JPH06187492A - Character string image pickup method - Google Patents

Abstract

PURPOSE:To provide a method for picking up the image of a character string in the identifiable state by calculating a, distance between two reference patterns provided at a prescribed gap parallelly to the outside of the character string by means of photographic data and adjusting the position or focal distance of a photographing device so that the distance becomes a prescribed value. CONSTITUTION:The square reference patterns smaller than characters are provided just above the first character in the top line of the character string to be marked on a billet end surface 21 and just above the end character. As a reference state, the distance between a TV camera 22 and the billet end face 21 is defined as Lo, and the distance between the reference patterns provided on the TV camera 22 at such a time is defined as Ho. When the distance between the reference patterns obtained at the time of picking up the image of the character string on the billet end surface 21 for any arbitrary billet is defined as H, an image with the distance Ho between the reference patterns can be obtained on the camera 22 by changing the distance between the end surface 21 and the camera 22 just by L=Lo(1-H/Ho). Instead of moving the TV camera 22, the focal distance of the zoom lens of the TV camera 22 may be changed as well.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character string imaging method for reading a character string.

[0002]

2. Description of the Related Art In many cases, a product being manufactured or shipped in a factory is marked with an attribute of the product. Although such markings are read and physical distribution management of the product is performed, an apparatus is used that performs this reading by an image pickup device and determines the written character string. Conventional marking and reading on steel materials have many achievements for large surfaces such as slabs, and rod-shaped steel materials (hereinafter called billets).
In this case, the longitudinal surface was marked, but recently, the end face of the billet has been marked.

The billet is flowing on the conveyor in a state where the length direction is perpendicular to the traveling direction and the end face is visible from the side of the conveyor. The conveyor is temporarily stopped and the marking on the end face is read. Work is done. The reading is performed by the image pickup device, and the marked character is identified from the read image.

[0004]

However, the distance between the end face of the billet flowing on the conveyer and the side face of the conveyer is not constant, and there is entrance and exit. For this reason, the distance between the imaging device and the end face of the billet is not constant, the camera is out of focus, the magnification of the imaged image fluctuates greatly, and the marking characters may not be identified. Further, the billet may be gently twisted in the length direction, and at this time, the square of the end face may be inclined by an angle Δθ with respect to the conveyor surface.
Also, since the billet has a nearly square cross section, the marked characters may be placed on the conveyor in a state rotated by 90 degrees or 180 degrees. Therefore, it has been difficult to identify the marked characters.

The present invention has been made in view of the above problems, and an object thereof is to provide a character string imaging method for imaging a marked character string in a distinguishable state.

[0006]

In order to achieve the above object, the present invention draws a predetermined character string, and provides two reference patterns parallel to the outside of the character string at predetermined intervals. Is imaged by the image pickup device, the distance between the two reference patterns is calculated from the image pickup data, and the distance between the character string and the image pickup device or the camera of the image pickup device is calculated so that the calculated distance becomes a predetermined value. The focal length of is adjusted.

Further, a predetermined character string is drawn on the rectangular cross section, and two reference patterns are provided at predetermined intervals in parallel to the outside of the character string. When the rectangular cross section is normal, 90 degrees, 180 degrees
4 states of the reference pattern when rotated by 270 degrees and 270 degrees respectively are stored, the rectangular cross section is imaged by an image pickup device, and the arrangement of the two reference patterns is detected to determine which of the 4 states is hit. However, when the image is different from the normal case, the image rotation processing is performed so that the captured image is in the normal case.

After detecting the arrangement of the two reference patterns, a straight line connecting the same positions of the two reference patterns and a scanning line of the image pickup device or a straight line orthogonal to the scanning line in the same plane. Find the intersection angle Δθ with
After the image is rotated so as to determine which of the above four states is to be met.

[0009]

A character string of a predetermined size is drawn, two reference patterns are provided in parallel to the outside of the character string at predetermined intervals, and the character string is imaged with the camera of the image pickup device in focus. , The distances of these two reference patterns on the image are stored as predetermined values. Next, this character string is picked up by an image pickup device, the distance on the image of these two reference patterns is calculated,
If the distance between the character string and the imaging device or the focal length of the camera of the imaging device is adjusted so that this becomes a predetermined value, the image input magnification of the imaging device for this character string can be held constant.

A predetermined character string is drawn on a rectangular cross section, and two reference patterns are drawn in parallel with the outside of the character string at predetermined intervals.
If this rectangular cross section is normal, the four states of the reference pattern are stored for each rotation of 90 degrees, 180 degrees, and 270 degrees. Next, this rectangular cross section is imaged by an image pickup device, it is judged which of these four states the arrangement of these two reference patterns corresponds to, and if it is different from the normal state, the image is rotated so that it becomes the normal state. To do. As a result, the character string can be recognized even when the rectangular cross section is rotated and arranged.

After detecting the two reference patterns, a straight line connecting the same positions of the two reference patterns, for example, the center of gravity, is obtained, and a crossing angle between the straight line and a scanning line of the image pickup device or a straight line orthogonal to the scanning line. Δθ is obtained, and the image is rotated so that this Δθ becomes zero. This makes it possible to correct small rotated images instead of large rotations of 90 degrees. If the character string is tilted more than a certain angle, it becomes difficult to identify the character, so this correction is performed.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of an image processing apparatus that realizes an embodiment of the present invention. An A / D converter 2 for converting the input data from the TV camera 1 from analog to digital, an input buffer 3 for storing the converted data, a bus 4, and the constituent characters of the input image data are separated to identify the characters. A CPU 5 that controls the entire operation, an image processor 6 that performs image processing of image data, a program memory 7 that stores a program that determines the operation of the CPU 5, and data that has been binarized and processed by the image processor 6 to be stored. A binary image memory 8, a grayscale image memory 9, an output buffer 10 for temporarily storing output data, a D / A converter 11 for converting digital data of the output data into analog data, and a CRT 12 for displaying the output data. To be done. It is assumed that the television camera 1 is configured so that its position can be adjusted in the direction of the object.

FIG. 2 is a diagram showing a first embodiment of the present invention. In the present embodiment, when the characters marked on the billet end face 21 are imaged by the TV camera 22 as an example of an image pickup device to recognize the characters, the distance L between the billet end face 21 and the TV camera 22 is kept constant and the image input magnification is increased. It is a method of keeping it constant. The billet is conveyed by the conveyor, but when it is installed on the conveyor, the side surface of the conveyor and the end surface of the billet 21
The interval between and is different for each billet. Therefore, when the TV camera 22 is used to capture the characters marked on the billet end face 21, if the TV camera 22 is fixed, the billet end face is fixed.
The distance L between the TV 21 and the TV camera 22 differs from billet to billet.
An image in which the image input magnification of the V camera 22 changes is also generated.
The image pickup device may be a combination of an image sensor and an encoder.

A square reference pattern smaller than a character is provided immediately above the first character and immediately above the last character of the uppermost line of the character string to be marked on the billet end face 21. As a reference state, the distance between the TV camera 22 and the billet end face 21 is L0, and the distance between the reference patterns obtained on the TV camera 22 at this time is H0. If the distance between the reference patterns obtained on the TV camera 22 when the character string on the billet end face 21 is imaged by the TV camera 2 for an arbitrary billet is H, then ΔL shown in the following equation Camera 22
If the distance between the reference patterns is changed, an image with the distance between the reference patterns H0 can be obtained on the TV camera 22. ΔL = L0 (1-H / H0) (1)

Distance L between billet end face 21 and TV camera 22
When adjusting ΔL by ΔL, the billet is heavy and difficult to move, so that the TV camera 22 can be moved.
The image of the marking characters on the billet end face 21 is taken while the conveyor is being moved relative to the image pickup device in the case of a TV camera. Also, only ΔL is a TV camera
Instead of moving 22, the focal length of the zoom lens of the TV camera 22 may be changed.

FIG. 3 shows an overall operational flow chart including the first and second embodiments. First, the process of extracting the reference pattern shown in FIG. 2 is performed (step 30). Next, using this reference pattern, it is checked whether the magnification of the TV camera 22 is a predetermined magnification, and if not, magnification correction processing is performed (step 50). Next, it is checked whether the image of the billet end face 21 is in the correct position so that the marking characters can be read, and if not, the rotation is corrected (step 60). The contents of each of these steps will be described in detail below.

FIG. 4 is a flow chart of the reference pattern extraction processing in step 30. First, enter the image (Step 3
1), inter-character connection processing is performed (step 32). As one of the inter-character connection processing methods, there is a method in which the maximum value filter processing is performed n times, and each character is thickened and then combined. This takes, for example, 3 × 3 pixels around the target pixel, and replaces the maximum density value in this area with the density of the target pixel. By doing so, each character becomes bold as shown in FIG. At the nth time, each letter is spread out and combined into a rectangle for each letter. As a result, the character area can be extracted. Before performing this filter processing, in order to remove noise, those smaller than the character area are erased. As a result, the reference pattern also disappears. Next, shading correction is performed (step 33). Shading refers to unevenness due to the location of illumination, uneven brightness within the lens area, uneven adhesion of marking ink, unevenness due to the location of the state of the marking background billet surface, and the like. By performing the shading correction, there is an effect that the contrast between the character and the background is made uniform.

The inter-character combined image subjected to shading correction is binarized to extract the inter-character combined image area (step 34), and the inter-character combined area is labeled to distinguish it from others (step 35). The combined area is extracted as a character area (step 36). Next, a character arrangement determination process for the extracted area is performed (step 37). This determination process is a process of calculating the number of character digits from the width of the character area based on the character layout design data. Explaining with FIG. 5, since the size of one character is known from the rectangular area at the bottom, the character arrangement of five characters is determined.

When each character arrangement is determined in this way, masking of the vicinity of the character area is performed (step 38).
Since the reference pattern is near the first character and the last character, the area containing these characters is masked as shown by the broken line in FIG. Next, reference pattern determination processing is performed (step 39).

As an example of the determination process of the reference pattern, a method of obtaining it from the shape coefficient and the particle area 4 of the reference pattern will be described. The following formula is used as the shape factor SF. SF = πL ² / (4A) (2) L is the maximum length of the particle (2) is 1 when the particle is a circle. Since the reference pattern is a square, L has a diagonal length and SF = π / 2. If the SF is close to π / 2 and the particle area is close to A, the reference pattern can be obtained. In this way, the reference pattern is obtained (step 40).

FIG. 7 shows a case where the billet end face 21 is conveyed on a conveyor in a state of being rotated by 90 degrees. In this case as well, mask the areas where the first and last characters of the top and bottom lines are located as shown by the broken lines, and step the reference pattern.
It can be extracted using the method shown in 40 or the like.

FIG. 8 is a flow chart of magnification correction processing. After extracting the reference patterns in the operation flow chart of FIG. 4, the distance H between the reference patterns is calculated as shown in FIG. 2 (step 51). Next, the distance between the billet end face 21 and the TV camera 22 is set to L0, the distance between the reference patterns at this time is set to H0, and these L0 and H0 are used as references. The ratio of H and H0 is used as the input magnification (step 52). In order to set this ratio H / H0 to 1, that is, H to H0, it is sufficient to move the TV camera 22 toward the billet end face 21 by ΔL using the equation (1) (step 53). As a result, the size of the image obtained by the TV camera 22 is always H0. The billet may be moved or the focal length of the zoom lens of the TV camera 22 may be changed.

Next, a second embodiment will be described. In the present embodiment, the billet end face 21 is mounted on the conveyor in a rotated state from the marked state. There are two types of this rotated state, that is, a small angle Δθ rotation of about 10 degrees as shown in FIG. 10 and a 90 degree rotation as shown in FIG. The rotation of a small angle Δθ is the case where the billet is twisted in the lengthwise direction, which causes a smaller angle Δθ rotation than when marking, and when rotating in 90 degree units, it is mounted on the conveyor after marking. It occurs when it is rotated and mounted.

FIG. 9 is an operation flow chart of this embodiment. First, it is checked whether the reference pattern direction is rotated by a small angle Δθ (step 61). As shown in FIG. 10A, the rotation of the small angle Δθ is such that the reference line P passing through the center of gravity of the reference pattern is in the 0 ° or 180 ° direction with respect to the scanning line A of the TV camera 22, and And a small angle Δθ rotation, (b)
As shown in, the reference line P is in the direction of 90 degrees or 270 degrees,
It may be rotated by a small angle Δθ with respect to the line B orthogonal to the scanning line A. The Δθ minute rotation processing is performed so that this small angle Δθ becomes 0 (step 62).

FIG. 11 shows that the billet end face 21 is in the normal state shown in (a), and this is I, and as shown in (b), the billet end face 21 is rotated 90 degrees II, 180 degrees III, 270 degrees IV. It is in a state. Since the arrangement of such reference patterns is extracted in steps 39 and 40 of FIG. 4, 90 degrees II, 180 degrees II
If it is in the I, 270 degree IV state, the image is rotated to the I state (step 63).

[0026]

As is clear from the above description, according to the present invention, it is not necessary to adjust the billet side by moving the image pickup device or adjusting the focal length of the zoom lens even if the position of the end face of the billet changes. There is no need for a positioning mechanism or billet position detection sensor. Further, even if the billet end face is changed from the marked state to the rotated state, the billet does not need to be rotated because the image processing technique performs the rotation process to a normal image. By these processes, it becomes easy to correctly recognize the characters marked on the end face of the billet.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a computer that implements an embodiment of the present invention.

FIG. 2 is an explanatory diagram of the first embodiment.

FIG. 3 is an overall flow chart including a first embodiment and a second embodiment.

FIG. 4 is a flowchart of reference pattern extraction processing.

FIG. 5 is a diagram illustrating maximum value filtering processing.

FIG. 6 is a diagram showing a masking process for extracting a reference pattern.

FIG. 7 is a diagram showing a mask process when rotated by 90 degrees.

FIG. 8 is a flowchart of magnification correction processing.

FIG. 9 is a rotation correction processing flowchart.

FIG. 10 is an explanatory diagram of a minute angle Δθ.

FIG. 11 is a diagram illustrating a rotation process in units of 90 degrees.

[Explanation of symbols]

 21 Billet end face 22 TV camera

Front page continuation (72) Inventor Katsuho Aikawa 2951 Ishikawa-cho, Hachioji-shi, Tokyo 4 Nireko Co., Ltd. 72) Inventor Mitsuo Yoshida, 12 Nakamachi, Muroran City, Hokkaido Inside Nippon Steel Co., Ltd. Muroran Works (72) Inventor Akio Ishiwata 12 Nakamachi, Muroran City, Hokkaido Shin Nippon Steel Co., Ltd. Muroran Works

Claims (3)

[Claims] 1. A predetermined character string is drawn, two reference patterns are provided in parallel to the outside of the character string at predetermined intervals, and the character string is imaged by an image pickup device, and between these two reference patterns. Is calculated from the image pickup data, and the distance between the character string and the image pickup device or the focal length of the camera of the image pickup device is adjusted so that the calculated distance becomes a predetermined value. Character string imaging method. 2. A predetermined character string is drawn on a rectangular cross section, and two reference patterns are provided parallel to the outside of the character string at predetermined intervals. When the rectangular cross section is normal, 90 degrees, 180 degrees, 27
The four states of the reference pattern when each is rotated 0 degrees are stored, the rectangular cross section is imaged by the image pickup device, the arrangement of the two reference patterns is detected, and it is determined which of the four states is hit, A character string imaging method, wherein when the image is different from the normal case, the image is rotated so that the captured image is in the normal case. 3. After detecting the arrangement of the two reference patterns, a straight line connecting the same positions of the two reference patterns and a scanning line of the image pickup device or a straight line orthogonal to the scanning line in the same plane. 3. The character string imaging method according to claim 2, wherein an intersection angle .DELTA..theta. With .alpha. Is obtained, and after which the image is rotated so that .DELTA..theta. .