be divided into optical resolution due also to the lenses of the camera and resolution of the sensor (allowing the definition also of that which is actually acquired).

For example, in the case of the first resolution, the first optical resolution may be between 5 mm/pixel and 0.1 micron/pixel and preferably around 0.1 mm/pixel. The first resolution of the sensor may be advantageously between 0.25 and 70 megapixels and, preferably, around 5 megapixels.

For example, as regards the second resolution, the second optical resolution may be between 2 mm/pixel and 0.01 mm/pixel and, preferably, around 15 micron/pixel. The second resolution of the sensor may be preferably between 512 and 16,384 pixels/line and, preferably, around 7142 pixels/line.

Moreover, the first head 13a may be chosen, also depending on its smaller resolution, so as to have a recording area greater than the recording area of the second head. For example, the recording area of the first head may be between 5 and 100,000 times, advantageously around 1800 times (preferably around 1850 times), the recording area of the second head.

For recording of the images, each head will comprise a suitable known acquisition unit (for example a camera). The heads may also comprise known illumination systems for suitably illuminating the recorded area.

At least the second head may advantageously comprise a known stereoscopic camera so as to supply three-dimensional images to the system for processing and cataloguing the defects. This allows easier classification of the defects which result in variations in height of the surface, with a greater distinction for example between projecting bubbles and dimples in the painted surface.

A suitable electronic control unit 41 , for example a suitable processing system suitably programmed, as will be clarified below, controls the movement of the heads along the scanning paths, by means of the positioner 16, in a manner synchronized with the acquisition of the images by the heads, such that the images which in each case are recorded by the head may be added together to form broader images of the surfaces.

For example, preferably the heads 13a and 13b may be designed so as to form linear image acquisition units, namely acquisition units formed by light sensors located substantially along a single line, so as to obtain along their optical axis, or the viewing direction of the head, an image which is formed essentially by a recording segment 18a, 18b (namely an image with a very limited amplitude in the direction of movement of the head along the path). Scanning of the surfaces 15 of the object 11 will therefore be performed in this case by recording segments 18 which are parallel to each other and transverse to the movement path of the head. The parallel recording segments 18a or 18b, which in each case are acquired by the scanning head 13a or 13b owing to a relative movement of head and surface to be scanned, transverse to the recording segments, thus compose the images of the surface as the sum of the single recording segments. Suitable linear cameras to be inserted in heads are for example known per se.

Figures 2 and 3 shows by way of example two heads 13a and 13b mounted on a single support 19 so as to be moved simultaneously by the positioner 16. As can be clearly seen in Figure 2, the heads may be mounted on the support 19 with divergent viewing directions 29a and 29b. During use, the heads may therefore be directed alternately with their viewing direction towards and perpendicular to the surface 15 so as to follow the desired paths 14 and be used alternately to scan these surfaces and obtain the desired images thereof.

By arranging the heads with divergent viewing directions, as shown in the figures, it has been found that it is possible to reduce the interference between the head not in use and the movement of the other head in use. In other words, with the divergent arrangement, the head not in use will be further from the surfaces 15 than the head in use and there will be less possibility of interference of the head not in use with the surfaces 15 subject to scanning with the other head. Figure 3 also shows the possible different recording segments 18a and 18b of the two heads 13a and 13b realized by means of linear image acquisition units.

The control unit 41 may also perform processing of the images using methods (known per se and therefore not shown or described in detail here since they may be easily imagined by the person skilled in the art) which will allow automatic identification of the defects on the surfaces 15 in the recorded images and if necessary classification thereof, with the degree of precision permitted by the resolution of the images produced by the heads.

In particular, during a first defect identification step, the system 41 will cause the movement of the positioner 16 so as to perform a first scan of the surfaces 15 by means of the first head 13a. The unit 41 will thus identify (using the image processing process known per se) the existence of any defects on the visible surfaces in the images produced by the head 13a. For each defect identified, the unit 41 will detect the spatial coordinates of the defect on the surface so as to be able to find it again. Once the first step of scanning the surfaces with the first head has been completed, the unit 41 will carry out a second scanning step for classification of the defects. During this second scanning step, the unit 41 will use the second scanning head 13b, operating it so that it directly scans the surfaces at the coordinates of the defects detected during the first scanning step.

Since the head 13a has a relatively low resolution and, preferably, a relatively broader recording range, the first step involving the scanning of the entire surfaces 15 will be faster than if it were carried out with a head having the resolution characteristics - and where applicable recording range - of the second head. On the other hand, the second head must carry out the scanning only of the zones of the surfaces with the defects already detected during the first step. In other words, the second head, which is slower owing to the higher resolution and where applicable the smaller recording range, must examine only small zones of the surfaces where a defect has already been detected so as to be to classify it (using the classification methods known per se) owing to its higher resolution. If required, the second head, owing to its higher resolution, may also refine the detection of the position of the defect based on the defect coordinates obtained by means of the first head.

The entire defect detection and classification procedure will therefore be optimized and much more rapid than that which would be possible using the systems of the prior art.

The system according to the invention therefore rapidly provides both the precise position of the defects and their classification. This information may then be passed to known manual, automatic or semi-automatic defect processing systems, as known in the sector.

Figure 4 shows a first variation of embodiment of a system for the detection and classification of defects according to the invention. In this variant, denoted overall by 110, there are again the two heads 13a and 13b as described above, but these heads are each supported by associated positioners, respectively indicated by 16a and 16b (which may be essentially of the same type as the positioner 16 of the preceding embodiment).

In this way, the control unit 41 may move in a substantially simultaneous and independent manner the two heads over the surface 15. If required, the two scanning steps, for detecting the defects and classifying the defects, respectively, may thus be carried in a parallel or semi-parallel manner. For example, as the control unit 41 detects defects by means of the first head, the second head may be operated so as to classify the defects, while the first head detects further defects, and so on. This speeds up even more the overall defect detection and classification operations.

Figure 5 shows a second variation of embodiment of a system for the detection and classification of defects according to the invention. In this variant, denoted overall by 210, there are again the two heads 13a and 13b as described above, but these heads are each supported by associated positioners, respectively indicated by 16a and 16b, and located in different scanning stations 50 and 51 into which the objects 11 are conveyed in sequence. The first station 50 therefore constitutes a station for automatic detection of the defects, while the second station 51 constitutes a station for automatic classification of the defects detected in the first station.

In this way, the control unit 41 may move in a substantially simultaneous and independent manner the two heads over the surfaces 15 also on different objects in the two stations.

With a plurality of objects 11 which are to be examined and are moved in sequence along the transportation line 12, the defect detection step performed in the station 50 on a following object may thus be carried out in parallel with the defect classification step performed in the station 51 on a preceding object which has already passed through the station 50 for identification of the position of the defects. The total time of the detection and classification cycle carried out on a continuous sequence of objects (such as motor vehicle bodies in a processing and painting plant) will therefore be substantially shorter.

At this point it is clear how the objects of the invention have been achieved.

Owing to the principles of the invention, the localization and cataloguing of the defects for possible subsequent removal thereof is rapid, precise and efficient. Obviously the description given above of embodiments applying the innovative principles of the present invention is provided by way of example of these innovative principles and must therefore not be regarded as limiting the scope of the rights claimed herein.

For example, the heads may have a form and/or proportions different from those shown, depending for example on the nature of the cameras and the illumination devices used to form them and/or the configuration and area of the surfaces to be scanned. For example, the head may have a square form in plan view, instead of a rectangular one, and may also have dimensions smaller than those shown in the drawings, so as to inspect for example surfaces which are smaller, with a greater concavity or within confined spaces. The recording segment of the camera (and therefore of the head) may also be more or less long depending on the extension of the surfaces and the desired scanning speed. The two heads may also be mounted inside the same housing.

As may be now easily imagined by the person skilled in the art it is also possible to provide heads with several cameras and/or several illumination devices so as to obtain recording areas or segments which are bigger or smaller depending on the specific requirements.

Moreover, a positioner for the head different from that shown and described by way of example may be used. In particular a Cartesian positioner or a gantry could be used, depending also on the geometrical form of the object to be scanned.

The control unit 41 may also be divided up or comprise a defect detection unit and a defect classification unit, with the former which passes on the defect coordinates to the latter, as may be now easily imagined by the person skilled in the art on the basis of the description provided above. In the system with separate stations, each station may comprise an associated control unit with the processing of the images provided by the corresponding head and the movement of this head. The control unit of the defect detection station may thus simply pass on the defect coordinates to the control unit of the classification station.

Obviously, in any case the paths 14 followed by the two heads may be different from each other.

Claims

1 . A method for the automatic detection and classification of defects on the surface (15) of an object (11), comprising:

- a first defect detection step during which a first scanning head (13a) for taking images at a first resolution is moved along scanning paths on the surface of the object, defects on the surfaces are located in the images taken by the first scanning head (13a) and the position of the defects is detected; and

- a second defect classification step during which a second scanning head (13b) for taking images at a second resolution, higher than the first resolution, is moved to the positions of the defects detected in the first step and the defects are classified on the basis of the images taken by the second scanning head (13b).

2. Method according to claim 1 , characterized in that the first resolution comprises a first optical resolution of between 5 mm/pixel and 0.1 micron/pixel and, preferably, around 0.1 mm/pixel and/or a first sensor resolution of between 0.25 and 70 megapixels and, preferably, around 5 megapixels.

3. Method according to claim 1 , characterized in that the second resolution comprises a second optical resolution of between 2 mm/pixel and 0.01 micron/pixel and, preferably, around 15 micron/pixel and/or a second sensor resolution preferably of between 512 and 16,384 pixels/line and, preferably, around 7142 pixels/line.

4. Method according to claim 1 , characterized in that the first head (13a) has a recording area greater than the recording area of the second head (13b) and, preferably, between 5 and 100,000 times, advantageously around 1800 times and preferably around 1850 times, the recording area of the second head.

5. Method according to claim 1 , characterized in that at least the second head comprises a stereoscopic camera.

6. Method according to claim 1 , characterized in that the heads (13a and 13b) form linear image acquisition units for obtaining, in a viewing direction of the head, an associated recording segment (18a, 18b) and the image of the surface (15) produced by a head is obtained as the sum of its parallel recording segments (18a or 18b), which are gradually acquired with the scanning head by means of a relative movement of head and surface to be scanned, which is transverse to the extension of the recording segment (18a, 18b).

7. Method according to claim 1 , characterized in that the objects (11) are vehicle bodies.

8. System for the automatic detection and classification of defects on the surface of an object according to the method of any one of the preceding claims, comprising the first scanning head (13a) for taking images at the first resolution; the second scanning head (13b) for taking images at the second resolution; at least one motorized positioner (16, 16a, 16b) for moving and positioning the heads on the surfaces (15); at least one electronic control and processing unit (41) connected to the heads (13a, 13b) and to the positioner (16, 16a, 16b). for receiving the images taken by the heads and moving the heads on the surfaces (15) of the object (11).

9. System according to claim 8, characterized in that the heads (13a, 13b) are mounted on the same positioner (16) so as to be moved together.

10. System according to claim 9, characterized in that the heads (13a, 13b) are mounted on the same positioner (16) with divergent viewing directions (29a, 29b).

11. System according to claim 8, characterized in that the heads (13a, 13b) are each mounted on an associated positioner (16a, 16b) so as to be movable independently of each other.

12. System according to claim 8, characterized in that the first head (13a) with its positioner (16a) is in a first defect detection station (50) and the second head (13b) with its positioner (16b) is in a second defect classification station (50).

13. System according to claim 12, characterized in that between the first station (50) and the second station (51) there is a line (12) for the sequential transportation of the objects (11).

14. System according to claim 8, characterized in that the positioner (16, 16a, 16b) is a programmed robot.

15. System according to claim 14, characterized in that the robot (16, 16a, 16b) is an anthropomorphic robot arm.

16. System according to claim 7, characterized in that it is intended for scanning objects (11) in the form of vehicle bodies.