Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
UNIVERSAL COIN SORTER AND COIN COUNTING MACHINE
Document Type and Number:
WIPO Patent Application WO/2016/199074
Kind Code:
A1
Abstract:
A coin counting and sorting device is capable of counting and sorting various denominations of coin irrespective of form and weight. A transporter system picks up coins from a coin hopper system to capture coin images with a camera and then store the images in a library of coin images for identification using machine vision. A sorting system sorts the coins into respective bins depending upon the coin denominations identified. A user interface component indicates the total value of the coins counted and performs other actions

Inventors:
ANAND MANJUNATH (CA)
CHAUHAN VEDANG DILIPKUMAR (CA)
JOSHI KEYUR DINISHCHANDRA (CA)
SURGENOR BRIAN WILLIAM (CA)
COLE JEFFREY (CA)
VARMA VIKRAM (CA)
ROTTI VINAY JAYATHIRTHA (IN)
Application Number:
PCT/IB2016/053417
Publication Date:
December 15, 2016
Filing Date:
June 09, 2016
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
9293507 CANADA INC (CA)
International Classes:
G07D3/16; G06F15/18; G07D5/00
Foreign References:
US5865673A1999-02-02
US20070125863A12007-06-07
US20130202184A12013-08-08
Attorney, Agent or Firm:
SMITH, Ryan T. et al. (1300 Yonge StreetSuite 50, Toronto Ontario M4T 1X3, CA)
Download PDF:
Claims:
We claim:

1. A device used for counting and sorting of any coin denominations irresptective of form and weight , the device compromising of:

a coin transporter system to pick up coins from coin hopper, transfer to camera zone and then drop into coin sorting area;

a machine vision system (MVS) with deep learning based machine vision system to capture coin images illuminated by dark field illumination, capture by a camera and stores in library /libraries of images of coins in computer system;

a coin hopper system to allow user load coins to be sorted, and allow controlled flow of coins towards rotating wheel;

a coin sorting system which allows sorting of coin into respective bin depending upon the coin denomination identified by the system; and

a user interface component to indicate total value of coins counted and perform other actions defined later in the claims.

2. The device of claim 1 , uses machine vision-based deep learning principles to identify coins of any denomination irrespective of form and weight.

3. The device of claim 1 , sorts and count coins from multiple countries.

4. The device of claim 1 , identifies the fake or damaged coin and sorts into a separate reject bin.

5. The device of claim 1 , sorts and counts tokens used for transit or other application along with coins of any denominations.

6. The device of claim 1 , allows introduction of new coins into the system without hardware change, requires only a software update through connected internet system.

7. The device of claim 1 , allows user to choose any bins for any coin

denominations, the selection is made using suitable user interface or software-based app.

8. The device of claim 1 , allows user to choose multiple bins for same

denominations.

9. The coin transporter of claim 1 , consist of a wheel connected to standard electric motor and mounted upon a structural chassis.

10. The coin transporter of claim 9, consists of compartment(s) spaced at equal intervals along the wheel diameter, allows pickup of coins from hopper area and transport towards camera zone.

11. The compartment(s) of claim 10, has unique form(s) that will allow only one coin to be picked up at a time.

12. The wheel of claim 9, has surface(s) with rubber finish or conveyor belt or other means that will provide frictional contact to pick up the coins against the gravity force.

13. The wheel of claim 9, has proximity metal sensor or conductive sensor or other means to detect non-metallic coins deemed to be fake.

14. The sensor(s) of claim 13, provides trigger information (delayed or non- delayed) to the camera to capture the coin image when the coin reaches the camera zone.

15. The sensor(s) of claim 13, also provides information to the camera to skip the image capture of the coin detected is considered to be fake.

16. The sensor(s) of claim 13, also provides information to the coin sorting system to route the fake coin into reject bin.

17. The MVS of claim 1 , consists of dark field illumination system (DFI), camera, computer unit (PC), deep learning algorithm(s) and database(s).

18. The DFI of claim 17, produces light source within dark filed zone such that only making(s) such as engraving(s) or projection(s) in the coin are illuminated.

19. The DFI of claim 17, has housing, light emitting diodes (LED) or other means of light source(s) and diffuser.

20. The DFI of claim 17, is mounted on the chassis such that the rotating wheel with coins passed under the DFI.

21. The camera of claim 17, is a standard camera with lens attached, captures the marking details on the coin illuminated by DFI and triggered by the sensor of claim 6.

22. The camera of claim 21 , transfers the captured image into PC for processing by the deep learning algorithm(s).

23. The PC of claim 17, consists of standard computer processor with or without graphics processing unit embedded with deep learning algorithm(s).

24. The database of claim 17, consists of library with images of coins pre-captured by the manufacturer using a pre-trained system.

25. The MVS of claim 17, captures the coin image, performs image processing tasks, and then compares with database of coin images using deep learning algorithm(s). Once the coin denomination is identified, the PC provides information to the coin sorting system to sort the coins into respective bins.

26. The image processing tasks of claim 25, consists of distinguishing the coin from background, crops the image, centers the coin within the image, and adjusts the image size.

27. The image processing tasks of claim 26, is one of many other possibilities to perform the same tasks and not necessarily restricted to the defined method.

28. The coin sorter of claim 1 , consists of coin receiver, deflector and bins.

29. The coin receiver of claim 28 is a trapezoid-shaped component which receives coins from the wheel and transfers to coin deflector.

30. The coin deflector of claim 28, is attached to a standard stepper motor which transfers the coins into respective bins based on the information received from MVS.

31. The bins of claim 28, receives the coins of particular denomination including rejected or fake coins.

32. The bins of claim 31 , allows attachment to external bins or bags or other means.

33. The bins of claim 31 , will be enclosed inside a housing with security or locks interface or biometrics or other means which allows only authorized users to access to collect the coins.

34. The coin hopper system of claim 1 , consists of a hopper, component with radial blades or fins and standard electric motor.

35. The coin hopper of claim 34, allows user to load the coins to be counted and sorted, the component with motor transfers the coins from hopper towards the rotating wheel.

36. The coin sorter of claim 1 , consists of receiving channel, deflector system attached to electric motor and receiving bins.

37. The receiving channel of claim 36, has V-shaped interface that receive coins from the rotating wheel and transfers into bins through deflector component.

38. The deflector of claim 36, is attached to a standard electric (stepper) motor, channels the coins into respective bins based on the information received from the MVS.

39. The MVS of claim 17, identifies the damaged coins and provides rejected signal to the coin sorter system of claim 36 which further directs into reject bin.

40. The device of claim 1 allows user to store and report on-demand historical information of value and volume of coins counted for the organization throughout the product life cycle.

41. The device of claim 1 , will communicate with a central computer to log the number of coins and the value counted in order to estimate the coins in circulations throughout the network of machines.

42. The device of claim 1 , provides an option where the user is allowed to choose the option to receive automatic texts to phone(s) or email notification(s) of the value and count of coins deposited into the machine per transaction and / or regular basis (daily, weekly, monthly, quarterly, annually) enabled by a communication unit within the machine.

43. The device of claim 1 , allows users to deposit coins and redeem value through e-gift certificates sent via email or SMS instantly through user interface of claim 7.

44. The device of claim 1 , further comprises a biometric authentication mechanism to authenticate the user who collects the bags of coins attached to bins of claim 26.

45. The authentication mechanism of claim 44 further comprises, but is not limited to finger print authentication, pattern recognition (similar to android phones), key to open the lock box (part of the machine) containing the bagged coins and the like.

Description:
UNIVERSAL COIN SORTER AND COIN COUNTING MACHINE

FIELD OF THE INVENTION

The present invention relates to the field of simultaneous coin sorting and counting machines. More particularly, the present invention relates to machine vision-based deep neural network (also called deep learning) principles used for identifying various coin denominations using multiple features of the coin including shape, size, color, engravings, etc. followed by counting and sorting into respective tray or bag. The invention is aimed to replace the human-assisted or form or weight based sorting / counting currently used in several countries. The existing methods are non-viable in countries where coin denominations overlap with respect to form and weight requiring advanced methods to do so.

BACKGROUND OF THE INVENTION

[oooi] The current application is the extension of provisional patent filed at

Intellectual Property India, reference number 2929/CHE/2015 with priority date on 2015 June 10.

[0002] In many countries, cash transactions are still predominant and particularly the use of coins are popular.

[0003] Significant efforts are made by the countries to replace paper-based bills to coins to increase the life expectancy of the currency.

[0004] Also existing coin denominations are replaced by smaller and light weight coins of the same denominations, posing challenge for existing coin sorting and counting technologies to adapt with new changes and introduction of new coin denominations.

[0005] Businesses in cash intensive economies has very large transactions each day in the form of coins, posing a huge labor requirement to sort and count.

[0006] Pilferage or inaccuracies or labor intense activities pose challenge for such organizations and thus reducing the circulation of coins [0007] The main aim of this invention is to eliminate the human intervention, increase the accuracy upto 100% and promote the use of coins thus reducing the dependency on paper-based bills and thereby increasing the recirculation of coins.

[0008] This invention also aims at helping staff / consumers focus on core tasks rather than count coins, saving time and costs.

[0009] Machines for counting relatively large quantities of consumer coins include those disclosed in, for example: U.S. Pat. Nos. 7,971 ,699, 7,874,478, 7,520,374, 8,033,375 and 8,332,313; each of which is incorporated herein by reference in its entirety.

SUMMARY OF THE INVENTION

[ooio] The present invention provides a novel system and method for coin sorting and counting using rotation-based coin transport and dynamic coin sorting which obviates or mitigates at least one disadvantage of the prior art.

[ooii] According to one aspect of the present invention, there is provided a method of identification of fake or plastic or deformed or damaged coins using a novel concept described later in the document.

[0012] According to one other aspect of the present invention, there is provided a method to identify the coin denomination using machine vision-based deep learning principle described later in the document.

[0013] Thus, the present invention allows easy adaptation for counting and sorting denominations of different countries or multiple countries in the same device without replacing or servicing the hardware.

[0014] The present invention counts and sorts coins of mutiple denominations of same country or multiple countries.

[0015] Other features and advantages of the present invention are described in more detail below.

[0016] These and other aspects of the embodiments herein will be better appreciated and understood when considered in conjunction with the following description and the accompanying drawings. It should be understood, however, that the following descriptions, while indicating preferred embodiments and numerous specific details thereof, are given by way of illustration and not of limitation. Many changes and modifications may be made within the scope of the embodiments herein without departing from the spirit thereof, and the embodiments herein include all such modifications.

BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Preferred embodiments of the present invention will now be described, by way of example only, with reference to the attached Figures, wherein:

Figure 1 shows the basic concept which is the primary claim of the patent;

Figure 2 shows other side of the concept with simple chassis integrated;

Figure 3 shows the details of coin hopper system;

Figure 4 shows the details of coin transporter system;

Figure 5 shows the machine vision system with camera, lens and dark field illuminator;

Figure 6 shows the details of coin sorter concept;

Figure 7 shows schematic of coin flow through the system;

Figure 8a shows the actual coin image captured by the camera under normal lighting conditions;

Figure 8b shows the coin image seen by camera under dark field illumination which is used for image processing and classification of denomination;

Figure 9 shows the example of filters used to extract information from the capture image of coin denomination using deep neural networks process;

Figure 10 shows the perspective view of the machine, according to an embodiment of the present invention;

DETAILED DESCRIPTION OF THE INVENTION

[0018] The proposed invention shown in Figure 1 , consists of coin hopper 10, coin transporter 20, machine vision system (MVS) 30 and coin sorter 40.

[0019] All components of the system are supported by a structural chassis 50 as shown in Figure 2. The system also consists of external covers and user interface panel not shown in the drawings.

[0020] The coin hopper system shown in Figure 3 consists of hopper 101 where the user will load the coins to be sorted and counted.

[0021] The coins are further loaded into the system using an axial turbine-shaped component 103 rotated by a standard electric motor 102.

[0022] The inclined ramp 104 and soft brush 105 ensures the coin flows due to gravity towards the coin transporter wheel 201.

[0023] Coin transporter system shown in Figure 3 consists of rotating wheel 201 , central rotation shaft 202, standard electric motor 204 and induction based proximity sensor 205.

[0024] The rotating wheel 201 has several unique form(s) 204a and 204b which helps in picking up the coins from 105 and transports the coins against gravity.

[0025] The form 204a and 204b helps in keeping the coin stable as it reaches the camera zone.

[0026] The proximity sensor 205 shown in Figure 4, detects the presence of metal coins and provides delayed trigger information for the camera to capture image to process further.

[0027] In case of non-metallic object which is considered to be fake, no signal is sensed by the sensor. No signal from the sensor provides information to the coin sorter system 40 to sort the fake object into reject bin 404.

[0028] MVS 30 shown in Figure 5 consists of a Dark Field Illuminator (DFI), camera 303 and lens 304.

[0029] Machine Vision System (MVS) 30 shown in Figure 5 consists of both hardware and software elements. The hardware elements are a digital camera 303, lens 304 and a dark-field illuminator (DFI) 301 and 302. [0030] The software is based on Deep Learning Algorithms (not detailed in this invention).

[0031] The camera acquires a single image of a coin when the coin is at the topmost position on the feeder wheel 20.

[0032] The camera 303 and lens 304 are positioned such that entire coin is visible in the field of view of the camera.

[0033] The image acquisition is performed under the DFI to highlight only edges and patterns of the coin.

[0034] The DFI used in this invention consists of housing 301 with light emitting diodes 302 mounted on an inclined surface at an angle between 60° - 85° w.r.t horizontal plane.

[0035] The DFI also consists of diffuser (not shown) which helps in diffusion and equal distribution of light intensity on the coin surface.

[0036] An example of coin image under normal lighting conditions captured by a digital camera 303 is shown in Figure8a. Image of the coin captured by the camera 303 of the MVS 30 under DFI 301 and 302 is shown in Figure 8b.

[0037] The acquired image is then transferred to a PC that runs a Deep Learning Algorithm with or without Graphics Processing Unit (GPU).

[0038] The deep learning principles consists, but not limited to two stages of development or the steps involved in the processs. First stage (Training stage), where the deep learning algorithm extracts features from the acquired coin images using image processing techniques desbribed in the next section but not necessariliy and it learns the relation between a coin denomination and its features.

[0039] Given a captured image as input data, we use N classifiers, each of which tests for one of the N different coin denominations used in a country. The employed classifiers, e.g., neural networks, are trained on a large database of coin images.

[0040] Second Stage (Testing stage), where a single coin image is acquired and the algorithm extracts the features from the coin image. It then compares the extracted features with the knowledge base (generated during the training stage) and assigns the coin a denomination with the highest match. [0041] Neural networks process the captured image in a hierarchical manner by first extracting edge information using automatically trained filters such as the ones shown in Fig. 9. The filter responses are then successively combined in the next levels of the network, until a probability is obtained in the final stage

[0042] We use the probabilities obtained from the N classifiers which are loaded into the processor of the coin sorter to decide into which of the N bins the respective coin should be sorted. More specifically, each classifier is assigned one bin, and the coin under consideration is sorted into that bin, that corresponds to the classifier returning the highest probability larger than a predefined threshold.

[0043] We sort the respective coin into the outlier bin if none of the probabilities is larger than the threshold. If the coin was assigned to one of the valid denominations, we add its value to the total count.

[0044] This design provides the unique capability to use the proposed approach for any number of coin denominations. Importantly, by means of a simple software update the proposed machine will be used for a different set of coins.

[0045] For a different country, as well as different number of coin denominations. The only required change is to load additional classifiers into the processor of the coin sorter.

[0046] For each loaded classifier the value that should be added to the total count shall be provided. No change of hardware is required assuming that the number of available bins is sufficiently large.

[0047] In case where a match is not found, the coin is considered as out of the database coin and it is transferred to the rejected bin by the coin sorter 40. The testing procedure is repeated for each acquired image of a coin.

[0048] In case where a match is found, the MVS 30 then signals the coin sorter 40 as shown in Figure 6 to transfer the coin in to its correct bin.

[0049] The electric motor 403 rotates at particular steps depending upon the inputs received from the controller which further rotates the trapezoid-shaped part 402.

[0050] Rotation of trapezoid-shaped part 402 directs each coin into respective bin 404 [0051] Each bin(s) 404 shall have sensors (not shown) implemented to indicate the user when the bin is filled with coins prompting the user to unload the bin 404 and replace with empty bin 404.

[0052] The sensor (not shown) shall also detect the presence of bin 404 before and during the coin sorting process.

[0053] The perspective view of the machine with external user interfaces is shown in figure 10.

[0054] The entire mechanism is enclosed by the housing 501 constructed using but not limited to plastics and provides protection to internal mechanism and MVS.

[0055] The user interface panel 502 consists of but not restricted to touch screen interface, allows user with or without authentication or service personnel to interact with the machine.

[0056] The user interface panel allows the user to select the bins configuration for each denomination, enter the passcode for authentication and enter the other information(s) related to user or machine.

[0057] Using the interface the user will redeem the collected currency value to equivalent e-gifts or transfer to bank account or other means of transferring the currency.

[0058] The user interface will allow the user to enter mobile number or electronic mail address to provide notifications with many other details when the counting has completed or other issues encountered by the machine.

[0059] The machine has communication to internet or network through wired connection at 507 or internal wireless modem (not shown).

[0060] The machine is interfaced with external printer through wired connection at

508 or internal printer of any type 505.

[0061] The machine will print the summary with quantity of each coin denominations counted, total value of each denominations counted and total value of the

denominations counted. The summary also includes the quantity of rejected coins.

[0062] The door 503 allows the user to access the sorting bin which is internal to the device but not limited this option alone. [0063] The authentication port 504 has means to allow the user to provide authentication which includes but not limited to key or biometrics or other means.

[0064] The authentication port 504 unlocks the door 503 once the authentication is successfully verified by the system.

[0065] The opening 506 allows the debris or other wastes to be rejected by the machine.