12 | P a g e
INDIAN COIN SEPARATOR AND COUNTING MACHINE USING EDGE DETECTION TECHNIQUE
Amith Salehittal
1, Vijay D Shetti
2, Medha M Naik
3, Vartur Sreekumar Sharanya
4Department of Information Science and Engineering, Canara Engineering College
ABSTRACT:
This project is focused on Separation and Counting of Indian currency coins is considered. Nearly all the temples in India have donation boxes (Hundi). More human intervention is required to separate such coins and hence it could be automated and as a result it improves efficiency and reduces the time consumption in the process.This project adapts automation for this time-taking process. It involves the usage of digital image processing technology that aids in detecting the coins and differentiating them; thus, making the process more fast and accurate.
The aim of this system is to identify and matches the Indian coins, finally count the number of one rupee, two rupees, five rupees, and ten rupee coins are used. There are various techniques are there to matches the coins. The proposed system utilizes the digital image processing template matching algorithm, by which the value of the coin is identified and also uses edge detection technique to detect the edge of the coin. That is to have computer read the image and matches the coins. Finally count the total value of the coins and it is displayed on the digital monitor.
KEYWORDS: Internet of Things, Metal Sensors, Server Motors
I. INTRODUCTION
In our Country, there are temples which have a lot of donation boxes. Many people use them and insert coins and currency notes into that as their religious donation. More human intervention is required to separate the currency coins that fell into the currency boxes (hundi). This is not reliable for today’s fast world. So, we have to take corrective actions against this process. Sorting of coins is a very time consuming and boring job.
The manual method of counting the coins does not have any recording device for future usage. This is not only happening in temples but also in banks which deals with more number of coins and currency every day. Banks uses bill counting machine to enumerate the money. But when the customer wants to pay a large number of cash into the bank, bank staffs may make mistakes to calculate the total value and number of one rupee, two rupee, five rupee, and ten rupee coins used. Some coins from different foreign currency look similar. So sometimes it is difficult to distinguish them by using human eyes, especially for large amount of coins. Moreover, because of the globalization, the banks often receive foreign currency that the staff may not recognize. The charities face the same situation as the bank, because the donators come from all over the world. So it is necessary to develop a system that can help them to recognize and calculate the money that they receive. As well as this time taking process will also need automation. Automation with flexibility provides a good result. Automation provides efficiency and accuracy. It does not need any human intervention. Coins and currencies are separated at the time of insertion of donation itself. Thereby it is reducing the time for separation of coins and currency manually.
With the help of a blower, it separates the currencies which are inserted into the donation boxes. Then coins are
13 | P a g e sorted according to their diameter. Counters are used for the counting purpose. The counted values are displayed on a digital display for the recording purpose. An electronic controller acts as a brain of this system which controls the sequence of operation. Separation of coins and currency and counting them is a time taking process which needs more manpower.
II. LITERATURE SURVEY
There are numerous techniques that are used to separate such coins of different denominations. The parameters such as size, weight and material have been used as the parameter to analyze and recognize the denomination of coin. Various programming languages like C, C++, C#, Java, Python etc. are also used.
Rotational invariant neural pattern recognition system for coin recognition. [1] In this work they have created a multi-layered neural network and preprocessor consists of many slabs of neurons. This preprocessor was used to get a rotational invariant input for the multi-layered neural network. For the weights of neurons in preprocessor, concept of circular array was used instead of square array. In 1993 tried to achieve 100% accuracy for coins. In this work they have used Back propagation and genetic algorithm to design neural network for coin recognition.
Back propagation is used to train the network.
To classify recently released Indian coins of different denominations and to count the total value of the coin in terms of Indian National Rupees [2] The objective of this work is to classify recently released Indian coins of different denominations and to count the total value of the coin in terms of Indian National Rupees (INR) [1, 2].
The system combines Robert’s edge detection method (93% accuracy), Laplacian of Gaussion edge detection method (94% accuracy), Canny edge detection method (97.5% accuracy) and Multi-Level Counter Propagation Neural Network (ML-CPNN) based approach (99.5% accuracy). Parameters like shape, size, surface, weight etc.
are considered and Matlab has been used as a medium for simulation and obtaining the result. The comparison between the existing technologies were discussed briefly by Modi et al., especially using image processing techniques and found that the maximum accuracy of 99.7% has been achieved in 1996 for Canadian coins using decision trees.
Recognition of Indian Currency and Denomination of Coin [3] This system used a FPGA (Field Processed Gate Array) controller coupled with MATLAB to detect the currency notes to check for faulty and fraudulent notes.
The denominations of the currency are counted by using the Artificial Neural Networks if the notes are genuine.
There is a DC gun for sorting process.
Indian Coin Detection and Sorting using SIFT algorithm [4] The system proposed by made uses of SIFT algorithm in addition with MATLAB tool to detect the coin. It makes use of feature extraction technique to distinguish between the coins. Similarly, they used edge detection algorithm to detect the coins
14 | P a g e Coin classification using a novel technique for learning characteristic decision trees by controlling the degree of generalization [5] In this work, presented an approach for coin classification using learning characteristic decision trees by controlling the degree of generalization. Decision trees constructed by ID3 like algorithms were unable to detect instances of categories not present in the set of training examples. Instead of being rejected, such instances get assigned to one of the classes actually present in the training set. Hieve 100% recognition accuracy rate. In pre-processing step, they have used Gaussian filter to smooth the image and then gray level thresholding is done to obtain binary image. For extracting the features, they have used Harris-Hassian detector. Next, the threshold value of feature points of the coin is calculated. Circular Hough Transform is used for circle detection. After circle detection coins are classified according to radius and threshold value
Three color selective stereo gradient method for fast topography recognition of metallic surfaces. [6] Similar to our approach translational invariance is achieved through segmentation, whereas rotational invariance is a result of a polar coordinate representation and correlation. Their system uses a special hardware to assure that no fraud coins are accepted by the system. In our case there is no risk that fraud images of coins will be presented to the system. Therefore, the use of colour seems to increase the computational costs unnecessarily. The basic idea of the method of detecting a straight line in coin’s image is discussed The criteria for coin classification based on gray- level, color, texture, shape, model, etc, are discussed. The method which specifically addresses coin segmentation based on color or gray value is reported. Many serious problems like shape of the coin, peak detection in surface of the coins are attempted but, it does not yield much result.
Rotation Invariant Coin Identification System along with pattern matching using a neural network. [7] This system is capable of collecting data and extracting the features of coin. In the first step the RGB coin image is acquired. Then RGB image is converted into Grayscale image. They have used Sobel filter for detecting the edges from image and to find radius and centre. In the next step they have generated feature vector. This feature vector is then given for trained Neural Network. Neural Network then classifies the coins.
Indian Coin Recognition using Image Subtraction Technique [8] This system checks the radius and Centroid of the coin image and then applies coarse and fine subtraction on input image. A subtraction between object image and database coin image is performed which provides fast recognition and good accuracy.
Harris-Hessian Algorithm for coin Apprehension [9] In pre-processing step, they have used Gaussian filter to smooth the image and then gray level thresholding is done to obtain binary image. For extracting the features, they have used Harris-Hassian detector. Next, the threshold value of feature points of the coin is calculated. Circular Hough Transform is used for circle detection. After circle detection coins are classified according to radius and threshold value
Coin Recognition by using Artificial Neural Network [10] They have used Sobel Edge Detection for detecting Circular Boundary. For removing shadow from the coin image they have used Circular Hough Transform. After
15 | P a g e they applied Discrete Cosine Transform and Discrete Wavelet Transform to segment the image into small blocks.
Then pattern averaged image is generated. In the last step feature vectors are generated and given as input to the Neural Network.
III. PROPOSED APPROACH
The intent of this paper is to identify and matches the Indian coins, finally count the number of one rupee, two rupees, five rupees, and ten rupee coins are used. There are various techniques are there to matches the coins. In this system Edge Detection technique was used. That is to have computer read the image and matches the coins. Finally count the total value of the coins. Techniques involved are image color segmentation, edge enhancement, edge detection, blob measurements, Hough transform. The proposed system utilizes the digital image processing template matching algorithm, by which the value of the coin is identified. It also makes use of the edge detection algorithm to detect the edges of the coins and counts them if the coin is turned up- side down and displays it on a digital monitor. It also matches color of the coin to segregate coins, as the new 1 rupee and 5 rupee coins are more or less similar in size. Coins are sorted according to their diameter.
Counters are used for the counting purpose. The counted values are displayed on a digital display for the recording purpose. This reduces the man power and increases the level of accuracy.
IV. COMPONENTS REQUIERED
Raspberry pi 4:
Figure 1: Raspberry pi 4 Board
The Raspberry Pi is a series of small single-board computers developed in the United Kingdom by the Raspberry Pi Foundation to promote teaching of basic computer science in schools and in developing countries.
Metal Sensors:
Figure 2: Metal Sensors
This is a module specifically designed to detect metal. The module operates by inducing currents in metal objects and responding when it occurs. A nice onboard buzzer signals when it detects something and an onboard potentiometer allow adjustment of sensitivity.
16 | P a g e Raspberry pi camera:
Figure 3: Raspberry pi Camera v2
The Raspberry Pi Camera v2 is the new official camera board released by the Raspberry Pi Foundation. The Raspberry Pi Camera Module v2 is a high quality 8 megapixel Sony IMX219 image sensor custom designed add-on board for Raspberry Pi, featuring a fixed focus lens.
MG995 Servo motors:
Figure 4: MG995 servo motor with all parts
MG995 is a servo motor providing precise rotation over 180º range its applications are many. The servo is suited for designing robotic arm in which wear and tear of motor is high.
V. ARCHITECTURE
Figure 5: Training and Testing phase of coin detection
17 | P a g e The proposed system consists following 7 steps as follows:
Step 1: Read input image
The captured image using Raspberry pi camera is taken as an input in testing phase.
Step 2: Convert input image into grey scale image
If there are color image then convert it into grayscale using weighted method.
The new equation that form is:
New grayscale image = ((0.3 * R) + (0.59 * G) + (0.11 * B))
According to this equation, Red has contributed 30%, Green has contributed 59% which is greater in all three colors and Blue has contributed 11%.
Step 3: Extract Features such as histogram, color feature and shape of the coin.
[counts,binLocations] = imhist(I) calculates the histogram for the grayscale image I. The imhist function returns the histogram counts in counts and the bin locations in binLocations. The number of bins in the histogram is determined by the image type.
O
ptionally can compute the histogram counts and bin locations using a GPU (requires Parallel Computing Toolbox™). For more information, see Image Processing on a GPU.Step 4: Perform normalization
After feature extraction, as it is the random process it create data redundancy. This can be overcome by normalization and this process also performs data integration of the coin and that data is sent to coin image feature trained database.
Step 5: Apply machine learning algorithm
Comparing the test image feature and coin image. Test image feature is a process which is having single image and that is obtained from training phase and coin image is obtained from trained database .Here 1:1 mapping is done between these two process using machine learning algorithm named SVM (Support Vector Machine) which is mainly used to solve classification or regression challenges.
Step 6: Perform class detection
Detecting the instance of a coin from a particular class .For example consider 3 classes, class1 ,class 2,class3 detecting for which class the particular coin belongs to and segregating the coins according to their respective class.
Step 7: Update count in server Updating the coin count of particular class. For example if 1 rupee is in class 1 then total count of class 1 is updated to the server.
VI. FUTURE SCOPE
Implementation is done only for coins. Integration of both coins and notes can be done by implementing principles of Machine Learning Algorithm. Database can be created for each notes and coins .Automated system can be developed which is able to distinguish between coins and notes in and around the acrylic box, segregate them and put them in respective slot. Finally outputting the total sum of coins and separately.
18 | P a g e
VII. CONCLUSION
Coin recognition using morphological operations shows positive signs for coin identification. Image segmentation used as the first step reduces total time requires executing the program. Machine Learning algorithm provides the clear edges of the coins to improve accuracy for coin detection. Different features such as histogram, color and shape are determined. Also blob measurements and normalization are performed which are used to provide and give precise results.
REFERENCES
[1] Minoru Fukumi, Sigeru Omatu, Fumiaki Takeda, and Toshihisa Kosaka,” Rotation invariant neural pattern recognition system with application to coin recognition”, IEEE Transactions on Neural Networks 3 (1992), no. 2, 272–279
[2] Velu CM, P.Vivekanadan , Kashwan K R, “Indian Coin Recognition and Sum Counting System of Image Data Mining Using Artificial Neural Networks”. International Journal of Advanced Science and Technology. Jun 2011; 31, 67–80p.
[3] Anija KP, Afna mamu, Aisha Chandhni, Kavya Madhusoodhanan, “Recognition of Indian Currency and Denomination of Coin”. International Journal for Scientific Research & Development. 2016; 4(1): 843–
846p.
[4] Rohan S, Akshat G Poi, Sahil G Khorjuvekar, Vaibhav Naik, “Indian Coin Detection and Sorting using SIFT algorithm”. International Journal of Science Technology & Engineering. Apr 2016; 2(10): 601–603p.
[5] Paul Davidsson, “Coin classification using a novel technique for learning characteristic decision trees by controlling the degree of generalization”, Ninth International Conference on Industrial & Engineering Applications of Artificial Intelligence & Expert Systems (IEA/AIE-96), Gordon and Breach Science Publishers, 1996, pp. 403–412.
[6] M. Adameck, M. Hossfeld, and M. Eich, “Three color selective stereo gradient method for fast topography recognition of metallic surfaces”, Proceedings of Electronic Imaging, Science and Technology (Martin A.
Hunt and Jeffery R. Price, eds.), Machine Vision Applications in Industrial Inspection XI, vol. SPIE 5011, January 2003, pp. 128–139.
[7] Sabita Pal, Gaurav Kumar and Sovan Raj Meher, “Indian Coin Recognition using Image Subtraction Technique”, IJCAES 2014
[8] Harveen Kaur and Neetu Sharma, “Modified coin identification using Neural Network”, IJETAE 2014 [9] Saranya das.Y. M and R.Pugazhenthi, “Harris-Hessian Algorithm for coin Apprehension”, IJARCET 2013 [10] Tushar N.Nimbhorkar and M.M.Bartere, “Coin Recognition by using Artificial Neural Network”, IJCITB 2013