TWI657412B - Coin classification method and device thereof - Google Patents

Abstract

A coin classification method includes: binarizing an image containing a coin image; locating a circle center of the coin image in the binarized image; and determining whether the bright area is bright if the binarized image includes a plurality of bright areas Containing the first characteristic bright area and the second characteristic bright area, the distance from the centroid of the first characteristic bright area to the circle center of the coin image falls within a first predetermined distance, and the centroid of the second characteristic bright area to the coin map The distance of the center of the image falls within the second predetermined distance range; if the bright area includes the first feature bright area and the second feature bright area, it is determined that the coin image is a new version of the NT $ 10 coin image.

Description

Coin sorting method and device

The present disclosure relates to a method and device for sorting coins.

To address the prevalence of counterfeit coins in the old version of the NT $ 10 coin, the government introduced a new version of the NT $ 10 coin with new anti-counterfeiting technology to replace the old version of the coin. In order to collect and eliminate the old NT $ 10 coin, a coin classification method that can distinguish the old and new NT $ 10 coins is needed.

The existing coin classification methods generally classify coins of different sizes, let the coins pass through a track, and the track has a plurality of openings, arranged in order from small to large, and the coins will fall into the openings corresponding to their sizes and enter the corresponding collection slots. . Considering that there are no obvious differences in the size of the new and old NT $ 10 coins, the conventional coin classification method obviously cannot be used to distinguish between the old and new NT $ 10 coins. In addition, manual classification of coins would consume more labor and time costs.

Therefore, how to propose a coin classification method that can solve the above-mentioned problems is one of the directions that are urgently needed.

In view of this, one of the objectives of this disclosure is to propose a coin classification method and device that can accurately and quickly discriminate new and old versions of NT $ 10 coins.

According to some embodiments of the present disclosure, a coin classification method includes: binarizing an image including a coin image; locating a circle center of the coin image in the binarized image; if the binarized image includes a plurality of Bright area, it is judged whether the bright area includes the first feature bright area and the second feature bright area, wherein the distance from the center of the first feature bright area to the center of the circle falls within the first predetermined distance range, and the second feature bright area The distance from the centroid to the center of the circle falls within the second predetermined distance range; and if the bright area includes the first feature bright area and the second feature bright area, it is determined that the coin image is a new version of the NT $ 10 coin image.

In one or more embodiments of the present disclosure, the coin classification method further includes performing noise removal on the binarized image.

In one or more embodiments of the present disclosure, the centroid of the first characteristic bright area, the centroid and center of the circle of the second characteristic bright area are arranged substantially along a straight line.

In one or more embodiments of the present disclosure, the number of bright spots in the first characteristic bright area falls within the first predetermined number of bright spots, and the number of bright spots in the second characteristic bright area falls within the second predetermined number of bright spots.

In one or more embodiments of the present disclosure, the aspect ratio of the smallest rectangle surrounding the first feature bright area falls within the first predetermined aspect ratio range, and the aspect ratio of the smallest rectangle surrounding the second feature bright area It falls within the range of the second predetermined aspect ratio.

In one or more embodiments of the present disclosure, the aforementioned positioning circle center The steps include: projecting the binarized image to a direction to generate an image histogram, wherein the first axis of the image histogram is the coordinates of the direction, and the second axis of the image histogram is the number of bright points; and if The image histogram includes a plurality of consecutive bright spot intervals, and the center of the largest range in the first axial direction in the consecutive bright spot intervals is determined as the coordinate of the circle center of the coin image in that direction.

In one or more embodiments of the present disclosure, the aforementioned step of locating the center of the circle further includes: determining whether the image histogram includes adjacent first consecutive bright point intervals and second consecutive bright point intervals, wherein the first consecutive bright point intervals include adjacent first The first endpoint of two consecutive bright spot intervals, and the second consecutive bright spot interval includes a second endpoint adjacent to the first consecutive bright spot interval, and the distance between the first endpoint and the second endpoint is less than a threshold; and The graph includes a first continuous bright spot interval and a second continuous bright spot interval. Then, the combination of the first continuous bright spot interval and the second continuous bright spot interval is regarded as one of the continuous bright spot intervals.

In one or more embodiments of the present disclosure, the coin classification method further includes performing contour detection on the binarized image before projecting the binarized image to the direction.

According to some embodiments of the present disclosure, a coin sorting device includes at least one camera and a processor. The camera is configured to capture at least one image of the coin, so that the image includes a coin image. The processor is electrically connected to the camera and configured to binarize the image. The processor is also configured to locate the center of the circle of the coin image in the binarized image. If the binarized image includes a plurality of bright areas, the processor is further configured to determine whether the bright areas include a first characteristic bright area And a second characteristic bright area, wherein the distance from the centroid to the circle center of the first characteristic bright area falls within a first predetermined distance range, and the distance from the centroid to the circle center of the second characteristic bright area falls within a second predetermined distance range . If the bright area includes the first feature bright area and the second feature bright area, the processor is further configured to determine that the coin is a new version of the NT $ 10 coin.

In one or more embodiments of the present disclosure, the coin sorting device further includes a track, and the camera is configured to photograph a coin sliding on the surface of the track.

In one or more embodiments of the present disclosure, the camera is located on a side of the surface facing the track.

In one or more embodiments of the present disclosure, the track has a transparent area, and the at least one camera includes a first camera and a second camera. The first camera is on the side of the surface facing the track, and the second camera is on the side of the surface facing away from the track, and is configured to capture coins through the transparent area.

In one or more embodiments of the present disclosure, the coin sorting device further includes a striker and a collection area. The striker is electrically connected to the processor. When the processor determines that the coin is a new version of the NT $ 10 coin, the striker pushes the coin into the collection area.

In one or more embodiments of the present disclosure, the coin sorting device further includes a roller rotatably disposed above the surface of the track, wherein the minimum distance between the rolling surface of the roller and the surface of the track is smaller than the thickness of the coin.

In one or more embodiments of the present disclosure, the track includes a plurality of balls that are rotatably embedded on the surface of the track.

In summary, the coin classification method disclosed in the present disclosure uses the new version of the NT $ 10 coin with the characteristics of two reflective areas to binarize the image containing the coin image. If the binarized image has a size that meets the bright area, If the two characteristic bright areas of the shape, position, and other conditions, the coin image is determined to be a new version of the NT $ 10 coin image.

100‧‧‧ Coin Classification Method

110, 120, 130, 140‧‧‧ steps

200a, 200b, 200c, 200d, 500a, 500b

210a, 210b‧‧‧ coin images

211a‧‧‧First reflective zone

211b‧‧‧The first characteristic bright area

212a‧‧‧Second reflective zone

212b‧‧‧Second characteristic bright area

215b‧‧‧center

300, 600, 700 ‧‧‧ image histograms

310, 610, 710‧‧‧‧ Maximum continuous highlight interval

311, 611, 711‧‧‧ Center

400‧‧‧coin sorting device

410‧‧‧first camera

420‧‧‧Second camera

430‧‧‧ track

431‧‧‧ surface

432‧‧‧Transparent zone

433‧‧‧ball

440‧‧‧Strike

450‧‧‧ Collection Area

460‧‧‧roller

461‧‧‧Rolling surface

470‧‧‧ Launching Agency

471‧‧‧Rotating disk

472‧‧‧ bezel

480‧‧‧ processor

490‧‧‧ coins

510a, 510b‧‧‧ Full Coin Image

520a, 520b ‧‧‧ partial coin image

530b‧‧‧Exclusion

FIG. 1 is a flowchart illustrating a coin sorting method according to an embodiment of the present disclosure.

Figure 2A shows an image containing a coin image.

Figure 2B is a binary image showing the image of Figure 2A.

Figure 2C is an image showing the contour detection of the image in Figure 2B.

Figure 2D is a noise-removed image showing the image in Figure 2B.

Fig. 3 is an image histogram showing the image of Fig. 2B.

FIG. 4 is a combination diagram illustrating a coin sorting device according to an embodiment of the present disclosure.

FIG. 5A shows an image including a complete coin image and a partial coin image.

Figure 5B is a binary image showing the image of Figure 5A.

Fig. 6 is an image histogram showing the image of Fig. 5B.

Figure 7 is another image histogram showing the image of Figure 5B.

In order to make the description of this disclosure more detailed and complete, reference may be made to the accompanying drawings and various embodiments described below. Elements in the figures are not drawn to scale and are provided merely to illustrate the present disclosure. Many practical details are described below to provide a comprehensive understanding of the disclosure. However, those of ordinary skill in the relevant arts should understand that the disclosure can be implemented without one or more practical details. Therefore, these details are not Apply to define this disclosure.

Please refer to FIG. 1, which is a flowchart illustrating a coin sorting method 100 according to an embodiment of the present disclosure. The coin sorting method 100 includes steps 110, 120, 130, and 140. Each step is explained in detail below.

Please refer to FIG. 2A, which shows an image 200a including a coin image 210a. In the image 200a, the coin image 210a is an image on the back of the new version of the NT $ 10 coin, which has a first reflective area 211a and a second reflective area 212a, which are not unique to the old version of the NT $ 10 coin. The coin classification method 100 uses the above-mentioned difference to distinguish the old and new versions of NT $ 10 coins.

In step 110, the image 200a including the coin image 210a is binarized. As shown in FIG. 2B, the binarized image 200b includes a coin image 210b.

In some embodiments, the coin classification method 100 further includes a step of reducing the size of the image 200a before binarizing the image 200a. By reducing the number of pixels included in the image 200a, the time required to execute the coin classification method 100 is reduced.

In step 120, the circle center 215b of the coin image 210b is located in the binarized image 200b.

As shown in FIG. 2B, the coin edge reflects so that the periphery of the coin image 210b is composed of bright points (ie, white pixels). The following describes in detail how to use this feature to locate the center 215b of the coin image 210b.

Please refer to FIG. 3, which is an image histogram 300 showing the image 200b of FIG. 2B. In some embodiments, step 120 includes the following two sub-steps: projecting the binarized image 200b along the Y direction to the X axis to generate an image histogram 300, wherein the vertical and horizontal axes are respectively the number of bright points and the X direction Coordinates, where the coordinates in the X direction are in pixels; and if the image histogram 300 includes a plurality of consecutive bright spot intervals, the center of the largest range in the consecutive bright spot intervals can be determined as the center 215b of the coin image 210b in the X direction. coordinate.

For example, the image histogram 300 has a maximum continuous bright spot interval 310, and the coin classification method 100 determines that the center 311 of the maximum continuous bright spot interval 310 is the coordinate of the circle center 215b in the X direction. The above-mentioned multiple consecutive bright spot intervals are defined as non-overlapping and unconnected plural intervals, and these intervals cover all X-direction coordinate points whose corresponding bright spots are greater than zero.

It should be noted that the background of the image 200a considered in this embodiment is dark and does not reflect light. Therefore, the part outside the coin image 210b on the image 200b does not have a large bright area, and most of the bright points fall on the edge or inside of the coin image 210b. The projection of the image 210b in the Y direction thus corresponds to the maximum continuous bright spot interval 310.

The coordinates of the center 215b in the Y direction can be calculated by changing the image 200b to be projected along the X direction to the Y axis. For details, refer to the related descriptions above, and details are not described herein.

In some embodiments, the coin classification method 100 further includes the step of performing contour detection on the binarized image 200b before projecting the binarized image 200b to the X direction to strengthen the bright area around the coin image 210b. To improve the accuracy of positioning the center 215b. The image 200c after contour detection is shown in FIG. 2C.

Taking into account the effects of wear, stains, and ambient light on the coin, the projection of the coin image 210b in the Y direction may correspond to a number of fairly close consecutive bright spot intervals. Therefore, these consecutive bright spot intervals need to be combined to form the coin image 210b. The continuous bright spot interval corresponding to the projection in the Y direction.

Therefore, in some embodiments, step 120 further includes the following two sub-steps: determining whether the image histogram 300 includes adjacent first consecutive bright point intervals and second consecutive bright point intervals, wherein the first consecutive bright point intervals include adjacent second consecutive bright point intervals The first endpoint of the highlight interval, and the second consecutive highlight interval includes the second endpoint adjacent to the first consecutive highlight interval, and the distance between the first endpoint and the second endpoint is less than a threshold; and The graph 300 includes a first continuous bright point interval and a second continuous bright point interval. The combination of the first continuous bright point interval and the second continuous bright point interval is regarded as one of the continuous bright point intervals, that is, the first continuous bright point interval is far from the second. The left endpoint of the continuous bright spot interval extends to the continuous bright spot interval away from the right endpoint of the first continuous bright spot interval of the second continuous bright spot interval.

In some embodiments, the coin classification method 100 further includes a step of removing noise from the binarized image 200b. In some embodiments, morphological dilation and erosion are used (erosion) method to strengthen the features of the first reflective area and the second reflective area on the back of the new version of the NT $ 10 coin to improve the accuracy of the coin classification method 100. The image 200d after noise removal is shown in FIG. 2D.

In step 130, the bright area on the image 200b is detected. If the binarized image 200b includes a plurality of bright areas, it is determined whether the bright area includes a first feature bright area and a second feature bright area, where the first feature The distance from the centroid of the bright area to the center of the circle 215b falls within a first predetermined distance range, and the distance from the centroid of the second characteristic bright area to the center of the circle 215b falls within a second predetermined distance range.

In some embodiments, blob detection can be used to detect bright areas on the image 200b.

In some embodiments, the first predetermined distance range is a range formed by the distance from the centroid of the first bright area of the standard NT $ 10 coin image to the center of the circle plus the tolerance, and the first The second predetermined distance range is a range formed by the distance from the centroid of the second feature bright area of the standard new version of the NT $ 10 coin image to the center of the circle plus the tolerance.

Taking into account the effects of wear, stains, and ambient light on the coins, the shapes of the first and second reflective areas on the new NT $ 10 coin images are not exactly the same. Therefore, adding the tolerance can improve the accuracy of identifying the first feature bright area and the second feature bright area.

In some embodiments, the centroid of the first characteristic bright area, the centroid of the second characteristic bright area, and the circle center 215 b are further substantially arranged along a straight line to improve the accuracy of the coin classification method 100. For example, the centroid of the first characteristic bright area, the centroid of the second characteristic bright area, and the center of the circle 215b and the line that best matches the three points described above (for example, by the least square method) Or the best straight line found by other analysis methods) The sum of the squared residuals is less than an upper limit.

In some embodiments, the number of bright spots of the first feature bright area is further limited to fall within the first predetermined number of bright spots, and the number of bright spots of the second feature bright area falls within the second predetermined number of bright spots to improve the method of coin classification. 100% accuracy.

In some embodiments, the first predetermined number of bright points is a range formed by the number of bright points included in the first characteristic bright area of a standard new version of the NT $ 10 coin image plus tolerance, and the second predetermined number of bright points range It is the range formed by the number of bright points included in the second feature bright area of the standard NT $ 10 coin image plus the tolerance.

In some embodiments, the aspect ratio of the smallest rectangle surrounding the first feature bright area falls within the first predetermined aspect ratio range, and the aspect ratio of the smallest rectangle surrounding the second feature bright area falls within the second predetermined Within the range of the aspect ratio to improve the accuracy of the coin classification method 100.

In some embodiments, the first predetermined aspect ratio range is a range formed by the aspect ratio of the smallest rectangle surrounding the first feature bright area of a standard new version of the NT $ 10 coin image plus a tolerance, and the second The predetermined aspect ratio range is a range formed by the aspect ratio plus tolerance of the smallest rectangle that surrounds the second characteristic bright area of the aforementioned standard new version of the NT $ 10 coin image.

In step 140, if the bright area includes the first feature bright area and the second feature bright area, it is determined that the coin image 210b is a new version of the NT $ 10 coin image. For example, the coin image 210b has a first characteristic light area 211b and a second characteristic light area 212b. The coin classification method 100 therefore determines a coin map. Image 210b is the new NT $ 10 coin image.

Please refer to FIG. 4, which is a combined diagram illustrating a coin sorting device 400 according to an embodiment of the present disclosure. The coin sorting apparatus 400 includes a first camera 410 and a processor 480. The first camera 410 is configured to capture an image of the coin 490 such that the captured image includes a coin image. For example, the image captured by the first camera 410 may be the image 200a shown in FIG. 2A, which includes the coin image 210a. The processor 480 is electrically connected to the first camera 410. The processor 480 receives the image captured by the first camera 410, and executes the coin classification method 100 to determine whether the coin 490 is a new version of the NT $ 10 coin.

In some embodiments, the coin sorting device 400 further includes a track 430, and the first camera 410 is configured to photograph the coins 490 sliding on the surface 431 of the track 430.

In some embodiments, the first camera 410 is located on a side of the surface 431 facing the track 430, and the first camera 410 is shooting toward the surface 431.

In some embodiments, the coin sorting device 400 further includes a ring light source (not shown), which is located between the first camera 410 and the track 430. The first camera 410 is photographed when the coin 490 passes under the ring light source. . The setting of the ring light source makes the coin 490 receive light more evenly, which is helpful to improve the accuracy of the coin classification method 100.

In some embodiments, the track 430 has a transparent area 432, and the coin sorting device 400 further includes a second camera 420, which is electrically connected to the processor 480, and is configured to capture coins 490 via the transparent area 432. In this embodiment, the processor 480 is configured to perform a first camera 410 and a second camera The image captured at 420 is subjected to the method 100 for sorting coins. When one of the coin images in the two images taken by the first camera 410 and the second camera 420 is determined by the coin classification method 100 as a new NT $ 10 coin image, it is determined that the coin 490 is the new NT $ 10 coin Yuan coin.

As the new version of the NT $ 10 coin only has two reflective areas on the back, when it is not possible to determine whether the coin 490 glides on the track 430 with the back facing up, the setting of the second camera 420 can ensure that the coin sorting device 400 can obtain a back view containing the coin 490 Like the image, it is correctly judged whether the coin 490 is a new version of the NT $ 10 coin.

In some embodiments, the lens of the first camera 410 and the lens of the second camera 420 face the transparent area 432. When the coin 490 slides on the surface 431 and passes through the transparent area 432, the coin 490 is photographed at the same time, and each coin 490 is included. Image of front and back images.

In some embodiments, the coin sorting device 400 further includes a striker 440 and a collecting area 450. The striker 440 is electrically connected to the processor 480, and the striker 440 and the collection area 450 are opposed to each other via the track 430. As the coins 490 glide on the track 430, they pass sequentially under the first camera 410 and between the striker 440 and the collection area 450. When the processor 480 determines that the coin 490 is a new version of the NT $ 10 coin, after the coin 490 slides between the striker 440 and the collection area 450, the striker 440 pushes the coin 490 into the collection area 450.

In some embodiments, the coin sorting device 400 further includes a roller 460 rotatably disposed above the surface 431 of the rail 430. When the coins 490 slide on the track 430, they pass through the roller 460, the first camera 410, and the striker 440 and the collection area 450 in order.

The roller 460 rotates in the direction of the track 430, and the minimum distance between the rolling surface 461 of the roller 460 and the surface 431 of the track 430 is smaller than the thickness of the coin 490. Therefore, when the coin 490 passes under the roller 460, the coin 490 is temporarily held in the Between the roller 460 and the track 430. With the rotation of the roller 460 and the frictional force between the roller 460 and the coin 490, the coin 490 is accelerated to be pushed out to increase the sliding distance on the track 430.

In addition, the scroll wheel 460 can help synchronize the operation of the striker 440 and the processor 480. The roller 460 rotates at a fixed speed, so that each coin 490 leaves the roller 460 at a fixed initial speed, and then the fixed coin 490 slides from below the first camera 410 to the taxi time required between the striker 440 and the collection area 450. Considering that the time required for the execution of the coin sorting method 100 is relatively short, the striker 440 receives the message from the processor 480 that the coin 490 is a new version of the NT $ 10 coin, and pushes the coin 490 into the collection area 450 after the above-mentioned taxi time.

In some embodiments, the track 430 includes a plurality of balls 433 rotatably embedded in the surface 431 of the track 430. The arrangement of the ball 433 can also increase the distance that the coin 490 can slide on the track 430.

In some embodiments, the coin sorting device 400 further includes a launching mechanism 470 including a rotating disk 471 and a baffle 472. The baffle 472 surrounds the rotating disk 471 and has an opening (not shown), which communicates with the track 430. When the coin sorting device 400 is used to sort coins 490, the coins 490 are first placed on the rotating plate 471, and the coins 490 are thrown out through the rotating plate 471 at high speed, so that the coins 490 enter the track 430 through the opening of the baffle 472.

To accelerate coin sorting, in some embodiments, the coin sorting device 400 allows a plurality of coins 490 to slide on the track 430 simultaneously. First A camera 410 is configured to capture an image having a width in the Y direction that is substantially equal to the diameter of the coin image (eg, image 200a in Figure 2A). Thereby, the coordinates in the Y direction of the coin image can be avoided, and the time required to execute the coin classification method 100 can be further reduced. In addition, the first camera 410 is configured to capture an image whose width in the X direction is less than twice the diameter of the coin image, so as to avoid that the image contains more than one complete coin image.

Please refer to Figures 5A and 5B. FIG. 5A shows an image 500a including a complete coin image 510a and a partial coin image 520a. FIG. 5B is an image 500b showing the image 500a of FIG. 5A after binarization and contour detection, which includes a complete coin image 510b and a partial coin image 520b. In the embodiment described in the previous paragraph, when a plurality of coins 490 are sliding on the track 430 at the same time, the image captured by the first camera 410 may include images of two coins that are quite close to or in contact with each other. For example, the image 500a shown in FIG. 5A includes a complete coin image 510a and a partial coin image 520a that are in contact with each other.

Please refer to FIG. 6, which is an image histogram 600 showing the image 500b of FIG. 5B. Because the full coin image 510b in the image 500b and the bright areas around the partial coin image 520b are connected, the method described above using the image histogram to calculate the center coordinates of the full coin image 510b corresponds to the full coin image 510b and The continuous bright spot intervals of some coin images 520b will be connected to form a maximum continuous bright spot interval 610, so that the center 611 of the maximum continuous bright spot interval 610 does not correctly correspond to the center coordinate of the complete coin image 510b.

Please refer to Figure 7 together, which is an image showing Figure 5B 500b of another image histogram 700. To solve the problem described in the previous paragraph, in some embodiments, the image histogram 700 generated by projecting the image 500b along the Y direction on the X axis, the number of bright points does not include pixels whose Y coordinate falls within the exclusion range 530b, where the exclusion range 530b covers where the full coin image 510b and the partial coin image 520b meet. For example, as shown in FIG. 5B, the exclusion range 530b may be an area where the image 500b falls between two dotted lines, that is, an area in the middle third of the image 500b. The image histogram 700 includes a maximum continuous bright spot interval 710, and its center 711 is determined to be the coordinate of the center of the circle of the complete coin image 510b in the X direction.

In summary, the coin classification method disclosed in the present disclosure uses the new version of the NT $ 10 coin with the characteristics of two reflective areas to binarize the image containing the coin image. If the binarized image has a size that meets the bright area, If the two characteristic bright areas of the shape, position, and other conditions, the coin image is determined to be a new version of the NT $ 10 coin image.

Although the present disclosure has been disclosed as above in the form of implementation, it is not intended to limit the present disclosure. Any person skilled in the art can make various modifications and retouches without departing from the spirit and scope of the present disclosure. Therefore, the protection of the present disclosure is protected. The scope shall be determined by the scope of the attached patent application.

Claims (15)

A coin classification method includes: binarizing an image including a coin image; locating a circle center of the coin image in the binarized image; and if the binarized image includes a plurality of bright Area, it is determined whether the bright areas include a first characteristic bright area and a second characteristic bright area, wherein a distance from a centroid to the circle center of the first characteristic bright area falls within a first predetermined distance range, The distance from the centroid to the circle center of the second characteristic bright area falls within a second predetermined distance range; and if the bright areas include the first characteristic bright area and the second characteristic bright area, the coin is judged The image is a new version of the NT $ 10 coin image. The coin classification method according to claim 1, further comprising performing noise removal on the binarized image. The coin classification method according to claim 1, wherein the centroid of the first characteristic bright area, the centroid of the second characteristic bright area, and the circle center are arranged substantially along a straight line. The coin sorting method according to claim 1, wherein the number of bright spots of the first characteristic bright area falls within a first predetermined number of bright spots, and the number of bright spots of the second characteristic bright area falls within a second predetermined number of bright spots Inside. The coin classification method according to claim 1, wherein an aspect ratio of a smallest rectangle surrounding the first characteristic bright area falls within a first predetermined aspect ratio range and a smallest rectangle surrounding the second characteristic bright area The aspect ratio falls within a second predetermined aspect ratio range. The method for classifying coins according to claim 1, wherein positioning the center of the circle comprises: projecting the binarized image to a direction to generate an image histogram, wherein a first axis of the image histogram is the The coordinate of the direction, one of the image histograms in the second axis is the number of bright points; and if the image histogram includes a plurality of consecutive bright point intervals, the largest of the consecutive bright point intervals in the first axis range The center is determined as the coordinate of the circle center of the coin image in the direction. The coin classification method according to claim 6, wherein the locating the center of the circle further comprises: determining whether the image histogram includes a first consecutive bright spot interval and a second consecutive bright spot interval, wherein the first consecutive bright spot interval Including a first endpoint adjacent to the second continuous highlight interval, the second continuous highlight interval including a second endpoint adjacent to the first consecutive highlight interval, and between the first endpoint and the second endpoint And the distance between the first consecutive bright point interval and the second consecutive bright point interval is regarded as the continuous if the image histogram includes the first consecutive bright point interval and the second consecutive bright point interval. One of the highlights. The coin classification method according to claim 6, further comprising performing contour detection on the binarized image before projecting the binarized image to the direction. A coin sorting device includes: at least one camera configured to shoot at least one image of a coin so that the at least one image includes a coin image; and a processor electrically connected to the at least one camera and configured to connect the at least one camera At least one image is binarized, and the processor is further configured to locate a circle center of the coin image in the at least one image that is binarized. If the at least one image that is binarized includes a plurality of bright areas, then The processor is further configured to determine whether the bright areas include a first characteristic bright area and a second characteristic bright area, wherein a distance from a centroid of the first characteristic bright area to the center of the circle falls at a first predetermined distance. Within the range, the distance from the centroid of the second characteristic bright area to the circle center falls within a second predetermined distance range. If the bright areas include the first characteristic bright area and the second characteristic bright area, the The processor is also configured to determine that the coin is a new version of the NT $ 10 coin. The coin sorting device according to claim 9, further comprising a track, and the at least one camera is configured to photograph the coin sliding on a surface of the track. The coin sorting device according to claim 10, wherein the at least one camera is located on a side facing the surface of the track. The coin sorting device according to claim 10, wherein the track has a transparent area, and the at least one camera includes: a first camera located on a side facing the surface of the track; and a second camera located away from the surface One side of the surface of the track is configured to photograph the coin through the transparent area. The coin sorting device according to claim 10, further comprising a striker and a collection area, the striker is electrically connected to the processor, and when the processor determines that the coin is a new version of the NT $ 10 coin, the striker Push the coin into the collection area. The coin sorting device according to claim 10, further comprising a roller rotatably disposed above the surface of the track, wherein a minimum distance between a rolling surface of the roller and the surface of the track is smaller than that of the coin. thickness. The coin sorting device according to claim 10, wherein the track includes a plurality of balls rotatably embedded in the surface of the track.