CN114550382B

CN114550382B - ATM cashbox management method and system

Info

Publication number: CN114550382B
Application number: CN202210093615.8A
Authority: CN
Inventors: 陈兵
Original assignee: Nanjing Financial Data Systems Ltd By Share Ltd
Current assignee: Nanjing Financial Data Systems Ltd By Share Ltd
Priority date: 2022-01-26
Filing date: 2022-01-26
Publication date: 2023-07-04
Anticipated expiration: 2042-01-26
Also published as: CN114550382A

Abstract

The invention discloses a cashbox management method and a cashbox management system for an ATM, which can be used for intelligently inspecting a cashbox, finding cracks of the cashbox in time and maintaining or replacing the cashbox, and comprises the following steps: step S1: extracting and inputting the use time of the cashbox; step S2: comparing the cashbox service time with the cashbox maintenance period data; step S3: when the service time of the cashbox meets the maintenance period, generating maintenance prompt information; step S4: and (3) carrying out crack inspection on the cashbox, entering the next service cycle if no crack occurs, and carrying out recovery treatment if the crack occurs.

Description

ATM cashbox management method and system

Technical Field

The invention relates to the technical field of ATM cashbox management, in particular to an ATM cashbox management method and system.

Background

ATM is a small machine set up in different places by bank, and the bank card is used for withdrawing money, depositing deposit and transferring money. An ATM has three to four cashboxes like drawers, and a recycling box (commonly called waste cashbox), and when the cash is fed and discharged, the cash which does not meet the recycling standard is received into the recycling box. The paper money is stacked in the cashbox from back to front in a stacking way, and the innermost part is a paper money outlet.

The current common maintenance modes of commercial bank ATMs mainly comprise periodic maintenance and fault maintenance, and in the use process, the cashbox is required to be replaced frequently. In the long-term use process of the cashbox, the cashbox is continuously disassembled, carried, installed and opened, such as banknote adding, banknote distributing and the like by an administrator. The phenomena of box body abrasion, damage and the like can occur in the process, and the regular management and inspection are also needed to ensure that the cashbox can be used continuously and ensure the normal operation of the ATM.

Disclosure of Invention

In order to overcome the defects of the prior art, one of the purposes of the invention is to provide an ATM cashbox management method which can be used for intelligently checking a cashbox, finding cracks of the cashbox in time and maintaining or replacing the cashbox.

One of the purposes of the invention is realized by adopting the following technical mode:

compared with the prior art, the invention has the advantages that:

an ATM cashbox management method comprising the steps of:

step S1: extracting and inputting the use time of the cashbox;

step S2: comparing the cashbox service time with the cashbox maintenance period data;

step S3: when the service time of the cashbox meets the maintenance period, generating maintenance prompt information;

step S4: and (3) carrying out crack inspection on the cashbox, entering the next service cycle if no crack occurs, and carrying out recovery treatment if the crack occurs.

Further, the crack inspection method in step S4 is as follows:

step S310: extracting picture information on the surface of the cashbox body, and obtaining an edge thermodynamic diagram according to the collected whole picture information;

step S320: obtaining flaw drawing vectors of the picture squares of each picture according to the edge thermodynamic diagram;

step S330: classifying the conclusion according to the set of the drawing vectors of the grid flaws of the whole picture to obtain undetermined cracks and the credibility of the undetermined cracks;

step S340: the final crack is obtained according to the edge thermodynamic diagram and the undetermined cracks with different credibility.

Further, the picture information obtaining method in step S310 is as follows:

s311: erecting a camera, and circumferentially distributing a plurality of light sources around the camera;

s312: each light source emits red, green and blue light respectively, and the direction of the light source faces towards the camera to realize overlooking irradiation of the surface of the cashbox body in the center;

s313: the camera captures a picture of the surface of the cashbox body every time the light source emits light of one color.

Further, the edge thermodynamic diagram extraction method in step S310 is as follows:

s314: extracting an edge single-color image of each picture in the whole pictures through an edge checking algorithm;

s315: extracting the average value of all gray scale values of all the picture grids on all the edge single-color pictures, and extracting a single-channel picture formed by the average value of all the picture grids;

s316: an edge thermodynamic diagram is obtained.

Further, the method for obtaining the defect drawing vector in step S320 is as follows:

s321: marking the gray scale value of all the picture grids with the gray scale value larger than 0 on the edge thermodynamic diagram as 1, and keeping the gray scale values of other picture grids unchanged to obtain an edge covering diagram;

s322: extracting any picture square grid Q with a gray level value of 1, extracting the gray level value of the picture square grid on an mth picture, and constructing a window by taking the picture square grid Q as a center;

s323: extracting gray scale values of all picture grids on an mth picture in a window, and extracting modes of the gray scale values;

s324: the absolute value of the gray level value and the mode difference value is the flaw degree of the picture square Q on the mth picture;

s325: obtaining all possible value conclusions of the flaw degree according to the defect degree of all the picture squares on all the pictures, and arranging the X value conclusions from small to large on the assumption that the X value conclusions exist; and then counting according to the flaw degree set of the picture square grid Q on the whole picture to obtain the probability of the X-th value conclusion appearing in the set, and combining the probability of the X value conclusions to form an X-dimensional vector to obtain the flaw drawing vector of each picture square grid Q.

The undetermined crack obtaining method in step S330 is as follows:

s331: performing mean shift set classification on flaw depiction vectors of all the picture squares, and obtaining all the first types;

s332: extracting all picture squares corresponding to all flaw drawing vectors in each first type;

s333: the coordinates of all the picture squares are subjected to aggregate classification through an aggregate classification algorithm to obtain all the second types;

s334: treating the whole picture grids in each second type as an edge formed by the whole picture grids in each second type, wherein the positions of the whole picture grids are continuously distributed and have similar flaw drawing vectors;

s335: when it is not possible to determine whether the edge is a crack or a footprint is used, the edge is referred to as an undetermined crack.

Further, the method for obtaining the reliability of the undetermined crack in step S330 is as follows:

s336: the average value of the characteristic values of the flaw depiction vectors of all the picture squares on each undetermined crack is regarded as the credibility of each undetermined crack.

Further, the method for obtaining the final crack in step S340 is as follows:

s341: when any one of the blended edges simultaneously comprises only a first undetermined crack and a third undetermined crack, attributing the third undetermined crack as the first undetermined crack;

s342: when the blended edge simultaneously comprises only the second undetermined crack and the third undetermined crack, attributing the third undetermined crack as the second undetermined crack;

s343: when the blended edge comprises a first undetermined crack, a second undetermined crack and a third undetermined crack at the same time, calculating the undetermined crack to which the third undetermined crack belongs according to the first undetermined crack and the second undetermined crack;

s354: when the blended edge includes only the third unidentified crack, the third unidentified crack is obtained as an unidentified crack, and the first unidentified crack as a whole is regarded as a detected final crack.

An ATM cashbox management system comprises a network module, a server, an ATM cashbox management system and a picture acquisition device;

the network module is used for communication, data transmission and interaction among the ATM cashbox management system, the picture acquisition device and the server;

the picture acquisition device is used for acquiring the picture information of the cashbox;

the ATM cashbox management system manages cashbox maintenance work.

The beneficial effects of the invention are as follows:

through an informatization technology, the crack condition of the cashbox body of the ATM is automatically searched and analyzed, and particularly, when the box body is carried out with actions such as carrying, moving, dragging, picking and placing in the using process, the abnormal contact surface of the crystal phase structure of the box body surface can be caused; because of the physical structures of the metal crystals and the crystal particles, the cracks are more sensitive to the illumination direction and the light color compared with the use of the imprinting, so that the intelligent recognition can be carried out on the surface picture friend-making system under different illumination directions and light source colors by the invention, thereby being convenient for the maintenance of an administrator and ensuring the normal operation of the ATM.

The foregoing description is only a brief summary of the technical aspects of the present invention, and it is intended that the technical aspects of the present invention be implemented according to the content of the present specification, and that the foregoing and other objects, features and advantages of the present invention be more apparent and understood, and that the following detailed description of the preferred embodiments will be given in conjunction with the accompanying drawings.

Drawings

Figure 1 is a block diagram of an ATM cashbox management system.

Description of the embodiments

The present invention will be further described with reference to the accompanying drawings and detailed description, which should be understood that, without conflict, new embodiments may be made by any combination of the embodiments described below or technical features. Unless defined otherwise, all technical and objective terms used herein have the same meaning as commonly interpreted by one skilled in the art to which the present invention pertains. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed names.

Referring to fig. 1, an ATM cashbox management method includes the steps of:

step S1: extracting and inputting the use time of the cashbox;

The crack inspection method in step S4 is as follows:

step S320: obtaining a flaw drawing vector of a picture square of each picture according to the edge thermodynamic diagram (the picture square refers to a minimum unit in a picture represented by a digital sequence, namely pixels of the picture);

When the box body is used for carrying, moving, dragging, picking and placing and the like, the contact surface of the crystal phase structure on the surface of the box body is abnormal, for example, the contact surface of the metal crystals is misplaced due to the stress between the metal crystals, and cracks are formed in macroscopic appearance; because of the physical structure of the metal crystals and crystal particles, such as the direction of the surface of the metal crystals and the crystal form of the metal crystals, the cracks are distinguished from the use of marks, and the cracks are more sensitive to the illumination direction and the color of light, so that the images of the surface of the box body under different illumination directions and light source colors need to be acquired through a camera.

And (3) obtaining a first undetermined crack from the undetermined crack with the reliability being greater than a first critical value, and describing a second undetermined crack with the reliability being less than a second critical value, and a third undetermined crack with the reliability being less than or equal to the first critical value and greater than or equal to the second critical value by taking the first critical value being equal to 0.8 and the second critical value being equal to 0.25 as an example.

Obtaining each edge on the edge thermodynamic diagram through an edge checking algorithm to obtain a blended edge, wherein the blended edge is also a set of some picture squares, and the picture squares of the blended edge can be from cracks or used marks, and the phenomenon that the used marks and the cracks appear along with each other exists, so that the blended edge can have the phenomenon that the cracks and the used marks appear along with each other; the undetermined crack in this example is also a collection of picture panes. A blend edge is considered to include an undetermined edge when more than eighty-five percent of the picture tiles in the undetermined edge are in the blend edge, and not otherwise.

The picture information obtaining method in step S310 is as follows:

s312: each light source respectively emits red, green and blue light, the light source direction faces the camera to realize overlooking irradiation of the surface of the cashbox body in the center, in the embodiment, 21 light sources are taken as an example for description, each light source is composed of three LED light emitting diodes, and the three light sources respectively emit red, green and blue light;

s313: when the light source emits light of one color, the camera collects a picture of the surface of the cashbox body, and 63 pictures are collected in total. Each picture is a gray-scale picture, and the gray-scale value of each picture square on the gray-scale picture is an integer from 0 to 255.

Cracks are edges in the picture, marks are also used in the picture, marks and cracks are often used together, and a common method cannot distinguish the two edges; the former is relatively noticeable in certain light source directions and illumination colors, while the latter is less noticeable in any light source directions and illumination colors, and the present embodiment, whether it is an inspection apparatus or an inspection object, has some great intervention and uncertainty, and the present embodiment employs the following method to obtain accurate cracking.

For convenience of description, flaws on the picture will be obtained together with the use of the printing. The construction edges of the surface of the case are present and can be checked by conventional methods, for example, whether the edge is a construction edge can be obtained by a difference in brightness between the pictures on both sides of the edge, if the difference is a construction edge, if the difference is no longer a crack or a imprinted edge is used, so that the construction edge is not investigated later.

The edge thermodynamic diagram extraction method in step S310 is as follows:

s314: extracting an edge single-color image of each picture in all pictures by an edge checking algorithm (such as a gamma edge detection algorithm), wherein all picture grids with gray scale value of 1 on the edge single-color image are the extracted edges, and the picture grids with gray scale value of 0 are not on the edges;

s315: extracting the average value of all gray scale values of all the picture grids on all the edge single-color pictures;

s323: and extracting a single-channel picture formed by the average value of all the picture squares to obtain an edge thermodynamic diagram.

The method for obtaining the defect drawing vector in step S320 is as follows:

the picture grids with the gray level value of 1 on the edge covering graph represent cracks or picture grids with printed marks, so that the calculation amount is reduced, and only the picture grids with the gray level value of 1 on the edge covering graph are researched;

s323: the gray scale values of all the picture squares on the mth picture in the window are extracted, and in this embodiment, the cracks or the using marks are considered to be only a small part in the window, and most of the picture squares in the window are not on the cracks. Extracting the mode of the gray scale value, and representing the gray scale value of a crack-free or printing place by the mode;

the larger the value, the more distinct the crack or gray scale value at the use print and the crack or gray scale value at the use print are in the local area under the influence of illumination. By the method, the defect degree is calculated, so that on one hand, the influence of different brightness caused by illumination at different positions of the picture is avoided, and on the other hand, the defect feature can be accurately depicted.

S325: obtaining all possible value conclusions of the flaw degree according to the defect degree of all the picture squares on all the pictures, and arranging the X value conclusions from small to large on the assumption that the X value conclusions exist; and then counting according to the flaw degree set of the picture square grid Q on the whole picture to obtain the probability of occurrence of the X-th value conclusion in the set, combining the probability of occurrence of the X-th value conclusion into an X-dimensional vector, and obtaining the flaw depicting vector of each picture square grid Q, wherein the flaw depicting vector is used for representing the flaw degree statistics and distribution phenomenon of each picture square grid under different light source directions and light source colors.

The undetermined crack obtaining method in step S330 is as follows:

s322: the flaw drawing vectors in each first type have larger approximation degree, and the whole picture square corresponding to the whole flaw drawing vectors in each first type is extracted;

s333: carrying out set classification on the coordinates of all the picture squares by a set classification algorithm (or DBSCAN density clustering algorithm) to obtain all the second types;

s324: the second type is a set of picture panes, the whole picture panes in each second type are regarded as being continuously distributed in position and have similar flaw drawing vectors, and one edge formed by the whole picture panes in each second type is formed;

The method for obtaining the reliability of the undetermined crack in step S330 is as follows:

s336: the average value of the characteristic values of the flaw depiction vectors of all the picture squares on each undetermined crack is regarded as the credibility of each undetermined crack. The more reliable the undetermined crack is, the more likely it is that the less reliable the undetermined crack is, the more likely it is that printing will be used.

The flaw drawing vectors of all the picture grids are classified by means of mean shift sets, all the first types are obtained, the flaw drawing vectors in each first type have larger approximation degree, all the picture grids corresponding to the all the flaw drawing vectors in each first type are extracted, the coordinates of all the picture grids are classified by means of DBSCAM set classification algorithm, all the second types are obtained, the second types are sets of the picture grids, all the picture grids in each second type are regarded as being continuously distributed in position and have similar flaw drawing vectors, and then one edge formed by all the picture grids in each second type cannot be determined temporarily, so that the edge is a crack or a print is used, and the edge is called as an undetermined crack.

For the picture pane Q, if the picture pane Q is on a crack, the picture pane has smaller flaw degree on most pictures and larger flaw degree on less pictures, so that the flaw drawing vector of the picture pane Q has larger values of the first few dimensions, smaller values of the middle few dimensions and larger values of the last few dimensions; if it is on a usage footprint, the picture pane has a larger defect level on almost the whole picture, so that the values of the first and the middle dimensions in the defect delineation vector of the picture pane Q are smaller and the values of the last dimensions are larger; based on this, the flaws of the picture pane Q are specifically depicted by the following method:

fitting a one-dimensional Gaussian mixture module through an EM algorithm, wherein the number of the Gaussian mixture modules in the Gaussian mixture module is set to be 2; respectively extracting average values y1 and y2 of the two sub-Gaussian modules; if y1 differs from y2 more, then it is stated that picture pane Q is more likely to be on a crack; if y1 and y2 differ less, or even are equal, then it is stated that the picture pane Q is more likely to be on the usage footprint; in this embodiment, sq= (y 1-y 2) 2/X2 is used to represent the characteristic value of the defect drawing vector of the picture square Q.

The final crack in step S340 is obtained as follows, where the final crack refers to a crack caused by abnormal alignment of metal crystals or contact surfaces of crystal grains on the cashbox body:

s341: when any one of the blended edges simultaneously comprises only a first undetermined crack and a third undetermined crack, attributing the third undetermined crack as the first undetermined crack; since no second undetermined crack is present, it is indicated that the third undetermined crack is likely to occur due to deviation and noise, which are still the first undetermined crack, and the noise and deviation occur due to the simplicity of the image acquisition device, the large noise on the surface of the inspected object and the deviation of the flaw depiction vector calculation of the image square;

s342: when the blended edge simultaneously comprises only the second undetermined crack and the third undetermined crack, attributing the third undetermined crack as the second undetermined crack; since no first undefined crack has occurred, indicating that the third undefined crack is likely to occur due to the deviation and noise, it remains a second undefined crack;

s343: when the blended edge comprises a first undetermined crack, a second undetermined crack and a third undetermined crack at the same time, calculating the undetermined crack to which the third undetermined crack belongs according to the first undetermined crack and the second undetermined crack; the specific method comprises the following steps:

firstly, combining flaw drawing vectors of all picture grids on a first undetermined crack and flaw drawing vectors of all picture grids on a second undetermined crack and flaw drawing vectors of all picture grids on a third undetermined crack into a set, then carrying out dimension reduction on all flaw drawing vectors in the set through a QCA algorithm, taking dimension reduction as 5 as an example, obtaining a dimension reduction conclusion of the flaw drawing vectors of all picture grids on the first undetermined crack, marking the dimension reduction conclusion as a first vector set Z1, obtaining a dimension reduction conclusion of the flaw drawing vectors of all picture grids on the second undetermined crack, marking the dimension reduction conclusion as a second vector set Z2, obtaining a dimension reduction conclusion of the flaw drawing vectors of all picture grids on the third undetermined crack, and marking the dimension reduction conclusion as a third vector set Z3; the intention of dimension reduction is to reduce the amount of calculation at the back on one hand and exclude the intervention of noise points on the other hand, so that the calculation at the back has stability and the accuracy of final crack extraction is ensured.

If the flaw depiction vectors of the whole picture square of Z3 are formed by cumulatively fusing the flaw depiction vectors in Z1 and Z2, then the third undetermined crack is shown to exist due to the concurrent generation of the crack and the using print, namely the third undetermined crack fuses with both the crack and the using print; if the flaw depiction vectors of the global picture square of Z3 cannot be cumulatively blended from the flaw depiction vectors in Z1 and Z2, then it is indicated that the third unidentified crack is an unidentified crack.

Multiplying the flaw depiction vector in Z1 by a first undetermined matrix, and obtaining a first vector set by the conclusion, wherein the first undetermined matrix is an unknown matrix to be determined and is a diagonal matrix with the size of 5 multiplied by 5; multiplying the flaw depiction vectors of all the picture squares on the second undetermined crack by a second undetermined matrix to obtain a second vector set, wherein the second undetermined matrix is an unknown diagonal matrix with the size of 5 multiplied by 5 to be determined; combining Z1 and Z2 into a set to obtain a blending set; and calculating the average value and the covariance matrix of the blending set, and calculating the average value and the covariance matrix of the third vector set.

Firstly, constructing objective functions related to a first to-be-determined matrix and a second to-be-determined matrix, and then obtaining the first to-be-determined matrix and the second to-be-determined matrix through a gradient descent method; after the first to-be-determined matrix and the second to-be-determined matrix are obtained, the difference between the blending set and the third vector set is obtained through a maximum average value difference algorithm, and the larger and smaller the difference is, the defect drawing vectors of all the picture squares of Z3 are indicated to be approximately formed by accumulating and blending the defect drawing vectors in Z1 and Z2; the greater this difference, the more likely the flaw depiction vectors for the entire picture grid of Z3 are to be blended from the flaw depiction vectors in Z1 and Z2. When the difference is less than the third threshold, the third unidentified crack is considered to be present due to the crack and the usage trace being co-produced, and then the third unidentified crack is attributed to the first unidentified crack; and when the difference is greater than or equal to a third critical value, obtaining an uncertain crack from the third uncertain crack.

S354: when the blended edge only comprises a third undetermined crack, the third undetermined crack is obtained to obtain an undetermined crack, after the whole undetermined crack is extracted, the first critical value is reduced to 0.25 times of the original critical value, the second critical value is increased to 0.25 times of the original critical value, and the first undetermined crack, the second undetermined crack and the third undetermined crack are calculated again according to the steps only aiming at the whole picture square on the whole undetermined crack until the undetermined crack does not exist. The first unidentified crack in the whole is regarded as the detected final crack. And taking the position and the length of each final crack as a check conclusion, wherein the length of each final crack refers to the number of picture squares included on the final crack.

the ATM cashbox management system manages cashbox maintenance work.

The above embodiments are only preferred embodiments of the present invention, and the scope of the present invention is not limited thereto, but all the non-realistic transformations and substitutions made by those skilled in the art based on the present invention are included in the scope of the present invention.

Claims

1. A ATM cashbox management method is characterized in that: the method comprises the following steps:

step S1: extracting and inputting the use time of the cashbox;

step S4: carrying out crack inspection on the cashbox, entering the next service cycle if no crack appears, and carrying out recovery treatment if the crack appears;

the crack inspection method in the step S4 is as follows:

step S340: obtaining an ultimate crack according to the edge thermodynamic diagram and the undetermined cracks with different credibility;

the picture information obtaining method in the step S310 is as follows:

s313: the camera collects a picture of the surface of the cashbox body every time the light source emits light with one color;

the edge thermodynamic diagram extraction method in step S310 is as follows:

s316: an edge thermodynamic diagram is obtained.

2. The ATM cashbox management method of claim 1, wherein: the method for obtaining the defect drawing vector in step S320 is as follows:

3. The ATM cashbox management method of claim 2, wherein: the undetermined crack obtaining method in step S330 is as follows:

4. A method of ATM cashbox management as in claim 3 wherein: the method for obtaining the reliability of the undetermined crack in step S330 is as follows:

5. The ATM cashbox management method of claim 4, wherein: the final crack obtaining method in step S340 is as follows:

6. A system for an ATM cashbox management method of claim 5, wherein: the system comprises a network module, a server, an ATM cashbox management system and a picture acquisition device;

the ATM cashbox management system manages cashbox maintenance work.