JPH0676107A - Signal processing method for photographed image - Google Patents

Abstract

PURPOSE:To provide a signal processing method for the photographed images which can improve the recognizing accuracy of the characters printed on a subject. CONSTITUTION:The end face of a subject is photographed with the spectral sensitivity suitable to the character spectral characteristic and a character image information signal is obtained. Meanwhile the end face of the subject is photographed with the spectral sensitivity suitable to the spectral characteristic of the background except the characters and a background image information signal is obtained. Then the background image information signal is subtracted from the character image information signal and an image recognition signal is obtained. In other words, the signal having a distribution that includes the small value of the character value and the large value of the background part is subtracted from the signal having a distribution that includes the large value of the character part and the small value of the packground value. Thus it is possible to obtain an image recognition signal that has an extremely big difference between the values of the character and backgound parts. As a result, the recognizing accuracy of the character image information can be improved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal processing method of a photographed image in recognizing a marking character for identifying a steel material.

[0002]

2. Description of the Related Art Conventionally, a character string printed on the surface of an object is photographed by a camera, the character string portion of the image is binarized and extracted, and each character is discriminated from the binarized data. There is a character reading device that reads the contents of a character string by doing so. As an example of such a device, JP-A-2-15348
In JP-A-8, by enclosing the front and rear of a printed character string with specific marks so that the start, end, length, extension direction, etc. of the character string can be easily recognized, the versatility and reliability of character reading are improved. A character reading device having improved property is disclosed.

On the other hand, conventionally semi-finished steel products such as slabs, blooms, billets, etc., which have been continuously cast, are temporarily stored as a buffer before moving to a rolling process, and for confirmation of the actual product when rolling on the end face of the steel product. , For example, marking a character composed of a plurality of numbers.

Since the above-mentioned characters are printed by a paint spraying type printing device, the character quality such as thickness, density and positional deviation is unstable, the characters may be adjacent to each other, or the end surface of the read steel material may be read. There is a possibility that it may be difficult to recognize the character portion and the background portion other than the character portion due to so-called uneven illumination of the illumination radiated on the background.

Therefore, in Japanese Unexamined Patent Publication (Kokai) No. 62-204665, the background density data obtained by previously reading the ground portion of the end surface of the steel material and the reciprocal value of the value corresponding to this density data are written in the memory. Then, the background density data is subtracted from the digital density signal obtained by reading the characters printed on the end surface, and then the reciprocal value is multiplied to obtain the density reading error caused by the uneven illumination on the end surface. An image input device that performs correction is disclosed.

However, when an image of an end face of a steel material is input using this apparatus, when there is only one image pickup camera as an image input means, for example, when the spectral characteristics of characters and background are remarkably different from each other. However, depending on the spectral sensitivity of the camera, it is not possible to clearly capture the characters and the background, and the recognition accuracy must depend on the spectral sensitivity of the imaging camera. was there.

In view of the problems of the prior art as described above, a main object of the present invention is to provide a signal processing method of a photographed image capable of improving the recognition accuracy of characters printed on an object. is there.

[0008]

According to the present invention, the above object is a signal processing method of a photographed image for obtaining an image recognition signal for recognizing a character printed on an end surface of an object. The process of obtaining the character image information signal by photographing the end face of the object with the spectral sensitivity suitable for the spectral characteristic of the character, and the same end face with the spectral sensitivity suitable for the spectral characteristic of the background other than the character of the end face. And a step of obtaining a background image information signal and a step of subtracting the background image information signal from the character image information signal to obtain an image recognition signal. It is achieved by

[0009]

In this way, the character image information signal obtained by photographing with the spectral sensitivity suitable for the spectral characteristic of the character is
It is possible to obtain a signal having a distribution in which the value of the image information signal in the portion where the character is present is high and the value of the image information signal in the other background portion is low, and the spectral sensitivity suitable for the spectral characteristic of the background is obtained. In the background image information signal obtained by photographing with, it is possible to obtain a signal having a distribution in which the value of the portion where characters are present is low and the value of the background portion is high. Since the background image information signal is subtracted from the character image information signal, it is possible to obtain an image information signal in which the signal of the character portion is emphasized and the signal component of the background portion is almost absent.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will now be described in detail with reference to the accompanying drawings.

FIG. 1 is a schematic configuration diagram showing an entire character automatic recognition apparatus to which the present invention is applied. The two imaging cameras 1 and 2 for photographing the end surface of the target object are arranged at positions facing the end surface 4a of the steel material 4 which is appropriately transported on the transport rail 3. These imaging cameras 1 and 2 have X-axis (left and right),
The traversers 5 and 6 are supported so as to be movable in two Y-axis (front and rear) directions. These imaging cameras 1,
A lighting device 7 is provided between the steel material 4 and the steel material 4.

On the end surface 4a of the steel material 4, a character string 9 consisting of 7 characters 8 matrix characters each having 5 dots horizontally and 7 dots vertically is previously printed by a paint spraying type printing device. This character string 9 is the number of characters, the character spacing,
The character shape, the character printing direction, and the character size are substantially the same, but the thickness and shade of the character 8 differ depending on the maintenance state of the paint device. The printing position of each character 8 is not always constant from the left, right, upper and lower ends of the steel material, but differs depending on the type of steel material.

On the other hand, as shown in FIGS. 3 and 4, the image pickup camera 1 has a spectral characteristic of the character 8, that is, a spectral sensitivity suitable for the peak wavelength Tλ _P of the color of the character 8. Therefore, as shown in FIG. 5, the image signal 11 of the steel material end surface 4a (this signal is referred to as a character image information signal) captured by the camera 1 has a high voltage value in the portion where the character 8 is printed. , The voltage value of the background portion 10 (the base portion of the end surface 4a of the steel material 4) other than the character 8 is low.

On the other hand, the image pickup camera 2 photographs the same portion of the end face 4a photographed by the image pickup camera 1, and at the same time, the spectral characteristic of the background 10, that is, the spectrum suitable for the peak wavelength Bλ _P of the color of the background 10. It has sensitivity. Therefore, the image signal 12 of the steel material end surface 4a photographed by the camera 2 (this signal is referred to as a background image information signal)
Contrary to the character image information signal 11 described above, indicates that the voltage value of the portion where the character 8 is printed is low and the voltage value of the background portion 10 is high.

The image pickup cameras 1 and 2 are electrically connected to an image processing device 13 for analyzing the image information signals 11 and 12, and the image information signals 11 and 12 are internally stored in an image memory (not shown). No.), and processing based on a predetermined image processing algorithm is performed so that the steel material printing end face position measurement, character reading processing, and the like are performed. The image processing apparatus 13 is a computer 14
Is connected to, and serves as a supply source of signals for driving the traversers 5 and 6, and indicates the illuminance of the imaging cameras 1 and 2 and the illumination device 7 and the process control computer 1
5 exchanges data with 5, and controls the image recognition apparatus as a whole.

A signal processing method of the above-mentioned image information signals 11 and 12 in the image recognition apparatus thus configured will be described with reference to the flowchart of FIG.

First, when the steel material 4 is conveyed in step 1, the illuminating device 7 is turned on and moved to the front of the end surface 4a of the steel material 4 so that the respective imaging cameras 1 and 2 can photograph the same place. Next, in step 2, the steel material end surface 4a is photographed by the imaging camera 1 and the character image information signal 11 shown in FIG. 5 is output to the image processing device 13. Similarly, in step 3, the steel material end surface 4a is photographed by the imaging camera 2 and the background image information signal 12 is output to the image processing device 13. Then, in step 4, the background image information signal 12 is subtracted from the character image information signal 11 to calculate the image recognition signal 16.

The image recognition signal 16 obtained by the signal processing method of the present invention has the character 8 as shown in FIG.
Only the image signal of the portion is detected as a positive component, and the boundary between the character 3 portion and the background 10 is clearly distinguished, and as a result, an image signal in which only the information of the character 8 exists is obtained. Therefore, the sensitivity for detecting the character 8 is dramatically improved. Further, even when so-called uneven illumination occurs on the characters 8 located at both ends of the steel material end surface 4a in the horizontal direction due to the lighting device 7, the respective image information signals 11 and 12 obtained by photographing the same portion are subtracted. As a result, the noises caused by the uneven illumination are canceled out with each other, so that noises are not superimposed on the processed image recognition signal 16.

In this embodiment, each image information signal 1
Two image pickup cameras 1 and 2 were used as the input means of 1 and 12, but as shown in FIG. 6, the transmission wavelength range is suitable for the spectral characteristics of the character 8 and the background 10 on the front surface of the lens of the image pickup camera 20, respectively. By arranging the two filters 21a and 21b in a replaceable manner, the same image recognition signal 16 as in the present embodiment can be obtained by one image pickup camera, so that the effect is great.

[0020]

As is apparent from the above description, according to the signal processing method of the photographed image according to the present invention, the character image information signal obtained by photographing with the spectral sensitivity suitable for the spectral characteristic of the character is obtained. , The background image information signal obtained by photographing with the spectral sensitivity suitable for the spectral characteristic of the background is subtracted, so that one of the image information signals has a high resolution of the image signal of the portion corresponding to the character. On the other hand, in the image information signal, since the image signal of the portion corresponding to the background has a high resolution, it is possible to complement each other for capturing images with different spectral sensitivities, which is a weak point of each image information signal. It is possible to cancel out the generation of noises due to each other. Therefore, a highly reliable image recognition signal can be obtained, and when recognizing a character printed on an object, the recognition accuracy can be dramatically improved.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing an entire automatic character recognition device to which the present invention is applied.

FIG. 2 is a flowchart illustrating a method of processing a captured image signal according to the present invention.

FIG. 3 is a diagram illustrating spectral characteristics of an object.

FIG. 4 is a diagram showing the spectral sensitivity of an imaging camera.

FIG. 5 is a waveform diagram illustrating a signal processing method of a captured image according to the present invention.

FIG. 6 is a diagram partially showing another embodiment of the present invention.

[Explanation of symbols]

 1, 2, 20 Imaging camera 3 Conveying rail 4 Steel 4a End face 5, 6 Traverser 7 Lighting device 8 Character 9 Character string 10 Background part 11 Character image information signal 12 Background image information signal 13 Image processing device 14 Computer 15 Process control computer 16 image recognition signal 21a, 21b filter

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Tetsuo Nakamura 46-59 Nakahara, Tobata-ku, Kitakyushu City Nippon Steel Corporation Machinery & Plant Division

Claims (1)

[Claims] 1. A signal processing method of a photographed image for obtaining an image recognition signal for recognizing a character printed on an end surface of an object, wherein an end surface of the object is spectroscopically adapted to a spectral characteristic of the character. A step of obtaining a character image information signal by photographing with sensitivity, a step of obtaining the background image information signal by photographing the same end face with a spectral sensitivity suitable for a spectral characteristic of a background other than characters on the end face, and the character image And a step of obtaining the image recognition signal by subtracting the background image information signal from the information signal.