JP4820971B2 - Surface inspection method - Google Patents

Description

The present invention relates to a surface inspection method for detecting minute irregularities in a pressed steel sheet.

  Conventionally, in a pressing process of a steel plate of an automobile or the like, minute unevenness may occur in the pressed steel plate due to the adhered zinc powder or iron powder. The unevenness generated at this stage is further emphasized by performing a subsequent coating process or the like. Therefore, at the stage of press working, a person detects a problem portion while visually observing, tentacles, and grinding stones.

  As a visual inspection, for example, the light of a fluorescent lamp is projected on a component, and the surface is viewed like a mirror, the line of sight is moved up and down, and the surface unevenness state is observed while changing the angle.

  The inspection by the tentacles refers to, for example, wearing gloves and touching the steel sheet along the flow of the vehicle state to see if the surface is uneven.

  As an inspection by grinding with a grindstone, for example, the grindstone is held lightly, the grindstone is applied straight along the flow of the vehicle state, and the unevenness of the surface is seen.

  However, it is difficult to say that the current manual detection work is sufficiently accurate because it is affected by the fatigue level and physical condition of the worker and the load state of the work. Furthermore, there is also a problem that the pressed steel sheet is scratched when polishing with a grindstone.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a surface inspection method capable of detecting minute irregularities in a pressed steel sheet without manual intervention.

In order to achieve the above object, the surface inspection method of the present invention irradiates a pressed steel sheet with light including infrared light, and detects reflected light including infrared light and visible light from the steel sheet, and A noise removal step for removing high-frequency noise in the image obtained in this imaging step, and a pixel having an approximate RGB value in the image from which high-frequency noise has been removed by the noise removal step are counted, and an appearance frequency value of each RGB value is obtained. Based on the color histogram process, and the appearance frequency value of each RGB value obtained by the color histogram process, a clustering process that classifies pixels as pixels having a high appearance frequency as normal parts, and pixels having a low appearance frequency as uneven defect areas, Identifying the concave and convex defect areas classified in the clustering process and obtaining the area, and the concave calculated in the labeling process Characterized in that it comprises a, and the area threshold value comparison processing step of recognizing the irregular defect performs a comparison with the threshold value for the area of the defect area.

In the above configuration, preferably, in the imaging step, a light source is arranged at an arbitrary angle of 60 ° ± 15 ° with respect to the surface of the pressed steel plate, and light including infrared light is irradiated toward the steel plate, An image pickup unit is arranged at an arbitrary angle of 60 ° ± 15 ° with respect to the surface of the steel plate pressed at a position facing to detect the reflected light from the steel plate.

According to the present invention , it is possible to detect an uneven defect with high accuracy by eliminating the influence of noise.
Since the color histogram process and the clustering process are performed, the normal area and the uneven defect area can be easily separated.

In carrying out the present invention, light including infrared light emitted from a light source is reflected by a steel plate and photographed by an imaging unit, and an image processing device detects irregularities from the captured infrared image, and this image The system may be configured such that information calculated by the processing device is output to the output device, which can reduce the number of inspection workers and reduce the burden on the inspection workers. At this time, since light including infrared light is used, even fine unevenness can be detected.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.
In the present embodiment, as an example, a method for automatically detecting a minute unevenness defect generated in a pressing process of a vehicle body manufacturing process in a short time without contact will be described. As the process, the flow from acquiring an image from the imaging unit, obtaining the feature points of the unevenness by image processing using a color histogram, and providing the inspection worker with the presence / absence of the unevenness detection and the occurrence position information will be described. .

  As shown in FIG. 1, the surface inspection system 1 of this embodiment includes a light 21 as a light source that irradiates light including infrared light toward a pressed steel sheet S, and red reflected by the pressed steel sheet S. As an image capturing unit that detects light including external light, an image processing device 3 that detects unevenness from an infrared image captured by the camera 22, and an output device that outputs a result obtained by the image processing device 3 Display device 4 and a recording device 5 for recording an image of the uneven portion. Here, the camera 22 removes the infrared cut filter in a normal digital camera, and can capture the wavelength in the infrared region. According to such a camera 22, among the color elements R, G, and B, green and blue of visible light can be detected by G and B, and infrared light and visible light can be detected by R.

[Imaging process]
The light 21 irradiates light including infrared light toward the steel sheet S at an angle of 60 degrees with respect to the surface of the pressed steel sheet S. The camera 22 is installed at an angle of 60 degrees with respect to the surface of the pressed steel sheet S at a position facing the light 21, and acquires a reflection component including infrared light and visible light from the steel sheet S. Here, among the color elements R, G, and B of the camera 22, G and B detect green and blue of visible light. On the other hand, R detects infrared light and visible light. Images are acquired as still images, but the number of acquired images may be small. The acquired image may have a resolution of about 250 dpi (dot per inch). By using light including infrared light as a light source, an image in which the unevenness is emphasized more than when photographing with a general camera is obtained due to the spectral characteristics of the pressed steel sheet. Thereafter, the acquired image is sent to the image processing apparatus.

  Next, an image processing process performed in the image processing apparatus will be described with reference to FIG.

[Noise removal process]
The noise removal step is a step of removing high-frequency noise in the image such as illumination unevenness, lens dust, and electronic noise entering the image with a filter, specifically a moving average filter. Actually, using a 3 × 3 pixel filter, step S1 for checking the presence or absence of noise and step S2 for removing the noise when it is determined that there is noise are repeated a plurality of times until the noise disappears.

[Abnormal area extraction process]
The abnormal region extraction step is a step of separating a normal region and an abnormal region in a photographed image, and includes step S3 for performing color histogram processing and step S4 for performing clustering processing. Color histogram processing refers to checking the appearance frequency of each RGB value by counting pixels having approximate RGB values in the image. Here, R of the RGB values includes infrared light and red of visible light. The clustering process refers to classifying pixels according to RGB values digitized by the color histogram process. The irregularity defect region has a smaller area than the normal region and has an RGB value different from that of the normal region, so that the appearance frequency of the RGB value of the irregularity portion is reduced. Accordingly, two types of pixels are clustered from the RGB values, and pixels having a high appearance frequency are defined as normal portions and pixels having a low appearance frequency are defined as an uneven defect region.

[Labeling process]
The labeling step is a step of identifying the concavo-convex defect region and obtaining the area thereof, and includes step S5 for performing numbering and step S6 for calculating the area of each region. In step S5, among the pixels subjected to the clustering process in the abnormal area extraction step, the areas determined as the uneven defect areas are numbered. In step S6, the area of each numbered region is obtained.

[Area threshold value comparison process]
The area threshold comparison processing step S7 is a step of recognizing the concavo-convex defect by comparing the area of the concavo-convex defect calculated in the labeling step with a reference area as a threshold. Those larger than the reference area are finally determined as irregularities. By this processing, noise that cannot be removed by noise removal is distinguished from uneven defects.

  Data calculated through the above steps is transmitted from the image processing device 3 to the display device 4 and the recording device 5. The presence / absence of the uneven defect and the portion finally determined as the uneven defect are displayed on the display device 4 and stored in the recording device 5. By inspecting the display device 4, the inspection worker P can know the presence or absence of uneven defects in the steel sheet S and the part finally determined as an uneven defect. Further, the data about the past steel plate can be known by referring to the recording device 5.

According to the surface inspection system of this embodiment, since it can be used as an alternative means of visual inspection in the press range, the number of inspection workers can be reduced and the burden on the inspection workers can be reduced. At this time, by using light including infrared light having a wavelength longer than that of visible light, it is possible to detect minute irregularities generated in the pressing process of the vehicle body manufacturing process . In addition, the product can be corrected at an early stage, thereby eliminating waste and reducing costs. Furthermore, since there is no fear of damaging the product as compared with the conventional grinding wheel polishing method, it is possible to prevent deterioration in quality. Moreover, since the surface inspection system of this embodiment is comprised from a simple image processing apparatus, while being able to shorten inspection time, the whole press steel plate can be test | inspected with the same precision, and a whole surface guarantee can be performed.

  As described above, the surface inspection system and the surface inspection method of the present invention automatically detect uneven defects without human intervention, and can be implemented in various forms without departing from the gist thereof. it can. For example, the light and the camera may be positioned at about 60 ° ± 15 ° with respect to the surface of the steel plate. Further, the components of the surface inspection system do not necessarily have to be separate, and may be integrated into the surface inspection apparatus.

It is a figure which shows the surface inspection system of this embodiment. It is a flowchart which shows the image processing process performed in the image processing apparatus of the surface inspection system of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Surface inspection system 3 Image processing apparatus 4 Display apparatus 5 Recording apparatus 21 Light 22 Camera P Inspection worker S Steel plate

Claims (2)

An imaging step of irradiating light including infrared light, detects reflected light including infrared light and visible light from the steel plate pressed steel plate,
A noise removing step for removing high-frequency noise in the image obtained in this imaging step;
A color histogram step for counting pixels having approximate RGB values in an image from which high-frequency noise has been removed by the noise removal step, and obtaining an appearance frequency value of each RGB value;
Based on the appearance frequency value of each RGB value obtained by the color histogram step, a clustering step of classifying pixels as pixels having a high appearance frequency as a normal part and pixels having a low appearance frequency as an uneven defect region,
A labeling step for identifying the uneven defect region classified in the clustering step and determining its area;
A surface inspection method comprising: an area threshold value comparison processing step for comparing the area of the concavo-convex defect region calculated in the labeling step with a threshold value to recognize the concavo-convex defect. In the imaging step, a light source is arranged at an arbitrary angle of 60 ° ± 15 ° with respect to the surface of the pressed steel plate, and light including infrared light is irradiated toward the steel plate, and a position facing the light source The surface inspection method according to claim 1, wherein the imaging unit is arranged at an arbitrary angle of 60 ° ± 15 ° with respect to the surface of the steel plate pressed in step 1 to detect reflected light from the steel plate.