TWI750687B - Finger vein identification risk assessment system and method - Google Patents

Abstract

A finger vein identification risk assessment system and method are provided. The system includes a data reading device and a credit evaluation device. The credit evaluation device includes a nail recognition module, a pulse recognition module, a finger vein recognition module, a category recognition module, a risk assessment module and a loan calculation module. The category recognition module analyzes the customer's category based on the shape of the nail, the amplitude, frequency and waveform of the pulse and the veins of the finger vein. The risk assessment module obtains the risk value corresponding to the category. The loan calculation module is to generate the recommended loan amount when the risk value is greater than the threshold value, and the loan interest rate and loan period corresponding to the risk value when the risk value is less than or equal to the threshold.

Description

Finger vein identification risk assessment system and method

The present invention relates to a risk assessment system, especially a risk assessment system based on finger vein recognition technology.

In today's financial lending, if the financial business required by the customer is related to financing or borrowing, the bank must first evaluate and measure the financial credit of the customer in accordance with internal regulations and procedures. However, the calculation and assessment of customers' credit risks are still carried out manually, which will result in a lengthy and inefficient assessment process, which will indirectly lengthen the review time of assessment cases.

In addition, traditional credit risk assessment cannot adjust the credit risk assessment model as the customer's default status changes. Therefore, it is not possible to immediately adjust the credit risk assessment model in response to the changing default conditions.

The bank pre-launched the credit loan fast loan approval service, and the service demands are fast response and funding. For customers in urgent need of funds, it can be a convenient loan channel. However, in order to provide fast credit loan approval services with quick response and funding, banks need to have a system that can quickly conduct credit risk assessment to assess the credit risk of customers.

In view of the above-mentioned problems to be solved and the reasons, in particular, the present invention provides a finger vein identification risk assessment system and method, which can analyze the category of the customer by identifying the customer's fingernail, pulse condition and finger vein according to the relevant information of the finger vein feature. , and quickly obtain the customer's credit risk through risk assessment.

The present invention is a finger vein identification risk assessment system. When a customer applies for a loan from a bank, the above finger vein identification risk assessment system is used to assess the fraud risk of the loan. The finger vein identification risk assessment system includes a data reading device and a credit assessment device. The above-mentioned data reading device includes a basic database, a finger vein database and a default database, wherein the finger vein database includes a plurality of nail data, a plurality of pulse condition data and a plurality of finger vein data. The above-mentioned credit evaluation device includes a nail identification module, a pulse identification module, a finger vein identification module, a category identification module, a risk assessment module, and a loan calculation module. The above-mentioned nail identification module divides the nail data into multiple nail categories according to the shape of the nail. The above-mentioned pulse condition identification module classifies the pulse condition data into a plurality of pulse condition categories according to the amplitude, frequency and waveform of the pulse condition. The above finger vein identification module divides the finger vein data into a plurality of finger vein categories according to the patterns of the finger vein blood vessels. The above-mentioned category identification module identifies the veins of the customer's finger, the shape of the nail and the amplitude, frequency and waveform of the pulse, and analyzes the category of the customer. The category corresponding to a certain pulse category. The above risk assessment module obtains the VaR corresponding to the above category according to the data in the basic database and default database, where the risk value is (AB)/(A+B), where A is the number of defaults in the category , B is the non-default within the category number. The above loan calculation module, when the risk value is greater than a threshold value, generates the recommended loan amount corresponding to the risk value, and generates the loan interest rate and loan period corresponding to the risk value when the risk value is less than or equal to a threshold value.

According to another embodiment, the above-mentioned finger vein identification risk assessment system further includes an appropriation device. After approving the loan request, the appropriation account information is checked to confirm the appropriation.

According to yet another embodiment, the basic database is for storing customer basic data.

According to yet another embodiment, the default database is for storing default data.

According to another embodiment, the above-mentioned finger vein identification risk assessment system further includes an input device for inputting the client's nail data, finger vein data and pulse condition data.

The present invention is a finger vein identification risk assessment method, which includes a customer applying for a loan from a bank; obtaining customer data, the above customer data includes the data of a basic database, a finger vein database and a default database, wherein the data of the above finger vein database Including multiple nail shape data, multiple pulse data and multiple finger vein data; according to the nail shape, the above nail data is divided into multiple nail categories; according to the pulse amplitude, frequency and waveform, the above pulse data is divided into multiple pulse categories; according to The pattern of finger veins divides these finger vein data into multiple finger vein categories; according to the above-mentioned customer's nail shape, pulse amplitude, frequency and waveform, as well as the pattern of finger veins, analyze the category to which one of the above customers belongs. Among them, the above category belongs to The category is the category corresponding to the above-mentioned nail category, the above-mentioned pulse condition category and the above-mentioned finger vein category; according to the data of the above-mentioned basic database and the above-mentioned default database, one of the risk values of the above-mentioned categories is obtained, wherein the above-mentioned risk value is (AB)/ (A+B), where A is the number of defaults in the above category, and B is the number of non-defaults in the above category; When the risk value is greater than a threshold value, a recommended loan amount corresponding to the risk value is generated; when the risk value is less than or equal to a threshold value, a loan interest rate and a loan period corresponding to the risk value are generated.

According to another embodiment, the above method further includes verifying the appropriation by checking the appropriation account data after approving the loan request.

According to yet another embodiment, the above-mentioned basic database is for storing a customer's basic data.

According to yet another embodiment, the above-mentioned default database stores a default data.

According to another embodiment, the above method further includes inputting the client's nail data, finger vein data and pulse data.

To sum up, the present invention analyzes the category of the customer by identifying the customer's fingernails, pulse and finger veins, and quickly obtains the customer's credit risk through the risk assessment module based on the information related to the characteristics of the finger vein. When the risk value is greater than a threshold value, a recommended loan amount is generated, and when the risk value is less than or equal to a threshold value, the loan interest rate and loan period are generated.

100: Finger Vein Identification Risk Assessment System

110: Customers

111: Input device

120: Data reading device

121: Basic database

122: Finger Vein Database

123: Default Database

130: Credit Evaluation Device

131: Nail recognition module

132: Pulse recognition module

1325: Finger Vein Recognition Module

133: Category identification module

134: Risk Assessment Module

135: Loan Calculation Module

140: Appropriation Device

200-211: Steps

In order to make the above and other objects, features, advantages and embodiments of the present invention more clearly understood, the accompanying drawings are described as follows: FIG. 1 illustrates a finger vein identification risk assessment system according to an embodiment of the present invention The structure diagram of the relationship between each device and module.

FIG. 2 is a schematic flowchart illustrating a risk assessment method for finger vein identification according to an embodiment of the present invention.

3-7 are pulse diagrams according to an embodiment of the present invention.

FIG. 8 is a flowchart illustrating a convolution according to an embodiment of the present invention.

Please refer to FIG. 1 . FIG. 1 is a diagram illustrating a relationship structure of various devices and modules in a finger vein identification risk assessment system according to an embodiment of the present invention. When a customer 110 applies for a loan, the finger vein recognition risk assessment system 100 is used to assess the fraud risk of the loan. The finger vein identification risk assessment system 100 of FIG. 1 includes a data reading device 120 and a credit assessment device 130 .

According to another embodiment of the present invention, the finger vein identification risk assessment system 100 further includes an input device 111 for the customer 110 to input personal information and loan amount, such as name, ID number, service company, seniority, loan amount, etc. .

The above-mentioned data reading device 120 includes a basic database 121 , a finger vein database 122 and a default database 123 . The finger vein database 122 includes a plurality of nail data, a plurality of pulse data and a plurality of finger vein data.

According to another embodiment of the present invention, the above-mentioned basic database 121 is used for storing basic data of customers.

According to another embodiment of the present invention, the above-mentioned default database 123 is used to store the default data of customers.

The credit evaluation device 130 includes a nail recognition module 131 , a pulse recognition module 132 , a finger vein recognition module 1325 , a category recognition module 133 , a risk assessment module 134 and a loan calculation module 135 .

The above-mentioned nail identification module 131 divides the nail data into a plurality of nail categories according to the shape of the nail. According to another embodiment of the present invention, please refer to Table 1 below, the nail data can be divided into straight long nails, horizontal short nails, round nails, egg-shaped nails, square nails, triangle nails, and inverse triangle nails. Nails, almond-shaped nails, sword-shaped nails and other categories.

According to another embodiment of the present invention, the above-mentioned method for obtaining the fingernail shape data is to obtain an image of the customer's fingernail shape data through image capture and recognition technology, and identify the fingernail shape data.

The above-mentioned pulse condition identification module 132 divides the pulse condition data into a plurality of pulse condition categories according to the amplitude, frequency and waveform of the pulse condition. According to another embodiment of the present invention, the pulse data can be classified into, for example, floating pulses, sinking pulses, delayed pulses, numbered pulses, rapid pulses, rapid pulses, and knotted pulses.

According to another embodiment of the present invention, the finger vein identification module 1325 uses near-infrared rays to capture a personal vein map. Its design principle is because the venous heme in the human body is in a hypoxic state. After being irradiated by near-infrared rays, the hypoxic heme will absorb the light source to present an image, and use a non-invasive photographic technology to obtain the venous blood vessel map in the body.

According to another embodiment of the present invention, the method for obtaining the pulse condition by the pulse condition identification module 132 is a transmission photoelectric method. Based on the absorption principle of light by substances, the green LED light of the device is combined with a photosensitive photodiode to illuminate the blood vessels for a period of time. Since blood is red, it can reflect red light and absorb green light. When the heart beats, the blood flow increases, and the absorption of green light increases; when the heart beats, the blood flow decreases, and the absorbed green light also decreases. Therefore, it can be measured through the absorbance of blood, and then the pulse characteristics such as pulse rate, intensity and pulse waveform width can be obtained.

Please refer to FIGS. 3-7 . FIG. 3-7 is a pulse diagram according to an embodiment of the present invention. The vertical axis of the pulse diagram is the pulse intensity, and its measurement standard is the calculated absorbance, the ratio of the incident light intensity before the light passes through the blood to the transmitted light intensity of the light passing through the blood. Because it is only a ratio, there is no unit. Since blood is red, it can reflect red light and absorb green light. When the heart beats, the blood flow increases, and the amount of green light absorption will increase with the increase of blood flow. Therefore, when the pulse intensity is stronger, the more blood flow through, and the greater the absorption of green light, the ratio of the incident light intensity before passing through the blood to the transmitted light intensity of the light passing through the blood tends to 0, Conversely, when the pulse intensity is weaker, the blood flow passing through is less, the absorption of green light is smaller, and the ratio of the incident light intensity before passing through the blood to the transmitted light intensity of the light passing through the blood tends to be 1. The horizontal axis of the pulse chart is the time of one breath. One breath means one breath and one breath. One breath lasts about 3-4 seconds, and the system takes 3.5 seconds.

Please refer to Figure 3. Figure 3 is a pulse diagram of the floating pulse. Floating pulse means that when the pulse strength is stronger, it will be in the first layer (floating) of the pulse diagram, and conversely, when the pulse strength is weaker, it will be in the third layer (sinking) of the pulse diagram. Any pulse position of the entire pulse, mainly located in the first layer (floating), or the second layer (middle), and the pulse rate is stable can be regarded as floating pulse.

Please refer to Figure 4. Figure 4 is a pulse diagram of the Shen meridian. Shen pulse refers to the pulse position of the whole pulse, mainly located in the third layer (shen), and the pulse rate is stable, it can be regarded as Shen pulse.

Please refer to Figure 5. Figure 5 is a pulse diagram of the delayed pulse. A delayed pulse is less than 60 beats per minute.

Please refer to Figure 6. Figure 6 is a pulse diagram of several pulses. The pulse count is 100-120 beats per minute.

Please refer to Figure 7, Figure 7 is the pulse diagram of the disease pulse. The rapid pulse is 120-140 beats per minute.

According to another embodiment of the present invention, when the pulse beats more than 120 beats per minute, it can be felt that the pulse suddenly stops and reappears after a while, and the interval of the stop is not fixed.

According to another embodiment of the present invention, when the pulse beat is less than 60 beats per minute, it can be felt that the pulse suddenly stops and reappears after a while, and the interval between the stops is not fixed.

The above-mentioned finger vein identification module 1325 divides the finger vein data into a plurality of finger vein categories according to the patterns of the finger veins. According to another embodiment of the present invention, please refer to Table 2 below, the finger vein data can be classified into, for example, busbar type, double rail type, I type, ladder type, network type, dichotomy type, trisection type, spider web type and other categories.

The above-mentioned category identification module 133 first identifies the customer's nail shape, pulse amplitude, frequency and waveform, and finger vein patterns, and then analyzes the customer's nail shape, finger vein patterns, and pulse characteristics of the category.

According to another embodiment of the present invention, the modeling method of the risk assessment model can be, for example, a large-scale data training mode of artificial intelligence. The massive data training mode is to divide the data of the finger vein database 122 into 4 equal parts. Each aliquot is divided into 80% training dataset and 20% testing dataset. 80% of the training data set was used for model modeling, and 20% of the test data set was used for model validity verification.

According to another embodiment of the present invention, the risk assessment model may be, for example, a convolutional neural network assessment model. The convolutional neural network evaluation model is to divide the data of the finger vein database 122 into 4 equal parts. Each aliquot is divided into 80% training dataset and 20% testing dataset. 80% of the training dataset was used for model modeling, and 20% of the test dataset was used for model validity verification. 80% of the training data set data is modeled by convolutional neural network, which is a computational model that imitates the structure and function of biological neural network.

According to another embodiment of the present invention, the method for evaluating the negative version of the above-mentioned model may be to use 20% of the test data set in each aliquot of data. 20% of the test data set data is used for customer feature similarity analysis using convolutional neural network. Calculate the average similarity value of customer characteristics using 20% of the test data in each aliquot. For example, assuming that in 4 equal parts of data, the calculated similarity of each aliquot of test data is 0.8, 0.7, 0.9, 0.8, then the average can be obtained. The similarity is as follows.

Average similarity=(0.8+0.7+0.9+0.8)/4=3.2/4=0.8

According to another embodiment of the present invention, whether the above-mentioned risk assessment model can be used is determined by checking whether the similarity of customer characteristics is greater than a threshold value. When the similarity of customer features is less than the threshold value, it means that the customer features of the test group are not similar to the customer features of the training group, and the customer features of the training group cannot be passed through. The identification and classification model established by pulse data, finger vein data and nail data can accurately identify the characteristic classification of customers. Therefore, the risk assessment model needs to be re-modeled, and the latest customer's pulse data, finger vein data and nail data need to be re-obtained to establish a risk assessment model; When the similarity of customer characteristics is greater than the threshold value, it means that the characteristics of the customers in the test group are similar to the characteristics of the customers in the training group. The risk assessment model established by the pulse data, finger vein data and nail data of the customers in the training group can accurately identify the characteristics of the customers in the test group Classification. For example, the customer's finger vein feature has 3 branch points, each branch point is divided into 2 paths, the finger vein is a dichotomous type, the pulse data has a pulse that promotes the pulse, and the nail is an inverse triangle nail.

The above-mentioned risk assessment module 134 obtains the risk value corresponding to the category according to the data of the basic database 121 and the default database 123, wherein the risk value is (AB)/(A+B), wherein A is the default within the category The number, B is the non-default number within the category.

According to another embodiment of the present invention, calculating the customer's finger vein data, pulse data and nail data for each type of successfully identified default risk value, the method is (the number of defaults - the number of non-defaults)/(the number of defaults + the number of non-defaults ), when the risk value is -1, it means that there are no defaulting customers in this category, when the risk value is 1, it means that all the defaulting customers in this category, and when the risk value is 0, it means that half of the category is defaulting customers and half are non-defaulting customers.

According to another embodiment of the present invention, please refer to Tables 3-5 below. Table 3 shows the information on the middle finger vein, pulse condition and nail information of 8 defaulting customers. Table 4 shows the middle finger vein data, pulse data and nail data of 2 non-defaulting customers. Table 5 shows the number of defaulting customers in each category.

According to another embodiment of the present invention, it is assumed that customers with high credit default risk have similar characteristic information, such as finger vein data, pulse data and nail data. Therefore, customers with high credit default risk with similar characteristic information can be identified through identification and classification, and customers with high credit default risk with similar characteristic information are prohibited from applying for loans quickly, thereby reducing the loss of banks in the rapid loan application business.

According to another embodiment of the present invention, for example, 10 customers with high credit default risk and 8 customers have finger vein characteristics with 3 branch points, each branch point is divided into 2 paths, and the pulse data has a pulse that promotes the pulse. The nails are in the shape of an inverse triangle. Therefore, as long as the identified customer's finger vein features have 3 branch points, each branch point is divided into 2 paths, which is a dichotomy type. High credit default risk.

The loan calculation module 135, when the risk value is greater than a threshold value, generates a recommended loan amount corresponding to the risk value, and when the risk value is less than or equal to a threshold value, generates the loan interest rate and loan period corresponding to the risk value.

According to another embodiment of the present invention, the following table 6 defines preset loan conditions for a customer 110 .

Table 6: Defining Pre-set Loan Conditions

According to another embodiment of the present invention, it is assumed that the similarity of the characteristic information of the customer 110 falls in No. 2, please also refer to the above Table 5, that is, the category risk is 0.34. The calculation method is as follows: Calculated loan interest rate = preset loan interest rate × (1 + (category risk - minimum category risk) / (maximum category risk - minimum category risk)), where the minimum category risk is -1 and the maximum category risk is 1. The above example loan interest rate is: 1.65%×(1+(0.34-(-1))/(1-(-1)))=1.65%×(1+(1.34)/2)=1.65%×(1+ 0.67)=1.65%×1.67=2.76%.

According to another embodiment of the present invention, when there is no default customer in this category, the category risk is -1, and the loan interest rate is: 1.65%×1+(-1-(-1))/(1-(-1)) =1.65%×(1+(0/2))=1.65%×1=1.65%.

According to another embodiment of the present invention, when the category is all defaulting customers, the category risk is 1, and the loan interest rate is: 1.65%×(1+(1-(-1)/1+(-1))=1.65 %×(1+(2/2))=1.65%×2=3.3%.

According to another embodiment of the present invention, when half of the class are defaulting customers and half are non-defaulting customers, the class risk is 0, and the loan interest rate is: 1.65%×(1+(0-(-1))/(1+ (-1))=1.65%×(1+1/2)=1.65%×1+0.5=1.65%×1.5=2.475%.

According to another embodiment of the present invention, in combination with the definitions in Table 6 above, the default loan conditions are calculated as follows: Calculated loan period = preset loan period × ((-1) × ((category risk + minimum category risk)/(maximum category risk - minimum category risk)). The minimum category risk is -1 and the maximum category risk is 1. The above example loan period is: 7×((-1)×((0.34+(-1))/1(-1))=7×((-1)×((-0.66)/2))=7 ×(-1×-0.33)=7×0.33=2.31.

According to another embodiment of the present invention, when there is no default customer in this category, the category risk is -1, and the loan period is: 7×((-1)×((-1)+(-1)/1-(- 1))=7×((-1)×((-2)/2))=7×((-1)×(-1))=7×1=7.

According to another embodiment of the present invention, when the category is all defaulting customers, the category risk is 1, and the loan period is: 7×((-1)×((1+(-1))/1-(-1 ))=7×((-1)×0/2)=7×((-1)×0)=7×0=0, when the loan period is 0, it means that the loan is not allowed due to the high risk of default.

According to another embodiment of the present invention, when half of the class are defaulting customers and half are non-defaulting customers, the class risk is 0, and the loan period is: 7×((-1)×((0+(-1))/ 1+(-1))=7×((-1)×((-1)/2))=7×((-1)×-0.5)=7×0.5=3.5.

According to another embodiment of the present invention, with the default loan conditions defined in Table 5 above, the calculation method is as follows: Calculate the recommended loan amount = the total preset loan amount × ((-1) × ((category risk + minimum category risk )/(Maximum category risk-minimum category risk)), where the minimum category risk is -1 and the maximum category risk is 1. The suggested loan amount in the above example is: 1 million × ((-1) × ((0.34+(- 1))/1-(-1))=1 million×((-1)×((-0.66)/2))=1 million×((-1)×(-0.33))=1 million×0.33 = 330,000.

According to another embodiment of the present invention, when there is no default customer in this category, the category risk is -1, and the recommended loan amount is: 1 million × ((-1) × (((-1)+(-1)) /(1-(-1))=1 million×((-1)×((-2)/2))=1 million×((-1)×(-1))=1 million×1=100 ten thousand.

According to another embodiment of the present invention, when the category is all defaulting customers, the category risk is 1, and the recommended loan amount is: 1 million×((-1)×((1+(-1))/(1 -(-1))=1 million×((-1)×(0/2))=1 million×((-1)×0)=1 million×0=0, when the recommended loan amount is 0, it means Loans are not allowed due to the high risk of default.

According to another embodiment of the present invention, when half of the category are defaulting customers and half are non-defaulting customers, the category risk is 0, and the recommended loan amount is: 1 million×((-1)×(((0)+( -1))/1-(-1))=1 million×((-1)×((-1)/2))=1 million×((-1)×(-0.5))=1 million× 0.5=500,000.

According to another embodiment of the present invention, the loan amount applied for by the customer and the recommended loan amount limit are checked, and the recommended loan amount limit is the maximum tolerable loan amount acceptable to the bank. The vein identification risk assessment system rejects the customer's application request because the customer's default risk is greater than the threshold of acceptable risk, and the customer can apply for a loan with reference to the recommended loan amount. When the customer's loan amount is less than the recommended loan amount, the risk assessment system based on finger vein identification will accept the customer's application request because the customer's default risk is less than the acceptable risk threshold, and ask the customer to confirm the loan application information.

According to another embodiment of the present invention, the finger vein identification risk assessment system 100 further includes an appropriation device 140 , after approving the loan request, the appropriation account information is checked to confirm the appropriation.

According to another embodiment of the present invention, when the credit evaluation device 130 performs the feature identification and classification of the customer 110, the convolutional neural network method is used to perform the feature identification and classification of the customer, and the finger vein data, pulse condition data and For nail data, risk assessment model training and customer 110 feature identification and classification are performed through convolution layer, linear rectification layer, pooling layer, full connection layer and loss function layer respectively.

According to another embodiment of the present invention, the main purpose of the above-mentioned convolution layer is to obtain different features of the input finger vein data, pulse condition data and nail data, respectively, using sliding convolution kernels and input images to perform element-corresponding product and sum operation. For example, as shown in Figure 8, the convolution kernel and the input image perform element-corresponding product and summation is: 0×0+0×0+0×0+0×1+1×1+0×1+0×0+ 1×1+1×0, the convolution result is 2; then slide the convolution kernel to the right to perform element-corresponding product with the input image, and the sum is: 0×0+0×0+0×0+1×1 +0×1+1×1+1×0+1×1+0×0, the convolution result is 3; and so on, the convolution kernel is corresponding to the input image element from left to right and from top to bottom. Product summation completes the convolution job.

According to another embodiment of the present invention, the main purpose of the above-mentioned linear rectification layer is to increase the nonlinear processing capability of the neural-like model, and its function is as follows: f(x)=max(0,x)

It means that when the element value is greater than 0, we will output the element, otherwise we will use 0 as the output of this function, which means that this event is not referenced.

According to another embodiment of the present invention, the main purpose of the pooling layer is to reduce the sampling number and feature size of finger vein data, pulse data and nail data. For example, the input image information is divided into an area every 2 elements, and the maximum value is taken from the elements in the area. This action will reduce the amount of data and reduce the complexity of image operations.

According to another embodiment of the present invention, the above-mentioned fully connected layer converts the output two-dimensional feature map of the previous layer into a one-dimensional vector, and each node of the fully connected layer is connected to each node of the previous layer. Multi-layer convolution and pooling layers of multiple finger vein data, pulse and nail features, classify the input finger vein, pulse and nail images, and predict the feature classification group of the risk assessment object by synthesizing the output features of the previous layer .

According to another embodiment of the present invention, the main purpose of the above-mentioned loss function (dropout) is to avoid the overfitting problem, and the output of each neuron in the hidden layer is set to zero with a probability of 50%, and the Neurons do not transmit information, which means that neurons set to 0 do not participate in forward propagation, nor do they participate in back propagation, so each time a new training data is input, it is equivalent to the network trying a different structure, But weights are shared among all structures. The loss function can reduce the interaction between hidden nodes, so the model does not depend on some local features, which makes the analysis of the model risk assessment module more accurate.

Please refer to FIG. 2 . FIG. 2 is a schematic flowchart of a method for risk assessment of finger vein identification according to an embodiment of the present invention. Please also refer to Figs. 1-2. Step 200 of Fig. 2 is the beginning.

In step 201 , a customer 110 applies for a loan to a bank, and the bank submits a loan evaluation request to the finger vein identification risk evaluation system 100 .

In step 202, the data reading device 120 obtains a customer data, the customer data includes the data of the basic database 121, the finger vein database 122 and the default database 123, wherein the data of the finger vein database 122 includes a plurality of nail shapes data, multiple pulse data and multiple finger vein data.

According to another embodiment of the present invention, the customer 110 inputs personal information and loan amount, such as name, ID number, service company, seniority, loan amount, etc., through the input device 111 .

According to another embodiment of the present invention, when the customer 110 has a bad credit record, the customer's application request will be rejected, and the customer will be informed of the reason for rejection. When the customer's 110 debt-to-income ratio is greater than the threshold, the customer's application request will be rejected and the customer will be informed of the reason for rejection.

In step 203, the nail identification module 131 divides the nail data into a plurality of nail categories according to the shape of the nail. According to another embodiment of the present invention, nail data can be classified into, for example, straight long nails, horizontal short nails, round nails, egg-shaped nails, square-shaped nails, triangle-shaped nails, inverse triangle-shaped nails, almond-shaped nails, Sword type nails and other categories.

In step 204, the pulse condition identification module 132 divides the pulse condition data into a plurality of pulse condition categories according to the amplitude, frequency and waveform of the pulse condition. According to another embodiment of the present invention, the pulse data can be classified into, for example, floating pulses, sinking pulses, delayed pulses, numbered pulses, rapid pulses, rapid pulses, and knotted pulses.

In step 205, the finger vein identification module 1325 divides the finger vein data into a plurality of finger vein categories according to the patterns of the finger veins. According to another embodiment of the present invention, Finger vein data can be classified into, for example, busbar type, double rail type, I type, ladder type, network type, dichotomy type, trisection type, spider web type, and the like.

According to another embodiment of the present invention, the above-mentioned method for obtaining the fingernail shape data is to obtain an image of the customer's fingernail shape data through image capture and recognition technology, and identify the fingernail shape data.

In step 206 , the category identification module 133 analyzes the category of the customer according to the customer's finger vein data, nail profile data and pulse data, wherein the category is the category corresponding to the finger vein category, the nail category and the pulse category.

In step 207, the risk assessment module 134, according to the data of the basic database 121 and the default database 123, finds a risk value of the category, where the risk value is (AB)/(A+B), where A is within the category The number of defaults, B is the number of non-defaults within the category.

According to another embodiment of the present invention, calculating the customer pulse data and nail data for each type of successfully identified default risk value, the method is (number of defaults - number of non-defaults)/(number of defaults + number of non-defaults), when the risk When the value is -1, it means that there are no defaulting customers in this category; when the risk value is 1, it means that all the defaulting customers in this category;

In step 208, the loan calculation module 135 determines whether the risk value is greater than a threshold value. In step 209, when the risk value is greater than a threshold value, a suggested loan amount corresponding to the risk value is generated. In step 210, when the risk value is less than or equal to a threshold value, a loan interest rate and a loan period corresponding to the risk value are generated. Finally, step 211 ends.

According to another embodiment of the present invention, please refer to the above Tables 5-6 at the same time, it is assumed that the similarity of the characteristic information of a customer 110 falls in No. 2, that is, the category risk is 0.34. Calculated loan interest rate = preset loan interest rate × (1+(category risk-minimum category risk)/(maximum category risk-minimum category risk)), where the minimum category risk is -1 and the maximum category risk is 1. The above example loan interest rate is: 1.65%×(1+(0.34-(-1))/(1-(-1)))=1.65%×(1+(1.34)/2)=1.65%×(1+ 0.67)=1.65%×1.67=2.76%.

According to another embodiment of the present invention, when there is no default customer in this category, the category risk is -1, and the loan interest rate is: 1.65%×1+(-1-(-1))/(1-(-1)) =1.65%×(1+(0/2))=1.65%×1=1.65%.

According to another embodiment of the present invention, when the category is all defaulting customers, the category risk is 1, and the loan interest rate is: 1.65%×(1+(1-(-1)/1+(-1))=1.65 %×(1+(2/2))=1.65%×2=3.3%.

According to another embodiment of the present invention, when half of the class are defaulting customers and half are non-defaulting customers, the class risk is 0, and the loan interest rate is: 1.65%×(1+(0-(-1))/(1+ (-1))=1.65%×(1+1/2)=1.65%×1+0.5=1.65%×1.5=2.475%.

According to another embodiment of the present invention, with the default loan conditions defined in Table 5 above, the calculation of loan period=preset loan period×((-1)×((category risk+minimum category risk)/(maximum category risk- Minimum category risk)). The minimum category risk is -1 and the maximum category risk is 1. The above example loan period is: 7×((-1)×((0.34+(-1))/1(-1)) =7×((-1)×((-0.66)/2))=7×(-1×-0.33)=7×0.33=2.31.

According to another embodiment of the present invention, when there is no default customer in this category, the category risk is -1, and the loan period is: 7×((-1)×((-1)+(-1)/1-(- 1))=7×((-1)×((-2)/2))=7×((-1)×(-1))=7×1=7.

According to another embodiment of the present invention, when the category is all defaulting customers, the category risk is 1, and the loan period is: 7×((-1)×((1+(-1))/1-(-1 ))=7×((-1)×0/2)=7×((-1)×0)=7×0=0, when the loan period is 0, it means that the loan is not allowed due to the high risk of default.

According to another embodiment of the present invention, when half of the class are defaulting customers and half are non-defaulting customers, the class risk is 0, and the loan period is: 7×((-1)×((0+(-1))/ 1+(-1))=7×((-1)×((-1)/2))=7×((-1)×-0.5)=7×0.5=3.5.

According to another embodiment of the present invention, in combination with the default loan conditions defined in Table 5 above, calculate the recommended loan amount = total default loan amount × ((-1) × ((category risk + minimum category risk)/(maximum Category risk-minimum category risk)), where the minimum category risk is -1 and the maximum category risk is 1. The suggested loan amount in the above example is: 1 million × ((-1) × ((0.34+(-1))/ 1-(-1))=1 million×((-1)×((-0.66)/2))=1 million×((-1)×(-0.33))=1 million×0.33=330,000.

According to another embodiment of the present invention, when there is no default customer in this category, the category risk is -1, and the recommended loan amount is: 1 million × ((-1) × (((-1)+(-1)) /(1-(-1))=1 million×((-1)×((-2)/2))=1 million×((-1)×(-1))=1 million×1=100 ten thousand.

According to another embodiment of the present invention, when the category is all defaulting customers, the category risk is 1, and the recommended loan amount is: 1 million×((-1)×((1+(-1))/(1-(-1))=1 million×((-1)×(0/2))=1 million×((- 1)×0)=1 million×0=0, when the recommended loan amount is 0, it means that the loan is not allowed due to the high risk of default.

According to another embodiment of the present invention, when half of the category are defaulting customers and half are non-defaulting customers, the category risk is 0, and the recommended loan amount is: 1 million×((-1)×(((0)+( -1))/1-(-1))=1 million×((-1)×((-1)/2))=1 million×((-1)×(-0.5))=1 million× 0.5=500,000.

According to another embodiment of the present invention, when the client default risk value is greater than the threshold value, the client's application request will be rejected, and the client will be informed of the suggested value; otherwise, when the client risk value is less than the threshold value, the client 110 will be reconfirmed whether Apply for a loan, and send customer information, loan application conditions, suggested loan application conditions and confirmation messages for customers to check, such as the customer's name, ID number, service company, seniority, loan amount, loan period, loan period and other information.

According to another embodiment of the present invention, the client 110 checks the client information and loan application condition information sent by the finger vein identification risk assessment system 100 to confirm the loan application information or set the loan amount according to the suggested loan application conditions.

According to another embodiment of the present invention, the finger vein identification risk assessment system 100 receives the loan application confirmation message from the customer 110, and performs subsequent loan operations according to the loan application conditions.

According to another embodiment of the present invention, the finger vein identification risk assessment system 100 checks whether the appropriation account is the account of the applicant who is applying for the loan. When the appropriation account is the account of the person applying for the loan, the appropriation will be entered.

According to the system and method disclosed above, the risk assessment system based on finger vein identification technology evaluates the default risk of customers by calculating the number of customers with high default credit risk of each type, and can dynamically adjust the default risk value of each type at any time as the default status of customers changes. Different from the traditional credit risk assessment model, the establishment and adjustment of the model requires a lot of time-consuming testing and verification, and the credit risk assessment model cannot be adjusted in time to cope with changes in default conditions.

Although the embodiments of the present invention are disclosed as above, they are not intended to limit the present invention. Those who are familiar with the art can make various changes according to the spirit of the present invention without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by those defined by the following patent applications.

100: Finger Vein Identification Risk Assessment System

110: Customers

111: Input device

120: Data reading device

121: Basic database

122: Finger Vein Database

123: Default Database

130: Credit Evaluation Device

131: Nail recognition module

132: Pulse recognition module

1325: Finger Vein Recognition Module

133: Category identification module

134: Risk Assessment Module

135: Loan Calculation Module

140: Appropriation Device

Claims (10)

A finger-vein identification risk assessment system, when a customer applies for a loan to a bank, uses the finger-vein identification risk assessment system to assess the fraud risk of the loan, the finger-vein identification risk assessment system comprises: a data reading device comprising: a basic database, a finger vein database and a default database, wherein the finger vein database includes a plurality of nail data, a plurality of pulse data and a plurality of finger vein data; and a credit evaluation device, including: a nail identification A module, which divides the nail data into a plurality of nail categories according to the shape of the nail; a pulse recognition module, which divides the pulse data into a plurality of pulse categories according to the amplitude, frequency and waveform of the pulse; a finger vein recognition module, according to The pattern of the finger veins divides the finger vein data into a plurality of finger vein categories; a category identification module uses a convolutional neural network to model a training data set, a risk assessment model, and uses the risk assessment model Identify the shape of the client's nails, the amplitude, frequency and waveform of the pulse, and the veins of the finger veins, so as to analyze the category to which one of the clients belongs, where the category corresponds to the nail category, the pulse category, and the finger vein category. Category; a risk assessment module, according to the data of the basic database and the default database, to obtain a risk value corresponding to the category, where the risk value is (AB)/(A+B), where A is the The number of defaults within the category, and B is the number of non-defaults within the category; a loan calculation module, when the risk value is greater than a threshold value, generates a recommended loan amount corresponding to the risk value. The risk value is less than or equal to a threshold value, A loan interest rate and a loan period corresponding to the value at risk are generated. A finger vein identification risk assessment system as described in claim 1, further comprising an appropriation device, after approving the loan request, checking the appropriation account information to confirm the appropriation. A finger vein identification risk assessment system as claimed in claim 1, wherein the basic database is for storing a customer's basic data. A finger vein identification risk assessment system as claimed in claim 1, wherein the default database is to store a default data. The finger vein identification risk assessment system as claimed in claim 1, further comprising an input device for inputting the nail data, the pulse condition data and the finger vein data of the client. A finger vein identification risk assessment method is executed by a finger vein identification risk assessment system. The method includes: obtaining an assessment request from a client applying for a loan from a bank; obtaining client information, the client information including a basic database , Data of a finger vein database and a default database, wherein the data of the finger vein database includes a plurality of nail shape data, a plurality of pulse data and a plurality of finger vein data; these nail data are divided into plural numbers according to the shape of the nail According to the pulse amplitude, frequency and waveform, the pulse data are divided into multiple pulse categories; according to the pattern of the finger veins, the finger vein data are divided into a plurality of finger vein categories; according to the customer's nail shape, pulse The amplitude, frequency and waveform of the client, as well as the vein pattern of the finger veins, use a convolutional neural network and training data set to model a risk assessment model, and use the risk assessment model to identify the client's nail shape, pulse amplitude, frequency and frequency. Waveform and pattern of finger veins to analyze one of the categories of the customer, wherein the category is the category corresponding to the nail category, the pulse category and the finger vein category; according to the basic database and the default database data, obtain a value at risk of the category, where the risk value is (AB)/(A+B), where A is the number of defaults in the category, and B is the number of non-defaults in the category; When the risk value is greater than a threshold value, a recommended loan amount corresponding to the risk value is generated; when the risk value is less than or equal to a threshold value, a loan interest rate and a loan period corresponding to the risk value are generated. The finger vein identification risk assessment method as described in claim 6, further comprising, after approving the assessment request for the loan, checking the appropriation account information to confirm the appropriation. The risk assessment method for finger vein identification as claimed in claim 6, wherein the basic database is for storing a customer's basic data. A finger vein identification risk assessment method as claimed in claim 6, wherein the default database is to store a default data. The risk assessment method for finger vein identification as described in claim 6, further comprising receiving the nail data, the pulse condition data and the finger vein data of the client.

Images