Claims:The scope of the invention is defined by the following claims:

Claim:
1. A system/method for segregating and counting currency using deep learning based image processing technique, said system/method comprising the steps of:
a) A segregating layer (1) to identify and differentiate cash in notes and coin form.
b) A notes layer (2) comprising a ventilation drum (3) to remove the folding’s in notes and segregating the different types of notes.
c) A coin layer (4) comprising a conveyer belt (5) where the coins are deposited followed by a robotic arm (6) with artificial intelligence to identify the different types of coins.
2. As mentioned in claim 1, the folding’s or any foreign particles in the notes are removed by using the ventilation drum.
3. According to claim 1, coins are pulled out from the coin layer through the conveyer belt at uniform speed.
4. As per claim 1, coins are identified and picked by using the deep learning artificial intelligence based robotic arm. , Description:Field of Invention
The present invention relates to, all the temples to count the number of coins and rupees notes there in the hundi by applying the deep learning based image processing. By introducing this type of machine, we can reduce the number of human’s require to complete the work as well as time.
Background of the invention
A number of coin sorting machines have already been developed. However one system, depicted in David Goh's (US1989/4995848), sorts coins using two ways, each dependent on the width of the coins. Coins are put into something like a bin within that device. Single coins are fed on with a developed this system by a spinning wheel. The back sides of the coins land it against Surface Area as they roll down the steps. When specified denomination flow into holes of varied sizes with in Surface Area, these are accepted. Another type of machine has been introduced in U.S. Pat. No. (US1977/4059122) to Yoshio Kinoshita, this can performed based on the denominator of the coins.
The range of applications is an extended version of a patent application, registration number (2929/CHE/2015), published at Intellectual Property India on June 10, 2015. Devices for measuring massive amounts of consumer coins are described, for example, in (U.S. Pat. Nos. (US2014/7971699), (US2013/7874478), (US2014/7520374), (US13/906,126) and (US09/646,446) each being included throughout in this invention as a references.
Another coin differentiating system is disclosed in (US09/646,446), which includes a light source irradiating mechanism for spotlighting the Surfaces of a coin to evaluate for Face unevenness and outer layer design. The ionizing radiation Sources consists of a number of optic fiber lightings which it expose the coin Surfaces. The reflected light from the coin face is converted into an electrical signal, which are processed and examined to ascertain the originality and kind of coin as differentiated by a photovoltaic conversion method.
The primary goal of this invention is to minimize human interaction, raise accuracy to 100%, and encourage the usage of coins, hence lowering reliance on paper-based currencies and promoting coin recycling.
Summary of the invention
The current invention is primarily concerned with improving methods for detecting, distinguishing, and validating coins of one or more denominations. Another goal is to recognize and validate coins of the same or various denominations using deep learning based image processing technique. The same mechanisms for coins used to identify and count the number of notes in hundi.
Brief description of Drawing
The invention will be described in detail with reference to the exemplary embodiments shown in the figures wherein:
Figure 1, 1 Deep Learning based Coin Counting Machine.
Detailed description of the invention
Throughout this document the Fig 1 is considered as for referencing. The preferable implementation will be detailed as employed for detecting rupees notes and sorting coins using a deep learning based image processing algorithm in the proposed invention. The components used in this proposed inventions are ventilation drum, robotic arm to pick the matched coins.
The deep learning based concepts is applied on the existing coins images, which can be maintained as dataset/new dataset. The important attributes of coin datasets is size, dimension, width, thickness, front view, and back view. This proposed algorithm is applied only on coins to identify the coin types and give to the counter machine. This entire task is made in terms of image based robot arm. Then comes to efficiency of this machine, upto now we will get around 60% to 65% of accuracy and efficiency. The detailed working process of this invention is discussed below.
We have to patch the two layers in Hundi. All currency notes and papers will be put in the first layer. The weighted materials come into hundi's second layer. All the notes are moved to the ventilation drum using the conveyor belt, then the notes are flattened. The currency counter system will then count the notes and provide the total amounts. The same method is carried out for coins, picking up the coins and placing them in the counter using the deep learning-based image processing robot. This image processing based robot arms has the camera to scan the coins and then validate the image with our coin dataset. If it is matched then, the robot arm is used to pick that coins and put it to the corresponding counter machine like one, two, five and ten rupees coins. If it is not matched then, the corresponding coins or some unwanted materials is removed from the conveyer belt. The speed of the conveyer belt is controlled by the weight of the hundi. The process for interpreting the coins employs an image processing system including several alignment and outer layer machine learning, whereby its specifications of the coin exterior corners and geometrical could be matched to a supermajority of reference coin images collected in a dataset, and in which the excellence of the match is determined by probability of an absolute match between both the origin and collected coin images. The matching score is calculated and compared with different classification algorithms, highest matching is picked based on the deep learning based decision making algorithms.
As described above the present invention relates to reducing the human efforts on counting the coins in the big temples like thirupathi, sabarimala, thiruvanamalai etc, The figure 1 has the exact detailed flow of the proposed method like; The novel hundi design has two layer one is for rupee notes and another one is for coins as shown in Fig.1. The proposed invention novel in terms of; It used deep learning based on image processing, Scanner over the conveyer belt to pick the odd one, Reducing the man power, Reduce the time.
4 Claims & 1 Figure