JP2000076513A - Stamp reader - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a stamp reader which is small-sized and has high read accuracy. SOLUTION: This stamp reader 10 is provided with an image inputting part 12 which inputs an image D1 including a coin and its background image D2, an image smoothing processing part 14 which smoothes the images D1 and D2, a threshold deciding part 16 for coin extraction which decides threshold D5 for coin extraction based on the smoothed background image D4, an image binarization processing part 18 which binarizes the smoothed image D3 including the coin based on the threshold D5 for coin extraction, a coin part image extracting part 20 that extracts a coin part from the image D1 based on the binarized image D6, a coin specifying part 22 that specifies a coin class D8 based on the image D6, a threshold deciding part 24 for characteristic extraction which decides threshold D9 for characteristic extraction based on the class D8 and a coin part characteristic extracting part 26 which extracts the characteristic of a coin part image D7 based on the threshold D9.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a marking reading device for reading a marking formed on a coin surface.

[0002]

2. Description of the Related Art As a stamp reading device for reading a stamp inscribed on the surface of coins such as coins and game coins,
For example, a coin identification device as disclosed in Japanese Patent Application Laid-Open No. 9-231432 is known. Such a coin identification device includes a line type optical sensor and a magnetic sensor, a first identification unit for detecting the outer diameter, material, and currency position of the coin, and an area sensor for reading an engraved mark on the surface of the coin. And two identification units. This coin identification device identifies the type of coin based on the outer diameter and the material of the coin detected by the first identification unit, and the second identification unit uses a threshold value according to the type of coin to identify the coin. Since the engraved mark on the surface is read, it is possible to accurately read the engraved mark on the surface of the coin regardless of the type of the target coin.

[0003]

However, the above-described stamp reading device (coin identification device) has the following problems. That is, first, in addition to the area sensor for reading the inscription imprinted on the surface of the coin, the engraving reading device includes a pre-processing device for identifying the type of the coin (the line type optical sensor, the magnetic sensor, etc.) It is necessary to provide the device, so that the device becomes large.

[0004] Second, even if the passing position of the coin is detected in the first discriminating section, the coin may be misaligned when being conveyed to the second discriminating section. In some cases, the exact position of the coin cannot be specified, and it may be difficult to accurately read the inscription imprinted on the surface of the coin.

SUMMARY OF THE INVENTION It is an object of the present invention to solve the above-mentioned problems and to provide a stamp reading device which can be formed in a small size and can accurately read a stamp inscribed on the surface of a coin.

[0006]

In order to solve the above-mentioned problems, an engraving reading apparatus according to the present invention comprises: input means for inputting a target image including coins and a background image not including coins; Smoothing means for smoothing each of the input target image and background image, and a first threshold value determination for determining a first threshold value based on the background image smoothed by the smoothing means Means, binarizing means for binarizing the target image smoothed by the smoothing means based on the first threshold value determined by the first threshold value determining means, and binarizing means A coin extraction unit that extracts a coin portion from the target image input by the input unit based on the target image binarized by the input unit; and a coin based on the target image binarized by the binarization unit. Type A target specifying means for constant, based on the type of the coin identified by the object specifying means, second
Second threshold value determining means for determining a threshold value of
Characteristic extracting means for extracting a characteristic image formed on a coin portion extracted by the target extracting means based on the second threshold value determined by the threshold value determining means. I have.

[0007] By determining a predetermined threshold value based on the background image and binarizing the target image based on the threshold value, the coin portion and the background portion in the target image can be efficiently separated. Therefore, the coin portion to be read can be extracted based on the binarized target image, and the position of the coin can be specified without providing a preprocessing device. Further, as described above, by determining the first threshold using the smoothed background image, the influence of local noise on the background image is reduced. Furthermore, since the type of coin is specified based on the binarized target image, the type of coin can be specified without providing a preprocessing device. In addition, a predetermined threshold value is determined based on the specified coin type, and a characteristic image formed in the coin portion is extracted based on the threshold value. It is possible to accurately read the engraved mark on the surface of the.

In the marking reading apparatus according to the present invention, the object specifying means is based on the number of 0 values or 1 values continuously arranged in a specific linear direction in the target image binarized by the binarizing means. It may be characterized in that the type of coin is specified by using

As described above, by specifying the type of coin based on the number of 0 values or 1 values continuously arranged in a specific linear direction, a plurality of types of coins having different diameters can be easily identified. It becomes possible.

[0010] The marking reading apparatus of the present invention is characterized in that the target specifying means specifies the type of coin based on the total number of 0 values or 1 values in the target image binarized by the binarizing means. Is also good.

As described above, by specifying the type of a coin based on the total number of 0 values or 1 values, the coins having substantially the same outer diameter and having a hollow portion inside and a coin having a hollow portion inside are provided. It is possible to efficiently discriminate between coins that do not (for example, a five-yen coin and a hundred-yen coin).

In the marking reading apparatus according to the present invention, the first threshold value determining means sets a maximum pixel value among the pixel values of the background image smoothed by the smoothing means as the first threshold value. The determination may be a feature.

The maximum pixel value among the pixel values of the background image smoothed by the smoothing means is determined as the first threshold value, so that the background image portion can be more efficiently processed by the binarization means. Removed.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A stamp reading device according to an embodiment of the present invention will be described with reference to the drawings. First, the configuration of the stamp reading device according to the present embodiment will be described.
FIG. 1 is a configuration diagram of the stamp reading device according to the present embodiment.

The marking reading device 10 includes an image input unit 12 as input means and an image smoothing processing unit 1 as smoothing means.
4. a coin extraction threshold determining unit 16 as a first threshold determining unit; an image binarization processing unit 18 as a binarizing unit;
A coin partial image extracting unit 20 as a target extracting unit, a coin specifying unit 22 as a target specifying unit, a threshold extracting unit 24 for feature extraction as a second threshold determining unit, and a coin as a characteristic extracting unit And a partial image feature extraction unit 26. Hereinafter, each component will be described in detail.

The image input section 12 is provided with an area sensor as an image pickup means (details will be described later), and a target image including a coin to be read (hereinafter referred to as a coin containing image).
And a background image that does not include coins, respectively, and can be stored in the memory. The image input unit 12 outputs the captured coin-containing image D1 and the background image D2 to the image smoothing processing unit 14, and outputs the coin-containing image D1 to the coin partial image extracting unit 20.

The image smoothing processing section 14 includes an image input section 12
And smoothing each of the coin-containing image D1 and the background image D2 output from. In this smoothing process, for example, the average of the pixel values of 9 adjacent pixels (3 vertical pixels × 3 horizontal pixels) in the original image (coin image D1 and background image D2) is calculated as the pixel value of the central pixel of the 9 pixels. This process is performed for the entire original image. The image smoothing processing unit 14 outputs the smoothed coin-containing image D3 to the image binarization processing unit 18, and outputs the smoothed background image D4 to the coin extraction threshold determining unit 16. Output.

The coin extraction threshold value determination unit 16 determines a coin extraction threshold value D5 (first threshold value) based on the smoothed background image D4 output from the smoothing processing unit 14.
Is determined, and this coin extraction threshold value D5 is output to the image binarization processing unit 18. Specifically, the maximum pixel value among the pixel values of the smoothed background image D4 output from the smoothing processing unit 14 is set to a coin extraction threshold D5.
To be determined.

The image binarization processing unit 18 generates a coin extraction threshold value D5 output from the coin extraction threshold value determination unit 16.
, The smoothed coin-containing image D3 output from the image smoothing processing unit 14 is binarized. Specifically, for each pixel of the smoothed coin-containing coin image D3, when the pixel value of the pixel is larger than the coin extraction threshold D5,
When the pixel value is set to 1 and the pixel value of the pixel is equal to or smaller than the coin extraction threshold value D5, a binarized image D6 having the pixel value set to 0 is formed. The image binarization processing unit 18 also
The binarized image D6 is output to the coin partial image extracting unit 22 and the coin specifying unit 22.

The coin portion image extracting section 20 extracts a coin portion in the coin containing image D1 output from the image input section 12, based on the binarized image D6 output from the image binarization processing section 18. . Specifically, for each pixel, a product of the pixel value of the coin-containing image D1 and the pixel value of the binarized image D6 is obtained, and the pixel value of the pixel is used to form a coin partial image D7. The coin partial image extracting unit 20 also outputs the coin partial image D7 to the coin partial image feature extracting unit 26.

The coin specifying section 22 is provided with an image binarization processing section 18.
The type of coin is specified based on the binarized image D6 output from. Specifically, the pixel values of the binary image are scanned in the horizontal direction for each row. At this time, the number of one values continuously arranged in the horizontal direction for each row is counted. This scanning is performed for all rows of the binarized image D6, and the maximum value of the number of continuous one-valued data is set as the outer diameter of the coin. The type of the coin is specified by comparing the outer diameter of the coin obtained as described above with the outer diameters of a plurality of types of coins input in advance. The coin specifying unit 22 also outputs the specified coin type D8 to the feature extraction threshold value determining unit 24.

The feature extraction threshold value determination section 24 determines a feature extraction threshold value D9 (second threshold value) based on the coin type D8 output from the coin identification section 22. Specifically, based on the coin type D8, a threshold value for feature extraction D9 is referred to by referring to a threshold table for feature extraction describing a predetermined threshold value for each coin.
To determine. Table 1 shows an example of a threshold table for feature extraction.
Shown in

[0023]

[Table 1] The threshold values in Table 1 are examples in the case where each pixel value is expressed in 256 gradations. Feature extraction threshold determination unit 2
4 also outputs the feature extraction threshold value D9 to the coin partial image feature extraction unit 26.

The coin partial image feature extracting section 26 outputs a coin partial image D7 output from the coin partial image extracting section 20 based on the feature extracting threshold value D9 output from the feature extracting threshold value determining section 24. Is extracted. In particular,
When the pixel value of the pixel of the coin partial image D7 is larger than the feature extraction threshold value D9, the pixel value is set to 1 and the pixel value of the pixel is set to the feature extraction threshold value D9.
In the case of 9 or less, a characteristic image having the pixel value of 0 is formed. The feature image is a feature image corresponding to a mark, such as a figure, a pattern, a number, or a character, carved on the surface of the coin. The coin partial image feature extraction unit 26 also outputs the feature image formed as described above to the outside.

Next, an example of a device suitably used as the image input unit 12 of the stamp reading apparatus 10 according to the present embodiment will be described. FIG. 2 is a perspective view showing the configuration of the image input unit 12. The image input unit 12 is provided above the imaging unit 28 having a rectangular parallelepiped shape, the plane plates 30 and 32 attached to two opposing sides of the upper surface of the imaging unit 28, and the imaging unit 28 and the plane plates 30 and 32, respectively. And a belt conveyor 34. The height of the upper surface of the imaging unit 28 and the height of the upper surfaces of the flat plates 30 and 32 coincide with each other, and a single plane is formed as a whole, forming a transport path B for transporting coins A. The transport path B and the belt conveyor 34
The lower surface of the coin is provided with a gap substantially equal to the thickness of the coin A,
By driving the belt conveyor 34, this gap is
Is transported.

As shown in FIG. 3, the imaging unit 28
A cubic case 36 having an open upper surface and a glass upper plate 38 covering the upper surface opening are provided.
Inside, an area sensor 40 serving as an imaging unit and a coin A
LED 42 for detecting the passage of the light, and LED 44 as illumination means are provided. The area sensor 40 is attached to a fixed base 46 in the center of the housing 36,
The surface of the coin A passing through the imaging region C of No. 8 is imaged.

The LED 42 is attached to a fixed plate 48 protruding from the inner side surface of the housing 36 and irradiates the upper plate 38 with light. Further, the LEDs 44 are arranged in a ring shape along the inner surface of the upper plate 38, and irradiate the coin A passing on the upper plate 38 with light for a short time.

The belt of the belt conveyor 34 shown in FIG. 2 is provided at the center of the conveying path B, and light emitted from a pair of LEDs 42 disposed on both sides of the belt is transmitted to the upper plate 38 and the belt conveyor 34. It travels upwards past both sides of the belt. A pair of PDs 50 is disposed on the optical path of the irradiation light above the belt conveyor 34,
Light that has passed through both sides of the belt of the belt conveyor 34 enters the PDs 50 facing each other. Based on the detection signal of the PD 50, it is detected whether or not the coin A has reached the imaging area C.

Next, the operation of the stamp reading device according to this embodiment will be described. FIG. 4 is a flowchart illustrating a flow of a process when reading a stamp engraved on the surface of a coin using the stamp reading device 10 according to the present embodiment. The description will be made in the following order.

The marking reading device 10 according to the present embodiment
When reading the engraved mark on the surface of the coin using
First, a background image is input by the image input unit 12 (S
1). The input background image D2 is, as shown in FIG.
A low-luminance, random background light image is obtained. Note that the input or processed image is a two-dimensional image, but in FIG. 5 and subsequent figures, it is represented by a one-dimensional image for convenience of explanation.

The background image D2 input by the image input unit 12 is smoothed by the image smoothing processing unit 14 (S2), and the smoothed background image D4 is as shown in FIG.

Thereafter, the coin identifying threshold value extracting section 16 determines a coin identifying threshold value D5 (S3). Specifically, the largest pixel value (point M in FIG. 6) among the pixel values of the smoothed background image D4 is determined as the coin extraction threshold D5.

By determining the maximum pixel value among the pixel values of the background image D4 smoothed as described above as the coin extraction threshold value D5, the influence of the background image can be effectively removed. It becomes possible. In particular, coin extraction threshold D
By using the smoothed background image D4 instead of the input background image D2 when determining 5, it is possible to remove the influence of local noise in determining the coin extraction threshold D5. Specifically, for example, the image input unit 1
2 cannot be input to a particular pixel of the area sensor, and the output value of that pixel is always at a high level (256
In this case, when the threshold value D5 for coin extraction is determined based on the input background image D2, the threshold value becomes 256 values / 256 gradations.
It becomes impossible to extract the coin portion to be read. On the other hand, if the coin extraction threshold value D5 is determined based on the smoothed background image D4, the influence of local noise is reduced by the averaging process between adjacent pixels, and the The threshold can be determined.

Subsequently, a coin-containing image is input by the image input section 12 (S4). The input coin image D1 is
As shown in FIG. 7, the image has a high luminance distribution in the coin portion.

The coin containing image D1 input by the image input unit 12 is smoothed by the image smoothing processing unit 14 (S5), and the smoothed coin containing image D3 is as shown in FIG.

Subsequently, the image binarization processing unit 18
The smoothed coin-containing image D3 output from the image smoothing processing unit 14 is binarized (S6). Specifically, FIG.
As shown in FIG. 9 and FIG. 9, when the pixel value of the pixel of the smoothed coin-containing image D3 is larger than the coin extraction threshold D5, the pixel value is set to 1 and the pixel of the pixel is If the value is equal to or smaller than the coin extraction threshold D5, a binarized image D6 with the pixel value set to 0 is formed. The binarized image D6 indicates the position of the coin to be read in the coin-containing image D1. By forming the binarized image D6, the binarized image D6 can be read without providing a separate preprocessing device. It is possible to specify the position of the coin to be processed.

After that, the binarized image D output from the image binarizing section 18 is output by the coin partial image extracting section 20.
6, a coin portion in the coin-containing image D1 output from the image input unit 12 is extracted. Specifically, for each pixel, the pixel value of the coin-containing image D1 and the binarized image D6
Is calculated, and the product is used as the pixel value of the pixel to form a coin partial image D7. That is, as shown in FIG. 10, the coin part image D7 has the pixel value of the coin containing image D1 input by the image input unit 12 for the pixel of the coin part, and the zero value for the pixels other than the coin part. Is obtained. Here, by setting the pixel values of the pixels other than the coin part to 0 values, it is possible to omit the processing for the pixels other than the coin part in the subsequent processing, and it is possible to shorten the reading time.

Subsequently, the coin type D8 is specified by the coin specifying unit 22 (S8). Specifically, the pixel values of the binarized image are scanned in the horizontal direction for each row, and the number of one values continuously arranged in the horizontal direction for each row is counted.
This scanning is performed for all rows of the binarized image D6,
The maximum value of the number of consecutive one-values is the outer diameter of the coin. The coin type D8 is specified by comparing this outer diameter with the outer diameter of the coin input in advance.

When the coin type D8 is specified, the feature extraction threshold value determining section 24 determines the feature extraction threshold value D9.
Is determined (S9). More specifically, as described above, the feature extraction threshold value D9 is referred to by referring to a feature extraction threshold value table describing a predetermined threshold value for each coin based on the coin type D8. Is determined.

Subsequently, based on the feature extraction threshold value D9 output from the feature extraction threshold value determination unit 24, the coin output from the coin partial image extraction unit 20 is performed by the coin partial image feature extraction unit 26. The feature of the partial image D7 is extracted (S10). Specifically, as shown in FIGS. 10 and 11, for each pixel of the coin partial image D7, when the pixel value of the pixel is larger than the feature extraction threshold D9, the pixel value is set to 1, and When the pixel value of the pixel is equal to or smaller than the feature extraction threshold value D9, a feature image O1 having the pixel value of 0 is formed. Such a characteristic image O1 corresponds to an inscription engraved on the surface of a coin. Here, since the feature extraction threshold value D9 is determined for each coin to be read, it is possible to accurately read the inscription imprinted on the surface of the coin regardless of the type of coin.

Further, in the case of reading an engraved mark on the surface of another coin, the above processing may be repeated. In this case, when the above processing is performed in the same reading environment, the processing of inputting the background image (S1), smoothing the background image (S2), and determining the threshold value for coin extraction (S3) are omitted. be able to. That is, as the coin extraction threshold value used in binarizing the coin-containing image (S6), the coin extraction threshold value D which has already been obtained is used.
5 may be used.

Next, the effect of the stamp reading device according to the present embodiment will be described. The marking reading device 10 according to the present embodiment is
By extracting the coin portion based on the binarized image D6, the position of the coin to be read can be specified without using a preprocessing device. Also,
Since the coin type D8 is specified based on the binarized image D6, the coin type can be specified without using a pretreatment device. As a result, it is not necessary to provide a preprocessing device for specifying the coin position and the coin type, and the size of the stamp reading device 10 can be reduced.

Further, the stamp reading apparatus 10 according to the present embodiment specifies the position of the coin by the pre-processing device, and thereafter conveys the coin to the area provided with the area sensor, and then captures the target image including the coin. Instead, the coin portion in the coin-containing image D1 is extracted by the coin portion image extracting unit 20, so that the problem of the displacement of the coin to be read does not occur. As a result, it is possible to read the mark imprinted on the surface of the coin to be read very accurately.

Further, the stamp reading apparatus 10 according to the present embodiment determines the coin extraction threshold value D5 using the background image D4 smoothed by the image smoothing processing unit 14, so that the coin reading threshold value D5 in the background image is determined. The effect of local noise is reduced. As a result, it is possible to accurately read the engraved mark on the surface of the coin to be read without being affected by local noise of the background image.

In addition, the marking reading apparatus 10 according to the present embodiment uses the feature image O1 based on the feature extraction threshold value D9 determined by the feature extraction threshold value determination unit 24.
Is formed, it is possible to accurately read the engraved mark on the surface of the coin regardless of the type of the coin.

The engraving reading apparatus 10 according to the present embodiment sets the maximum pixel value among the pixel values of the background image D4 smoothed by the image smoothing processing unit 14 as a coin extraction threshold value D5. With the determination, the background image portion is efficiently removed by the image binarization processing unit 18.
As a result, erroneous detection or the like due to background light is prevented, and reading accuracy is improved.

Further, in the stamp reading device 10 according to the present embodiment, the coin specifying unit 22 specifies the coin type D8 based on the number of one values continuously arranged in the horizontal direction. As a result, a plurality of types of coins having different diameters can be easily identified, and the reading speed is improved.

The marking reading device 1 according to the above embodiment
At 0, the coin specifying unit 22 specifies the coin type D8 based on the number of 1 values arranged continuously in the horizontal direction, but this is based on the number of 0 values arranged continuously. The coin type D8 may be specified. The scanning direction may be a vertical direction.

Furthermore, in the stamp reading apparatus 10 according to the above embodiment, the coin specifying unit 22 specifies the coin type D8 based on the number of one values arranged continuously in the horizontal direction. , By the image binarization processing unit 18
The coin type D8 may be specified based on the total number of 0 values or 1 values in the binarized image D6.

As described above, by specifying the coin type D8 based on the total number of 0 values or 1 values, a coin having substantially the same diameter and having a hollow portion inside and a coin having no hollow portion inside is provided. For example, a five-yen coin and a hundred-yen coin can be efficiently distinguished.

In the stamp reading apparatus 10 according to the above-described embodiment, coins are read, but other types of coins such as game coins may be used.

[0052]

According to the stamp reading apparatus of the present invention, the coin portion is extracted based on the target image binarized by the binarizing means, so that the coin to be read can be read without using a preprocessing apparatus. The position can be specified. Further, since the type of the coin is specified based on the binarized target image, the type of the coin can be specified without using the pretreatment device. As a result, it is not necessary to provide a preprocessing device for specifying the position of the coin and for specifying the type of the coin, and the size of the stamp reading device can be reduced.

Further, in the stamp reading device of the present invention, the position of the coin is specified by the pre-processing device, and then the coin is conveyed to the portion provided with the input means, and then the target image including the coin is captured. Instead, since the coin portion is extracted from the target image by the target extracting means, the problem of the positional deviation of the coin to be read does not occur. As a result, it is possible to read the inscription imprinted on the surface of the coin to be read extremely accurately.

Further, in the marking reading device of the present invention, the first threshold value is determined using the background image smoothed by the smoothing means, so that the influence of local noise on the background image is reduced. Is done. As a result, it is possible to accurately read the engraved mark on the surface of the coin to be read without being affected by local noise of the background image.

Further, the stamp reading device of the present invention has a
By extracting the characteristic image based on the second threshold value determined by the threshold value determining means, it is possible to accurately read the inscription imprinted on the coin surface regardless of the type of the coin. Become.

Further, in the marking reading apparatus of the present invention, the target specifying means specifies the type of coin based on the number of one values continuously arranged in the horizontal direction. As a result, a plurality of types of coins having different diameters can be easily identified, and the reading speed is improved.

Further, the stamp reading apparatus of the present invention specifies the type of coin based on the total number of 0 values or 1 values, so that a coin having substantially the same diameter and having a hollow portion inside and a hollow portion inside is used. It is possible to efficiently identify a coin having no portion.

Further, the marking reading device of the present invention determines the maximum pixel value among the pixel values of the background image smoothed by the smoothing means as the first threshold value, thereby performing the binarization. By this means, the background image portion is efficiently removed. As a result, erroneous detection or the like due to background light is prevented,
The reading accuracy is improved.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a stamp reading device according to an embodiment of the present invention.

FIG. 2 is a perspective view of an image input unit.

FIG. 3 is a perspective view of an imaging unit.

FIG. 4 is a flowchart showing a flow of processing when reading the surface of a coin using the stamp reading device according to the embodiment of the present invention.

FIG. 5 is a diagram showing a background image D2.

FIG. 6 is a diagram showing a smoothed background image D4 and a coin extraction threshold D5.

FIG. 7 is a diagram showing a coin containing image D1.

FIG. 8 is a diagram showing a smoothed coin-containing coin image D3 and a coin extraction threshold D5.

FIG. 9 is a diagram showing a binarized image D6.

FIG. 10 shows a coin partial image D7 and a threshold value D for feature extraction.
FIG.

FIG. 11 is a diagram showing a characteristic image O1.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 10 ... Engraving reading device, 12 ... Image input part, 14 ... Image smoothing processing part, 16 ... Threshold determination part for coin extraction, 1
8 image binarization processing unit, 20 coin part image extraction unit, 2
2 ... Coin identification unit, 24 ... Threshold extraction unit for feature extraction, 2
6 ... Coin partial image feature extraction unit

Claims (4)

[Claims] 1. A stamp reading device for reading a stamp inscribed on a surface of a coin, comprising: an input unit for inputting a target image including the coin and a background image not including the coin; A smoothing means for smoothing each of the target image and the background image, and a first threshold for determining a first threshold value based on the background image smoothed by the smoothing means. Value determination means, and binarization means for binarizing the target image smoothed by the smoothing means based on the first threshold value determined by the first threshold value determination means; A target extraction unit that extracts the coin portion from the target image input by the input unit based on the target image binarized by the binarization unit; A target specifying unit that specifies the type of the coin based on the binarized target image, and a second threshold value that determines a second threshold based on the type of the coin specified by the target specifying unit. And a characteristic image formed on a portion of the coin extracted by the target extracting means based on a second threshold value determining means and a second threshold value determined by the second threshold value determining means. And a feature extracting means for extracting a mark. 2. The method according to claim 1, wherein the target specifying unit is configured to determine the value of the coin based on the number of 0 values or 1 values continuously arranged in a specific linear direction in the target image binarized by the binarizing unit. The stamp reading device according to claim 1, wherein the type is specified. 3. The coin identification method according to claim 2, wherein the target identification unit identifies the type of the coin based on the total number of 0 values or 1 values in the target image binarized by the binarization unit. 3. The marking reading device according to 1 or 2. 4. The first threshold value determining means determines a maximum pixel value among pixel values of the background image smoothed by the smoothing means as the first threshold value. 4. The method according to claim 1, wherein
An engraving reading device according to the item.