JPH0792105A - Evaluation method for substrate treatment of steel plate - Google Patents

Abstract

PURPOSE:To evaluate the substrate treatment of steel plate accurately and automatically with no influence of illumination by normalizing the gray level of an image obtained by picking up the magnified image of a steel plate subjected to substrate treatment, further binarizing the normalized gray level, and determining the statistical distance value of a depth map. CONSTITUTION:The magnified image of a steel plate 1 subjected to substrate treatment is picked up and delivered to an image processor 3. The image processor 3 normalized the magnified image thus improving the overall balance of brightness and the contrast while eliminating the influence of illuminating conditions. The normalized image is further binarized and a set of dark or bright parts is extracted. The size of the set is then expressed in terms of distance value for the extracted image thus obtaining a distance image. A decision unit 4 determines statistical values, e.g. an average value or dispersion, by weighting the large distance values for the depth map and then evaluates the substrate treatment based on the statistical values.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for evaluating a steel sheet subjected to a base treatment.

[0002]

2. Description of the Related Art Steel sheets used for ships and the like are usually blasted. For the evaluation of the steel sheet subjected to the blast base treatment, visual inspection by an inspector and off-line inspection by a contact blast meter were performed.

At the laboratory level, a steel plate was photographed with a wide field of view, and the obtained image was binarized to obtain 2
Attempts have been made to discriminate macroscopically good and bad portions from a binarized image.

[0004]

However, in the method of visual observation, there are individual differences in the evaluation due to the experience of the inspector. Also, the inspection with the blast meter
Automation is difficult because a person contacts the steel plate with a blast meter for inspection. Moreover, in the method by the binarization processing,
Since this is not a microscopic evaluation that shows a difference in the accuracy of the base treatment, it is doubtful whether the evaluation method truly reflects the accuracy. In addition, it is easily affected by illumination conditions (irradiation angle, illuminance, etc.), and there is a problem in accuracy stability.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to solve the above-mentioned problems and to provide an evaluation method for steel plate undercoating which can automatically and accurately evaluate the steel plate undercoating without being affected by illumination. .

[0006]

In order to achieve the above-mentioned object, the present invention enlarges an image of a steel plate subjected to a base treatment, normalizes the obtained image with respect to a gray value, and performs a normalization with respect to the normalized image. Binarization is performed to extract the dark part and the light part with a predetermined gray level threshold, and the distance conversion is performed on the set of the dark part or the light part of the obtained binarized image, and the obtained distance image The statistical amount such as the weighted average value of the distance values is calculated with respect to, and the quality of the base treatment of the steel sheet is evaluated / determined based on the obtained statistical amount.

[0007]

According to the above construction, the steel plate subjected to the base treatment is replaced with C
A CD camera is used to magnify an image at a magnification so that the field of view has a characteristic in which the accuracy of the background processing appears. The data of the enlarged and photographed image is sent to the image processing apparatus. In the image processing device, the influence of the illumination condition is removed by performing the normalization process on the original image of the steel plate. Binarization processing is performed with a predetermined grayscale threshold value to extract a dark portion. By performing the distance conversion, the size of the set of dark areas is represented by the distance value. A statistic of the distance value (an average value (a cube mean, etc.) in which a large value is weighted, a variance, etc.) is obtained. The determination device evaluates the accuracy of the background processing by comparing the statistical values of the distance values according to a predetermined determination database.

[0008]

An embodiment of the present invention will be described in detail below with reference to the accompanying drawings.

FIG. 2 shows an embodiment of an apparatus to which the evaluation method for steel plate undercoating of the present invention is applied.

In the figure, reference numeral 1 is a steel plate which has been subjected to a blast base treatment, and 2 is a CCD camera for enlarging and photographing the steel plate to obtain image information. An image processing device 3 takes in image information from the CCD camera 2 and performs normalization, binarization, distance conversion, and the like. Reference numeral 4 denotes a determination device that determines / evaluates the accuracy of the background processing accuracy based on the distance information obtained from the image processing device, and includes a calculation device, an image processing device, and the like.

FIG. 1 is a flow chart showing the procedure of an evaluation method for steel plate undercoating according to the present invention.

In FIG. 1 and FIG. 2, an image of the steel plate 1 having been subjected to the base treatment is magnified and photographed by the CCD camera 2 and the image is inputted to the image processing device 3. As a result, the state (unevenness, etc.) of the underlying surface of the steel sheet 1 can be accurately grasped. Since the blast treatment is performed uniformly on the steel sheet 1, it is possible to grasp the state of the entire steel sheet by evaluating a part of it (S
-1).

By performing a normalization process on the image enlarged by the image processing apparatus 3, the brightness is balanced as a whole, the contrast becomes clear, and the influence of the illumination condition is eliminated (S-2). .

Here, the normalization processing will be described.

FIG. 3A is a grayscale histogram of the image before the normalization process, and FIG. 3B is a grayscale histogram after the normalization process. 3 (a), (b)
In the figure, the horizontal axis represents the gray value of the image, and the vertical axis represents the frequency.

The image from the CCD camera 2 may be too dark or too bright depending on the shooting conditions such as lighting conditions.
Alternatively, there may be a state where there is almost no contrast. Further, even for the same steel sheet, the light and shade distribution may differ in this way, and it is difficult to evaluate the steel sheet only with the input image. Further, it is difficult to automate because the grayscale threshold at the time of binarization cannot be fixed.

Therefore, by normalizing the grayscale value of the input image by the formula shown in Formula 1, it becomes possible to compare the grayscale value with the same average value and standard deviation irrespective of the illumination conditions. Further, since the distribution of the gray value is normalized, the gray threshold can be almost fixed.

[0018]

## EQU1 ## N (i, j) = (I (i, j) -μ) / σ where I (i, j) is the gray value of the input image of pixel (i, j), and N (i, j) ) Is the grayscale value of the normalized image of pixel (i, j), μ is the grayscale average value of the input image, and σ is the standard deviation of the input image.

From FIG. 3A, it is understood that the grayscale histogram before the normalization process is dark as a whole and has no contrast. On the other hand, in FIG. 3B, it can be seen that the brightness is well balanced and there is contrast.

Returning to FIG. 1, the normalized image is converted into 2
A binarized image is obtained by performing the binarization process, and a set of dark areas or bright areas is extracted. A range image is obtained by converting the size of the set into a distance value for the extracted image (S- 3, 4).

Here, the binary image and the range image will be described.

FIG. 4A is a schematic diagram of an image obtained by binarizing a normalized image with a predetermined grayscale threshold value and extracting a dark portion and a light portion.

In the figure, the shaded area is the dark area 5,
The portions other than the diagonal lines are the bright portions 6a and 6b. FIG. 4B is a schematic diagram of a distance value image obtained by performing distance conversion on a set of dark areas of this binarized image.

Here, the distance conversion will be described.

Generally, when an input image is divided into a figure and a background, the process of giving each point (pixel) in the figure the shortest distance from the background to that point is called distance conversion. There are several definitions of distance values, but typical ones are 4 neighborhood distances, 8
There is a near distance and an octagonal distance.

For two pixels, x = (i, j) and y = (k,
l)

[0027]

## EQU00002 ## d ₄ (x, y) = | i−k | + | j−1 |

[0028]

D ₈ (x, y) = max (| i−k |, | j−l |)

[0029]

D ₀ (x, y) = max {| i−k |, | j−
1 |, 2 (| i−k | + | j−1 | +1) / 3} In the example shown in FIG. 4, the dark portion 5 is the figure and the bright portions 6a and 6b are the background, and the distance value is Eight neighborhood distances were used. In FIG. 4 (c), a pixel having a distance value of 1 is indicated by an arrow from the pixels corresponding to the shaded portion in FIG. 4 (a). In FIG. 4B, the outer peripheral portion of the dark portion 5 and
The distance value around the bright portion 6b in the dark portion 5 is 1 and
It can be seen that the distance value increases from 2 to 3 as approaching the center of the dark portion 5.

Returning to FIG. 1, statistical values such as an average value (three-square average value) and a variance obtained by weighting a large distance value with respect to the distance image obtained by the determination device 4 are obtained (S-4). Since the base processing is evaluated based on the statistical value, accurate evaluation is performed (S-5).

(Concrete Example) FIG. 5A is a normalized image in the case where the steel sheet subjected to the base treatment is a defective product, FIG. 5B is its binarized image, and FIG. The distance-converted image, the same figure (d) is the histogram. Figure 6 (a)
Is a normalized image when the ground-treated steel plate is good, FIG. 7B is the binarized image, FIG. 7C is the distance conversion image, and FIG. Is. 5 (d) and 6 (d), the horizontal axis represents the gray value of the image and the vertical axis represents the frequency.

From FIGS. 5 (a) to 6 (d), the quality of the surface treatment of the steel sheet is reflected in the degree of variation in the histogram, so that the quality can be seen at a glance.

In the above, according to the present embodiment, the steel sheet subjected to the base treatment is magnified and photographed, the obtained image is normalized with respect to the gray value, and the gray scale threshold value determined in advance with respect to the normalized image is set. In order to extract the dark part and the bright part, the binarized image is binarized, distance conversion is performed on the set of dark parts or bright parts, and the weighted average of the distance values is obtained for the obtained distance image. Values such as statistics are obtained, and the quality of the undercoating of the steel sheet is evaluated / determined based on the obtained statistics, so the steel sheet undercoating can be accurately and automatically evaluated without being affected by lighting. It is possible to realize an evaluation method for steel plate undercoating.

In the present embodiment, the case where the steel plate is magnified and photographed using the CCD camera has been described, but the present invention is not limited to this, and if it is possible to magnify and photograph and convert it into an electric signal, an image pickup tube. Needless to say, an image intensifier or the like may be used.

[0035]

In summary, according to the present invention, the following excellent effects are exhibited.

The steel plate subjected to the base treatment is enlarged and photographed, normalized and binarized, and a distance conversion is performed on a set of dark portions or bright portions of the obtained binarized image to calculate a weighted average value of the distance values. Since the statistic is calculated and the quality of the groundwork of the steel sheet is evaluated / determined based on the obtained statistic, the steelwork groundwork can be accurately and automatically evaluated without being affected by the illumination.

[Brief description of drawings]

FIG. 1 is a flowchart showing a procedure of an evaluation method for steel plate undercoating of the present invention.

FIG. 2 is an example of an apparatus to which the evaluation method for steel plate undercoating of the present invention is applied.

FIG. 3A is a grayscale histogram of an image before being subjected to normalization processing, and FIG. 3B is a grayscale histogram after being subjected to normalization processing.

FIG. 4A is a schematic diagram of an image obtained by binarizing a normalized image with a predetermined grayscale threshold value and extracting a dark portion and a light portion, and FIG. 4B is the binarized image. It is a schematic diagram of a distance value image obtained by performing distance conversion on a set of dark areas of the image.

FIG. 5 (a) is a normalized image when the ground-treated steel plate is a defective product, and FIG. 5 (b) is a binarized image thereof.
(C) is the distance conversion image, and (d) is the histogram.

FIG. 6 (a) is a normalized image when the ground-treated steel plate is a good product, and FIG. 6 (b) is a binarized image thereof.
(C) is the distance conversion image, and (d) is the histogram.

[Explanation of symbols]

 1 Steel plate 2 CCD camera 3 Image processing device 4 Judgment device

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] June 23, 1994

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] 0032

[Correction method] Change

[Correction content]

From FIG. 5A to FIG. 6D, the quality of the base treatment of the steel sheet is reflected in the degree of variation in the histogram, so that the quality can be seen at a glance. Incidentally, FIG.
The magnifications of (c) and FIGS. 6 (a) to 6 (c) are all 1
It is 00 times. ─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] June 23, 1994

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] Fig. 4

[Correction method] Change

[Correction content]

FIG. 4A is a schematic diagram of an image obtained by binarizing a normalized image with a predetermined grayscale threshold value and extracting a dark portion and a bright portion, and FIG. 4B is the binarized image. FIG. 3 is a schematic diagram of a distance value image obtained by performing distance conversion on a set of dark areas of the image.
(C) is a pixel corresponding to the shaded area in (b)
Pixels having a distance value of 1 are indicated by arrows.

[Procedure Amendment 2]

[Document name to be corrected] Drawing

[Name of item to be corrected] Figure 5

[Correction method] Change

[Correction content]

[Figure 5]

[Procedure 3]

[Document name to be corrected] Drawing

[Name of item to be corrected] Figure 6

[Correction method] Change

[Correction content]

[Figure 6]

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shohei Misono, 1-1-15 Toyosu, Koto-ku, Tokyo Ishikawajima Harima Heavy Industries Co., Ltd. Toni Technical Center

Claims (1)

[Claims] 1. A magnified image of a base-treated steel sheet is normalized with respect to a gray value of the obtained image, and a dark portion and a light portion are separated from the normalized image by a predetermined gray level threshold value. It is binarized for extraction, the distance conversion is performed on the set of dark parts or bright parts of the obtained binarized image, and the statistical value such as the weighted average value of the distance values is obtained for the obtained distance image. A method for evaluating a steel sheet undercoating method, comprising: evaluating / determining the quality of the undercoating treatment for the steel sheet based on the obtained statistics.