JPH05223522A - Method and apparatus for detecting mending position of vehicle painting - Google Patents

Abstract

PURPOSE:To obtain a mending-position detection method and its apparatus wherein an irregularity in the brightness of a sensed image due to an irregularity in an illumination operation is removed, a detection error can be prevented and a mending detection can be performed surely and stably. CONSTITUTION:The image of the painting face of a vehicle is sensed by using a color camera 4. Three primary-color video signals VR, VG, VB which have been sensed by using the color camera 4 are taken into a brightness correction processor 24; they are first normalization-converted individually into a reference scale. Then, correction coefficients are found regarding a color level and a white level on the basis of reference values; the brightness deviation of a normalization signal is corrected regarding the correction coefficients; correction video signals CVR, CVG, CVB are obtained. The correction video signals CVR, CVG, CVB are input to an image processor 25; they are analyzed and processed; a color mark which is attached to a correction position is detected, i.e., an image is recognized.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for automatically photographing a painted surface of a vehicle with a color camera to automatically detect a color mark attached to a repair position by image recognition. The present invention relates to a vehicle painting repair position detection method and apparatus for correcting the lightness deviation of the obtained three primary color signals.

[0002]

2. Description of the Related Art It is well known that a coating defect on a coating surface of an automobile body is corrected by water-polishing, and as a repairing technique for vehicle coating, for example, Japanese Patent Laid-Open No. 58-64157 (B05).
C 9/08) and the like, the worker visually inspects the conveyed car body, and inputs the data on the spot and the state of the coating defect found by the instruction device. There is known an automatic water polishing method and device for automatically performing water polishing by operating a water polishing device such as a robot on the downstream side. However, in this technique, since the operator has to input data one by one each time the operator finds a defective portion, an input error is likely to occur, and input is troublesome and complicated.

Therefore, by developing the above-mentioned automated repair technology, a defective portion found by visual inspection of the painted surface of the automobile body by an operator on the upstream side of the transfer line is marked with a mark for each defective grade, and the downstream side thereof. The detection robot equipped with a color camera detects the type of mark and the position of the mark on the painted surface, and a polishing robot is arranged further downstream, and the information of the detection result of the detection robot is used for the polishing robot. It is considered that the robot for polishing is used to correct and polish a defective coating portion.

As the detection robot, a color camera at the tip of the hand picks up an image of the surface to be inspected, that is, the coating surface of the automobile body, and automatically detects the mark attached to the defective portion by image recognition. The hand is moved along the automobile body to image the surface to be inspected for each predetermined section, and the images are sequentially captured to perform image recognition, that is, detection of the repair position.

[0005]

However, at this time, due to the displacement of the automobile body on the transfer line,
The relative relationship between the illuminating device that illuminates the car body as the surface to be inspected and the car body is misaligned, and it is not possible to uniformly illuminate the surface to be inspected under the preset conditions, resulting in uneven illumination. There was the inconvenience that the brightness of the captured image was uneven.

Moreover, since the automobile body as the surface to be inspected has a complicated shape consisting of many curved surfaces, it is difficult to illuminate it evenly, and therefore the displacement of the automobile body is slight. However, the illumination setting is greatly affected, and the image captured by the color camera has uneven brightness, resulting in a detection error in image recognition, which is not preferable because the mark cannot be detected stably.

The present invention has been made in view of the above circumstances, and an object thereof is to eliminate unevenness in brightness of a picked-up image due to uneven illumination so as to prevent a detection error at the time of image recognition, and to ensure reliability and stability of mark detection. It is an object of the present invention to provide a method and apparatus for detecting a repair position of a vehicle painting that can improve the property.

[0008]

In order to achieve the above-mentioned object, the present invention is to automatically image a painted surface of a vehicle with a color camera and automatically recognize a color mark attached to a repair position of the painted surface by image recognition. In the vehicle paint repair position detection method to be detected, for each of the three primary color signals taken by the color camera, normalization conversion is performed on the reference scale, and the correction coefficient is obtained based on the reference values for the color level and the white level. It is characterized in that the correction coefficient corrects the brightness deviation of the normalized signal.

Further, the present invention relates to a vehicle painting repair position detecting device for picking up an image of a painting surface of a vehicle with a color camera and automatically detecting a color mark attached to the repairing position of the painting surface by image recognition. , For each of the three primary color signals taken by the color camera, a normalization conversion means for performing a normalization conversion on a reference scale, and a correction coefficient for the color level and the white level are obtained based on the reference values, and the correction coefficients are normalized. And a lightness correction means for correcting the lightness deviation of the signal.

[0010]

The operation of the present invention will be described. Each of the three primary color signals captured by a color camera for image recognition is first normalized and converted to a reference scale, and then corrected with respect to the color level and the white level based on the reference values. Coefficients are obtained, and the brightness deviation of the normalized signal is corrected by these correction coefficients.
Therefore, even if illumination unevenness occurs on the surface to be imaged, the brightness of the captured image can be flatly and evenly adjusted to the reference without deviation, and the unevenness of brightness can be removed.

[0011]

Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a plan view of an automobile manufacturing line showing a preferred embodiment of the present invention. The figure shows a paint repair line L for inspecting and correcting the painted surface of an automobile body conveyed from upstream.

The coating repair line L is provided with an inspection marking station L1, a mark detection station L2, a repair station L3 and a water washing station L4 in this order from the upstream side.

At the inspection marking station L1, inspectors P1 and P2 are arranged facing the left and right of the vehicle body B, respectively. Here, the coated surface of the vehicle body B conveyed to the station is visually inspected by the inspectors P1 and P2. When a coating defect such as a pinhole or foreign matter is found on the coated surface, the detected defect is detected. The inspectors P1 and P2 make marks M on the parts by using color markers corresponding to the polishing grades corresponding to the parts. In this embodiment, the polishing grades to be repaired on the defective coating portion are three ranks of heavy polishing, medium polishing and light polishing, and a predetermined color is assigned to each rank.

At the upstream side of the mark detecting station L2, a light emitting section 1 and a light receiving section 2 for detecting the type of the vehicle body B are provided.
A plurality of vehicle type detection sensors 3 paired with and are provided facing each other on the left and right of the transport line, and mark detection robots R1, R2, R3 for detecting the mark M attached to the vehicle body B are provided in the downstream side portion. Are arranged so as to face the right and left sides and the upper surface of the vehicle body B, respectively. Each mark detection robot R1, R
2 and R3, a color camera 4 is provided at the tip of each hand, and the hand is moved along the vehicle body B to detect the position and color of the mark M attached to the vehicle body B. The painted surface of B is imaged for each predetermined section.

A washing shower 5 is attached to the repair station L3.
Are provided around the vehicle body B, and the polishing robots R4, R5, R6 are provided so as to face the left and right sides and the upper surface of the vehicle body B, respectively. Each polishing robot R4, R
The polishing tools 6 are respectively provided at the tips of the hands 5 and R6, and the polishing tool 6 is guided to the position of the mark M to perform the correction polishing based on the data from the mark detection station L2.

A washing shower 7 is installed at the washing station L4.
Are provided around the vehicle body B, and the washing brushes W1, W2, W3 are provided so as to face the right and left sides and the upper surface of the vehicle body B, respectively, so that the vehicle body B after the repair and polishing is washed with water. There is.

A control panel 8 for operation control is provided in each of the stations L2 to L4 except the inspection marking station L1.
2 are provided, as shown in the control system block diagram of FIG. 2, the vehicle type detection sensor 3 is a vehicle type detection panel 8 and the mark detection robots R1, R2, R3 are mark detection robot control panels 9, 10, respectively. 11, the color cameras 4, ... Are respectively marked by the mark detection panels 12, 13, 14, and the polishing robots R4, R5, R6 are respectively designated by the polishing robot control panels 15, 16, 17, and the polishing tools 6 ,.
With each polishing tool control panel 18, 19, 20
The washing shower 7, ... And the washing brushes W1, W2, W3 are controlled by the washing control board 21. The control panels 8 to 20 are connected to each other by a bus line 22 and are connected to a production information net terminal device 23 that receives production information from the outside.

The mark detection boards 12, 13, 14 all have the same structure and, as shown in FIG. 3, are provided with a brightness correction processor 24 and an image processor 25. The image of the painted surface of the vehicle body B taken by the color camera 4 is taken into the brightness correction processor 24 as so-called three-primary-color video signals V _R , V _G , and V _B , and the mark detection boards 12, 13, 14 are used. As the brightness correction processor 24
In the corrected video signal C
V _R , CV _G , and CV _B are sent to the image processor 25 for analysis processing, and detection of the position of the mark M indicating the repair position and color indicating the polishing grade, that is, image recognition is performed. There is.

The operation of the brightness correction processor 24 will be described below with reference to the block diagram of FIG. The brightness correction processor 24 is provided with a plurality of arithmetic units 26 to 35, and constitutes an arithmetic logic for correcting the brightness deviation.

The video signals V _R , V _G and V _B input from the video camera 4 are first normalized by the normalizing operation units 26, 27 and 2.
8 and each of them is subjected to subtraction and removal of zero biases B _R , B _G , and B _B from their video signals and multiplied by scale factors G _R , G _G , and G _B to obtain normalized signals S _R and S.
Normalized as _G , S _B.

Next, they normalized signal is each incorporated into the color level correction calculating unit 29, 30, 31, the reference value SB _R for the color level, SB _G, SB _B is normalized signal S
It is divided by _R , S _G , and S _B , and is raised to a power of a predetermined value α to obtain the ratio to the reference color level, that is, the color level correction coefficient. Then, the color level correction coefficient is multiplied to the normalized signals S _R , S _G , and S _B to first correct the color level. Which is the output normalized signal S _R of the normalization arithmetic unit 26, 27, 28, S _G, S _B is incorporated into the white level correction coefficient calculating unit 32, three reference values SB _R for the color level, SB _G, SB The sum of the squares of each of the _Bs , that is, the reference values SB _R , S
The vector sum of B _G and S _{B B} is converted into the normalized signals S _R , S _G , and
The white level correction coefficient K2 is obtained by dividing by the vector sum of S _B to obtain the ratio to the reference white level, and taking the power of a predetermined value β. Then, the color level correction signal SK _R ,
SK _G and SK _B are brightness deviation correction calculation units 33, 34 and 35.
Incorporated into, their correction coefficient white level K2 white level correction by the multiplied made in the correction video signal CV _R, CV _G, is CV _B obtained.

That is, as shown in the flow chart of FIG. 5, in the brightness correction processor 24, each of the three primary color video signals V _R , V _G , and V _B taken in from the color camera 4 is first normalized and converted onto the reference scale. normalized signal S _R by, S _G, and S _B (T1), a reference value with respect to the next color and white levels SB _R, SB _G, SB _B
Each of the correction coefficients K _R , K _G , K _B , and K2 is calculated based on (T2, T3), and the normalized signal S _R ,
S _G, and to obtain and correct the brightness deviation is S _B have corrected video signal CV _R, CV _G, the CV _B (T
4).

As described above, since the lightness deviation is corrected for each of the three primary color video signals V _R , V _G , and V _B taken by the color camera 4 for image recognition, the vehicle body B is corrected.
Even if there is illumination unevenness on the painted surface of, that is, the surface to be imaged,
As a captured image, the lightness can be adjusted to be flat and uniform without any deviation. For example, FIG. 6 is a graph showing the output level of the image signal with respect to the aperture value of the camera lens. The two characteristics V _i and CV _i shown in FIG. 6 are obtained by increasing / decreasing the aperture value in order to realize uneven illumination. It has gained. When the brightness deviation is not corrected, the output level of the image signal naturally decreases as the camera lens is narrowed down as indicated by the broken line characteristic V _i. However, according to the present invention, the solid line characteristic CV _As indicated by _i , the output level can be flat and uniform regardless of the aperture value. Since such a correction is performed for each part of the captured image, it is possible to remove unevenness in brightness. As a result, it is possible to prevent detection errors during image recognition, improve the certainty and stability of mark detection, and reliably and stably detect the repair position.

[0024]

As described in detail above, according to the vehicle painting repair position detecting method and apparatus according to the present invention, the lightness deviation is corrected for each of the three primary color signals taken by the color camera for image recognition. Therefore, even if illumination unevenness occurs on the surface to be imaged, the brightness of the captured image can be flatly and evenly adjusted to the reference without deviation, and the unevenness of brightness can be removed. As a result, it is possible to prevent a detection error in image recognition and improve the certainty and stability of mark detection.

[Brief description of drawings]

FIG. 1 is a plan view of an automobile manufacturing line showing an embodiment of the present invention.

FIG. 2 is a block diagram showing a control system of a paint repair line.

FIG. 3 is a block diagram of a mark detection board.

FIG. 4 is a block diagram showing a calculation logic for brightness deviation correction.

5 is a flowchart illustrating the operation of the arithmetic logic shown in FIG.

FIG. 6 is a graph illustrating an output level characteristic of an image signal with respect to an aperture value of a camera lens.

[Explanation of symbols]

4 Color camera 12, 13, 14 Repair position detection device (mark detection board) 26, 27, 28 Normalization conversion means (normalization calculation section) 29, 30, 31 Brightness correction means (color level correction calculation section) 32 Brightness correction means (white level correction coefficient calculating unit) 33, 34, 35 brightness correction means (brightness deviation correction calculation unit) B vehicle (vehicle body) M color mark _{(mark) V R, V G, V} B three primary color signals (video signals) SB _R , SB _G , SB _B reference values K _R , K _G , K _B , K2 correction coefficients S _R , S _G , S _B normalized signal

Claims (2)

[Claims] 1. An image of a painted surface of a vehicle is imaged by a color camera,
A method for detecting the repair position of a vehicle painting that automatically detects the color mark attached to the repair position on the painted surface by image recognition, wherein each of the three primary color signals taken by the color camera is normalized and converted to a reference scale. Then, a correction coefficient for the color level and the white level is obtained based on the reference values, and the lightness deviation of the normalized signal is corrected with these correction coefficients. 2. An image of a painted surface of a vehicle is picked up by a color camera,
In a vehicle paint repair position detection device that automatically detects the color mark attached to the repair position on the painted surface by image recognition, for each of the three primary color signals taken by the color camera, normalization conversion to a reference scale is performed. And a brightness correction unit that calculates a correction coefficient for the color level and the white level based on the reference value and corrects the brightness deviation of the normalized signal with these correction coefficients. Vehicle paint repair position detection device.