JPH0812682B2 - Stamped stamp detection method and device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a stamp detection method and device, and more particularly,
The present invention relates to a method and an apparatus for detecting an embossed mark in which a marking stamped on a relatively small-sized primary rolled steel material is photographed by a television camera and the image is processed to recognize the content.

[Prior Art] For example, a relatively small primary rolled steel material (for example, billet, bloom) is marked with numbers, letters, symbols, etc., and this marking is automatically read and recognized,
This is used for manufacturing, processing, inventory control, etc.

The marking is formed by pressing a printing member (character die) in which numbers, letters, symbols, etc. are reversed in the left-right direction on the end surface of the primary rolled steel material to engrave it, and engraving the numbers, letters, symbols, etc. It is formed in a groove shape.

As one of the methods, a stamped and printed image formed on a primary rolled steel material (hereinafter referred to as a marking material) is obtained as a multivalued original image signal by an image pickup device such as a television camera, and a binary image obtained by conversion is obtained. Some recognize characters using the final image. As image processing for obtaining a final image for character recognition, there is, for example, JP-A-61-80372. The technique shown therein uses one threshold value for the entire image in the binarization process for converting a multi-valued original image into a binary final image. It is characterized in that it is updated and determined until the number of low-luminance areas in (3) matches the set value.

[Problems to be Solved by the Invention] However, in the conventional stamp detection device as described above, since the binarization processing is performed on the entire image using only one threshold value, There is a case of misrecognition.
That is, the stamped surface is not flat, and rust and contamination are often generated, and the degree of occurrence and the shape condition are not constant. For this reason, stains and the like are detected together with the marking, resulting in erroneous recognition and unrecognizable result.

The present invention has been made in order to solve the problems of the prior art, and it is an object of the present invention to provide a method and an apparatus for detecting an embossed mark that can obtain a clear final image regardless of the state of the imprinted surface.

[Means for Solving the Problems] In order to achieve the above object, the present invention captures an embossed print image of an embossing surface, and from the image signal, the direction of the marking material and the image of each embossed print image. The position on the upper side is detected, and based on this, the image signal is divided for each embossed print image to obtain a plurality of embossed print images, and binarization processing is individually performed for each of the plurality of embossed print images. It is applied to obtain information for character recognition.

Then, in order to obtain a clear engraved image, the image signals are further moved by a plurality of predetermined directions and distances to obtain a plurality of moving image signals, and the moving image signals are added and obtained. It is desirable that the marking image-enhanced image signal is subjected to the marking material orientation detection processing.

Further, as an apparatus configuration, an illuminating means for illuminating the stamped surface of the stamping material from an oblique direction, an imaging means for imaging the stamped surface, and a stamping material based on the density projection of a specific range of the original image signal by the means. A marking material orientation detection processing unit that detects the orientation of the marking image, a marking position detection processing unit that obtains the position of the marking image based on a change from the integrated value of the density projections of the marking material in a specific range of the original image signal, and the processing unit. A marking image separation processing unit that separates the marking image into print units based on the marking position information, and a binarization processing unit that independently binarizes each marking-specific image signal separated by the means. A configuration consisting of is recommended.

Further, in order to obtain a clear marking image, the image is moved by a plurality of predetermined directions and distances with respect to the original image signal applied to the marking material direction detection processing unit and the marking position detection processing unit, and emphasis processing is performed. It is possible to provide a stamped image enhancement processing unit for performing.

[Operation] With the configuration described above, it is possible to detect which rotation angle of the marking material is in 90 degree units from the normal image position with respect to the original image signal obtained by imaging the marking surface of the marking material. Further, the position of each marking is obtained from the unevenness of the integrated value of the density projection in the specific range of the marking material. Based on this, the final image for character recognition can be obtained by binarizing the images for each marking separated for each marking. Therefore, since the optimal binarization threshold can be set for each character,
Binarization can be performed for each character independently, and a character pattern can be satisfactorily extracted without causing a situation such as a missing character.

Further, since the density difference between the background and the character can be clarified by performing the marking image enhancement process, a clear marking image can be obtained.

[Embodiment] FIG. 1 is a block diagram showing an embodiment of the present invention.

The imaging device 1 captures an image of a stamped surface of a marking material (for example, a billet), and a television camera is used, for example. An embossing detection device 2 is connected to the imaging device 1, and the character recognition processing units 3 _{-1 to} 3 _-N are connected to the embossing detection device 2.

As shown in FIG. 2, the image pickup device 1 is arranged so as to face the engraved surface of the marking material 4, and a plurality of lamps 5 are arranged to illuminate the engraved surface.

In order to reliably extract the stamped image, it is desired that the density difference (contrast) between the background and the character is large. This density difference is the straight line length of a character that is determined by the illumination direction of the illumination light and the character shape. (Here, when a straight line parallel to each illumination direction is considered, the length of the line that overlaps the character is the longest. The thing is defined as the effective straight line length), and is related to the character's embossing depth. In order to improve the extraction, it is preferable that the characters are deeply stamped and the effective straight line length is shortened. To shorten the effective straight line length, the illumination by the lamp 5 is oblique to the stamped surface. It is better to start from

The marking detection device 2 is a marking image enhancement processing unit that performs processing for enhancing only the marking image in the original image signal 11 from the imaging device 1.
The marking material orientation detection processing unit 7, the marking position detection processing unit 8, the marking image separation processing unit 9, and the binarization processing units 10 _-1 to 10 _-N .

Next, the general operation of the configuration shown in FIG. 1 will be described.

First, the engraved surface of the marking material 4 is imaged from the direction shown in FIG. 2 by the imaging device 1 to obtain the engraved image as a multivalued original image signal 11. On the other hand, the marking image emphasizing processing unit 6 performs processing for emphasizing only the marking image, and outputs the marking image emphasizing signal 12. Then, the marking material orientation detection processing unit 7 detects the orientation of the marking material and outputs the marking material orientation information 13.

Further, the marking position detection processing unit 8 detects the marking position based on the marking image enhancement signal 12 and the marking material orientation information 13. Based on the marking position information 14 and the original image signal 11 obtained by the marking position detection processing unit 8, the marking image separation processing unit 9 separates the original image signal 11 for each marking, and N marking-specific image signals. Get 15. Image signal according to this stamp 15
Contains only one of the N imprints. Therefore, the binarization processing units 10 _-1 to 10 _-N independently perform binarization processing on each of the markings to obtain final images 16 _{-1 to} 16 _-N.
Is obtained. The character recognition processing units 3 _{-1 to} 3 _-N perform character recognition processing on the final images 16 _{-1 to} 16 _-N .

Next, details of the marking image enhancement processing unit 6 will be described with reference to FIG.

The marking formed on the marking material 4 is captured as an image by photographing a shadow generated by illumination (hereinafter, an image formed by this shadow is referred to as a marking image). Therefore,
For the imaging apparatus 1, if the marking image is clear enough to perform the marking material orientation detection processing and the marking position detection processing, the marking image enhancement processing is not necessary. However, it is generally difficult to obtain a marked image clearly, and good image detection cannot be obtained unless the marked image enhancement processing is performed.

The stamped image enhancement processing unit 6 includes image movement processing units 61 _{-1 to} 6-1.
1- _N , weight value setting units 62 _{-1 to} 62 _-N , multiplication units 63 _{-1 to} 63 _-N , and an addition unit 64.

The image movement processing units 61 _{-1 to} 61 _-N move the original image signal 11 by a predetermined direction and distance with respect to the original image signal 11 to obtain a plurality of moving image signals 65 with different directions and distances. . For each of the moving image signals 65, the weight value setting unit 62
_The weighted moving image signal 66 is obtained by multiplying the respective weight value settings of _{-1 to} 62 _-N by the multiplication units 63 _{-1 to} 63 _-N . Multiplier
The output signals of 63 _{-1 to} 63 _-N can be added by the addition unit 64 to obtain the stamped image enhancement signal 12.

If the process in the stamped image enhancement processing unit 6 is expressed by a mathematical expression,
It looks like this:

That is, the density value of the coordinates (x, y) of the original image signal 11 is f
Assuming that (x, y), the density value g (x, y) at the coordinates (x, y) of the marking-enhanced image is given by the following equation.

Here, xi and yi are movement amounts in the x and y directions when the original image signal is subjected to image movement processing, and ai is a weighting constant. Further, the constants (N, xi, yi, ai) in the above equation can determine the target original image signal from the characteristics. By selecting the optimum constant, it is possible to enhance only the imprinted image without enhancing the noise component.

Next, the details of the processing of the marking material orientation detection processing unit 7 will be described with reference to FIG.

The marking of the marking material 4 (in this embodiment, a combination of numbers in three stages, the first stage is composed of 5 digits, the second stage is composed of 2 digits, and the third stage is composed of 1 digit) is most suitable for the image pickup apparatus 1. It is not always sent at the location. Therefore, the stamped image enhancement signal 12 (or the original image signal in the configuration without the stamped image enhancement processing unit 6)
Vertical and horizontal are determined from the image in 11), and the image is converted to an image in the correct position. Specifically, the density projections in the vertical and horizontal directions of the stamped image are obtained, and the markings are arranged with the center of the stamping material 4 as the center of rotation.
If it is not a point object, vertical and horizontal integrated values depending on the orientation of the marking material 4
The maximum value position c, d of a, b or the size is different.

By performing an analysis based on this information, the marking material 4
The direction of can be detected. For example, it is determined that the rotation angle is 0, 90 degrees, 180 degrees, or 270 degrees depending on whether the intersection of the maximum value positions is the upper left, upper right, lower right, or lower left of the marking material 4. .

Next, details of the processing of the marking position detection processing unit 8 will be described with reference to FIG.

In this process, the density projection is calculated for the marking image emphasis signal 12 or the marking material orientation information 13 in a specific range of the marking material 4 (in this embodiment, the density projection calculation range 1 in FIG. 5), and the integrated value thereof is obtained. The position of the marking is detected by detecting the unevenness of the. In FIG. 5, 5 in the upper line of the marking
Although two numbers are detected, it is possible to detect the positions of all the markings by repeating the same process.

Next, the details of the binarization processing units 10 _-1 to 10 _-N will be described with reference to FIG.

This processing is to binarize the marking-specific image (multi-valued image) obtained by the marking image separation processing unit 9. The binarization processing units 10 _-1 to 10 _-N have the same configuration, and FIG. 6 shows one configuration thereof.

As shown in FIG. 6, the binarization processing unit 10 _-1 includes a threshold value determination processing unit 101 that determines a threshold value for the marking-based image signal 15, a threshold value determination unit 102 that determines the threshold value, The binarization processing unit 103 and the binarization processing unit 10 based on the image signal 15 for each marking
It is composed of a noise removal processing unit 105 that performs noise removal processing on the binarized image 104 according to 3.

The general operation in this configuration is to detect the density histogram of the marking-based image (multi-valued image) and binarize the marking-based image based on the binarization threshold value determined thereby. As it is, since it contains various noises, the noise removal processing unit 105 performs noise removal processing on the binarized image.

This binarizing process will be described with reference to FIGS. 7, 8A, 8B and 9 by taking each billet shown in FIG. 5 as an example.

Next, details of operations of the binarization processing units 10 _-1 to 10 _-N will be described.

As shown in FIG. 7, the input image is divided so that each character is surrounded by a frame, and the frames w1 to w8 are assigned. This frame
For each of w1 to w8, the inside of the frame is binarized by the discriminant threshold method (threshold value: th1, separation b1), and a binarized image wa1 is obtained (S
81). Then, measure the area ar1 of the remaining area (S82),
As shown in FIG. 9, binarization processing (threshold value: th2, degree of separation: b2) is performed on a point having a density equal to or greater than the threshold value th1 in the frame (the number of pixels) by the discrimination threshold method, A binarized image wa2 is obtained (S83).

Subsequently, the area ar2 of the remaining region is measured (S84) and the degree of separation is determined.
b1 and b2 are compared (S85). In S85, if b1> b2, the flow moves to the flow of FIG. 8B, and if b2> b1, the flow moves to the flow of FIG. 8B.

In the flow, the area ar1 is compared with the area threshold arth, and a
If r1 <arth, the image to be processed thereafter is set to wa1 (S
87). If ar1> arth, the image to be processed thereafter is set to wa2 (S88). The process of S88 is also executed when the flow is selected in S85. The above S81 to S88 correspond to the processing contents of the threshold value determination processing unit 101 and the threshold value 102 shown in FIG.

When the processing of S86 or S88 is completed, the feature amount in the processing target image is measured (S89). Next, the area is 30 pixels (pi
Those smaller than (x: pixels) are removed (S90), and those having an area ratio of 0.5 or more are removed (S91). S89 to S91 correspond to the processing contents of the noise removal processing unit 105 shown in FIG.

By the way, the marking material 4 is not always conveyed to the photographing region of the image pickup apparatus 1 one by one, but may be conveyed in a bundle of several marking materials. In this case, it is necessary to identify each stamp after identifying them one by one. This process will be described below.

Here, as shown in FIG. 10, there are three marking materials in the photographing area, one marking material is located in the center of the photographing area, and two marking materials are positioned on both sides of the marking material at a predetermined interval. An example will be explained. Then, elongated frames w10, w20, w30, and w40 are set horizontally and vertically from the center of the photographing area. The resolution of the input image is, for example, 0.305 × 0.305 mm / 1 pixel level, the width of the frame w is 100 pixels (pix), and the length of the frame is 256 pixels.

Next, the specification of the marking material will be described with reference to FIGS. 11, 12, and 13.

The three marking materials are bundled in close contact with each other, but since the peripheral edges of both ends are rounded, a groove deeper than the marking is formed between the adjacent markings and outside the peripheral edges. ing. Therefore, when the inside of the frame is binarized, a change point of a binary level that is clearly different from the marking occurs, and the position of this changing point can be regarded as the peripheral edge of the marking surface.

Specifically, first, binarization processing is performed on each image in the above-mentioned frames w10, w20, w30, w40 by the discrimination threshold method (S121). Then, using the binarization result, the frames w10, w20,
For each of w30 and w40, when the profile is sequentially measured from the center to the outside in the lengthwise direction, as shown in FIG. 11, the peripheral edge of the marking surface of the marking material 4 to be extracted and the marking The profile changes greatly at the position where the gap between the materials (or between the open spaces) comes into contact, and the change point h1 (frame w1
Existing in 0), change point h2 (existing in frame w20), change point h3 (existing in frame w3
Existing in 0), the change point h4 (existing in the frame w40), (S1)
twenty two). The part surrounded by these change points is determined to be the end face of the marking material 4 (S123), and this is set to white as shown in FIG. 13 on the image processing, and then shown in FIGS. 8A and 8B. The marking detection process is performed by the procedure described above to obtain the final image.

 Next, specific examples of the present invention will be described.

Here, it is assumed that the marking material 4 is a square billet of about 120 mm × 120 mm, and the product number is stamped on the end face. The product number is detected by the marking detection device 2, and this is detected by the character recognition processing unit 3 _-1. ~ 3 _-N recognizes characters. Further, the respective constants in the marking image enhancement processing unit 6 are set as follows.

N = 5 (xi, yi, ai) = (0,0,4), (4,0, -1), (-4,0, -1), (0,4, -1), (0, -4, -1) As a result, in the detection by the conventional technique, partial rust or stain remained on the final image, and the recognition rate was about 76%. However, in the detection according to the present invention, it was confirmed that a clear final image having only the marking can be obtained, and the recognition rate reaches 95%.

[Effects of the Invention] Since the present invention is configured as described above, it has the effects described below.

According to the method of detecting an embossed mark of claim (1), an embossed printed image of the embossed surface is captured, and the direction of the engraved material and the position of each embossed printed image on the image are detected from the image signal, Based on this, the image signal is divided for each stamp print image to obtain a plurality of stamp print images, and each of the plurality of stamp print images is individually binarized to obtain information for character recognition. Therefore, the binarization can be performed for each character independently, and the character pattern can be satisfactorily extracted without causing a situation such as a missing character.

According to the stamp detection method of claim (2), a plurality of moving image signals are obtained by moving the image signals by a plurality of predetermined directions and distances, and the moving image signals are added together. Since the obtained marking image-enhanced image signal is used for the direction detection processing of the marking material, the density difference between the background and the character can be clarified, and a clear marking image can be obtained.

According to the embossing mark detection apparatus of claim (3), the illumination means for illuminating the embossing surface of the engraving material from an oblique direction, the imaging means for imaging the embossing surface, and the specification of the original image signal by the means. A marking material orientation detection processing unit that detects the orientation of the marking material based on the density projection of the range, and a marking that determines the position of the marking image based on the change from the integrated value of the density projection of the marking material in the specific range of the original image signal. A position detection processing unit, a marking image separation processing unit that separates the marking image into printing units based on the marking position information by the processing unit, and a binary value independent for each of the marking-specific image signals separated by the means. Since the binarization processing unit for performing the binarization is provided, the binarization can be performed for each character independently, and the character pattern can be satisfactorily extracted without causing a situation such as missing of the character.

According to the marking detection device of claim (4), the image is moved by a plurality of predetermined directions and distances with respect to the original image signal applied to the marking material orientation detection processing unit and the marking position detection processing unit. Since the engraved image emphasizing processing unit for performing the emphasizing process is provided, the density difference between the background and the character can be clarified, so that a clear engraved image can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is an explanatory view showing attachment and illumination means of the image pickup apparatus shown in FIG. 1, and FIG. 3 is a block showing details of a stamped image enhancement processing section. 4 and 5 are explanatory views showing the processing of the marking material orientation detection processing unit, FIG. 5 is an explanatory drawing showing the processing of the marking position detection processing unit, and FIG. 6 is a block diagram showing the details of the binarization processing unit. , FIG. 7 is an explanatory view showing the framing process of the input image, FIGS. 8A and 8B are flowcharts showing the process of the binarization processing unit, FIG. 9 is an explanatory view showing the threshold setting, and FIG. The figure is an explanatory view showing the process of specifying one target from a plurality of marking materials, Fig. 11 is a change point detection characteristic diagram showing a profile process for specifying the marking material, and Fig. 12 is a process for obtaining a change point. 13 is a mask image diagram of the marking material specified based on the change points. In the figure, 1: imaging device, 2: marking detection device 3 _{-1 to} 3 _-N : character recognition processing unit 4: marking material, 5: lamp 7: marking material orientation detection processing unit 8: marking position detection processing unit 9 : Stamped image separation processing unit 10 _-1 to 10 _-N : Binarization processing unit 11: Original image signal 12: Stamped image enhancement signal 13: Stamped material orientation information 14: Stamped position information 15: Stamped image signal 16 _-1 ~ 16 _-N : Final image 61 _-1 ~ 61 _-N : Image movement processing part 62 _-1 ~ 62 _-N : Weight value setting part 63 _-1 ~ 63 _-N : Multiplying part 64: Addition part 101: Threshold value Decision processing unit 103: Binarization processing unit 105: Noise removal processing unit

Claims (4)

[Claims] Claims: 1. An embossed print image of an embossed surface is imaged, density projections in the vertical and horizontal directions of the imprinted image obtained by the imaging are obtained, and the maximum value position in the vertical and horizontal directions of the integrated value is obtained. The direction of the marking material and the position of each stamp print image on the image are detected from the unevenness, and based on this, the image signal is divided for each stamp print image to obtain a plurality of stamp print images. A method for detecting an embossed mark, which comprises individually binarizing each of the embossed printed images to obtain information for character recognition. 2. An imprinted image-enhanced image signal obtained by adding a plurality of moving image signals by moving the image signals by a plurality of predetermined directions and distances and adding the engraved image signals. The method for detecting an embossing mark according to claim 1, wherein the method is used for a process for detecting the orientation of a material. 3. An illuminating means for illuminating the stamped surface of the stamping material from an oblique direction, an image pickup means for capturing an image of the stamped surface, and a stamping material of the stamping material based on the density projection of a specific range of the original image signal by the means. A marking material orientation detection processing unit that detects the orientation, a marking position detection processing unit that obtains the position of the marking image based on a change from the integrated value of the density projections of a specific range of the marking material for the original image signal, and the processing unit A marking image separation processing unit that separates the marking image into printing units based on the marking position information, and a binarization processing unit that independently binarizes each marking-specific image signal separated by the unit. An embossing mark detection device, comprising: 4. A marking image enhancement process for performing enhancement processing by moving an image by a plurality of predetermined directions and distances with respect to an original image signal applied to the marking material orientation detection processing unit and the marking position detection processing unit. The stamp detection device according to claim 3, wherein a portion is provided.