The method and device of bank note classificationTechnical field
The present invention relates to the method and devices that field of computer technology more particularly to a kind of bank note are classified.
Background technology
For the ease of arranging a large amount of bank note, need to carry out taxonomic revision to the bank note of different editions.Present bank note versionThis identification technology is based primarily upon manual identified, after the completion of the identical bank note of face amount is classified automatically, then artificially carries out versionThis classification.However, this method needs to waste a large amount of human cost, and due to by manually being differentiated, it is less efficient simultaneouslyAnd it is easy error.
For Hongkong dollar bank note, corresponding denomination size include 10 yuan, 20 yuan, 50 yuan, 100 yuan, 500 yuan,100 yuan, and each denomination size has the different versions of different issued by banks, and which results in Hongkong dollar bank note is differentAmounting under denomination version of different sizes has 31 versions, corresponding version, denominations difficulty higher.
That is, for the method (in particular for Hongkong dollar) of paper money recognition classification, there are classification effectivenesses in the prior artProblem low, accuracy is low.
Invention content
Based on this, there is classification effect (in particular for Hongkong dollar) to solve the method that traditional technology classifies for paper money recognitionThe technical problem that rate is low, accuracy is low, spy propose a kind of method of bank note classification.
A kind of method of bank note classification, including：
Obtain the target image of bank note to be identified；
The target signature data of the target image are extracted, the target signature data include denomination characteristic, timeAt least one of characteristic and/or distribution row characteristic；
The target signature data are identified, determine target bank note classification letter corresponding with the target signature dataBreath；
The target category of the bank note to be identified is determined according to the target bank note classification information.
Optionally, the target image includes obtaining institute by contact-type image sensor in one of the embodiments,State the target RGB color image of bank note to be identified；
The step of target signature data of the extraction target image, further include：
The target RGB color image is converted into target HSV images, obtains the H component datas of the target HSV imagesAs the denomination characteristic；
It is described that the target signature data are identified, determine target paper money corresponding with the target signature dataThe step of other information, further include：
It is searched in the H component informations for each denomination that the feature templates include matched with the denomination characteristicH component informations determine denomination corresponding with the matched H component informations as target bank note type as target signature templateInformation, the feature templates include the H component informations of each denomination；.
Optionally, the target image includes obtaining institute by contact-type image sensor in one of the embodiments,State the Infrared Targets transmission image of bank note to be identified；
The step of target signature data of the extraction target image, further include：Obtain the Infrared Targets transmissionThe quantity of safety line is as target signature data in image；
It is described that the target signature data are identified, determine target paper money corresponding with the target signature dataThe step of other information, further include：
According to the quantity of safety line in the target signature data, time version corresponding with the bank note to be detected is determinedInformation is as target bank note type information.
Optionally, in one of the embodiments, the target signature data of the extraction target image the step of alsoIncluding：
It determines the topography of the predeterminated position for the bank note reversed image that the target image includes, obtains and the partThe corresponding binary image of image；
It is described that the target signature data are identified, determine target paper money corresponding with the target signature dataThe step of other information, further include：
What it is by binary image input training in advance includes the various graders for issuing capable characteristic image informationIn, the emission of banknotes row parameter to be detected is exported as target bank note type information.
Optionally, in one of the embodiments, it is described by the binary image input in advance training include eachBefore step in the grader of the characteristic image information of kind distribution row, further include：
It obtains including the various predeterminated position image informations for issuing row bank note using convolutional neural networks classifier training,And the predeterminated position image information of the various distribution row bank note is stored to the grader.
Optionally, in one of the embodiments, the target signature data of the extraction target image the step of itBefore, further include：
The target image is normalized, will be waited for described in the target image replacement after the normalizedIdentify that the target image of bank note, the image size of the target image after the normalized meet preset picture size ginsengNumber.
In addition, to solve the method (in particular for Hongkong dollar) that traditional technology classifies for paper money recognition, there are classification effectivenessesA kind of technical problem low, accuracy is low, it is also proposed that device of bank note classification.
A kind of device of bank note classification, including：
Target image acquisition module, the target image for obtaining bank note to be identified；
Characteristic extraction module, the target signature data for extracting the target image, the target signature dataIncluding at least one of denomination characteristic, time characteristic and/or distribution row characteristic；
Bank note classification information determination module determines special with the target for the target signature data to be identifiedLevy the corresponding target bank note classification information of data；
Bank note classification identification module, the target for determining the bank note to be identified according to the target bank note classification informationClassification.
Optionally, the target image includes obtaining institute by contact-type image sensor in one of the embodiments,State the target RGB color image of bank note to be identified, the target that obtains the bank note to be identified by contact-type image sensor it is redOuter transmission image；
The characteristic extraction module is additionally operable to the target RGB color image being converted into target HSV images, obtainsThe H component datas of the target HSV images are as the denomination characteristic；Obtain safety in the Infrared Targets transmission imageThe quantity of line is as target signature data；Determine the Local map of the predeterminated position for the bank note reversed image that the target image includesPicture obtains binary image corresponding with the topography；
The bank note classification information determination module is additionally operable to the H component informations for each denomination for including in the feature templatesMiddle lookup, as target signature template, determines and the matched H components with the matched H component informations of the denomination characteristicThe corresponding denomination of information is as target bank note type information；According to the quantity of safety line in the target signature data, determine withThe corresponding time version information of the bank note to be detected is as target bank note type information；Binary image input is advanceTrained includes to export the emission of banknotes row parameter to be detected in the various graders for issuing capable characteristic image informationAs target bank note type information.
In addition, also, in the present invention, it is proposed that a kind of computer equipment, including memory, processor and it is stored in describedOn memory and the computer program that can run on the processor, which is characterized in that the processor executes the calculatingThe method that foregoing bank note classification is realized when machine program.
Also, in the present invention, it is proposed that a kind of computer readable storage medium, including computer instruction, when the computerWhen instruction is run on computers so that the method that computer executes foregoing bank note classification.
Implement the embodiment of the present invention, will have the advantages that：
After the method and apparatus for using above-mentioned bank note classification, in the process that the classification of the bank note such as Hongkong dollar is identifiedIn, corresponding characteristic in the image of bank note to be identified can be obtained according to the feature that be based on of bank note classification identification, then rootCorresponding classification information is identified respectively according to each characteristic, so that it is determined that the classification where bank note to be identified.Especially needleFor the Hongkong dollar bank note more to classification, from 3 dimensions such as denomination size, time version, bank of issue respectively to bank note classificationIt is identified, effectively the classification of Hongkong dollar bank note can be identified, and improve the accuracy of paper money recognition.
Description of the drawings
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show belowThere is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only thisSome embodiments of invention for those of ordinary skill in the art without creative efforts, can be withObtain other attached drawings according to these attached drawings.
Wherein：
Fig. 1 is a kind of flow diagram of the method for bank note classification in one embodiment；
Fig. 2 determines the stream of the denomination size of bank note according to denomination characteristic in classifying for a kind of bank note in one embodimentJourney schematic diagram；
Fig. 3 determines the version year information of bank note according to time characteristic in classifying for a kind of bank note in one embodimentFlow diagram；
Fig. 4 is a kind of banknote image schematic diagram in one embodiment；
Fig. 5 is a kind of banknote image schematic diagram in one embodiment；
Fig. 6 is to determine emission of banknotes bank according to distribution row characteristic in a kind of classification of bank note in one embodimentFlow diagram；
Fig. 7 is to determine emission of banknotes bank according to distribution row characteristic in a kind of classification of bank note in one embodimentFlow diagram；
Fig. 8 is a kind of structural schematic diagram of the device of bank note classification in one embodiment；
Fig. 9 is the structural schematic diagram of the computer equipment for the method that aforementioned bank note classification is run in one embodiment.
Specific implementation mode
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, completeSite preparation describes, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based onEmbodiment in the present invention, those of ordinary skill in the art are obtained every other without creative effortsEmbodiment shall fall within the protection scope of the present invention.
It is low, accurate that there are classification effectivenesses to solve the method (in particular for Hongkong dollar) classified for paper money recognition of traditional technologyThe low technical problem of exactness, in the present embodiment, spy propose a kind of method of bank note classification, and the realization of this method can be dependent onComputer program, the computer program can run on the computer system based on von Neumann system, the computer programIt can be the application program that bank note progress currency type, classification are identified or are classified to bank note.The computer system can be withIt is the server apparatus such as smart mobile phone, tablet computer, PC, cash inspecting machine for running above computer program.
It should be noted that in the present embodiment, the object that the method for above-mentioned bank note classification is classified can be portCoin can also be the bank note of other currency types.
Specifically, as shown in Fig. 2, the method for above-mentioned bank note classification includes the following steps S102-S108：
Step S102：Obtain the target image of bank note to be identified.
In the present embodiment, the image of bank note to be identified is detected and identifier determines the class of the bank note to be identifiedNot or classification, for example, identify the corresponding denomination size of the bank note to be identified, distribution row (bank for issuing the bank note) andThe version time limit, so that it is determined that the bank note is specifically classified.For example, for comparison Hongkong dollar bank note, it is thus necessary to determine that the bank note is correspondingDenomination size is 10 yuan, 20 yuan, 50 yuan, 100 yuan, 500 yuan or 1000 yuan；Also, the version of the Hongkong dollar bank note is also predefined,For example, belonging to 2010 editions or 2003 editions；Further, it is also necessary to determine the emission of banknotes bank, for example, 10 yuan notes byHong Kong Hong Kong Monetary Authority issues, but 20 yuan, 50 yuan, 100 yuan, 500 yuan, 1000 yuan of the bank of issue include Standard Chartered Bank, Huifeng silverRow and Bank of China (Hong Kong).
In the present embodiment, may include feux rouges image, green light image, blue light to the target image of bank note to be identified acquisitionImage, ultraviolet reflectance image, infrared transmission image, green light transmission image, infrared external reflection image, also, wherein feux rouges image, greenLight image, blue light images, ultraviolet reflectance image further comprise positive, reversed two of bank note to be identified towards image.SpecificallyIn implementation, different images can be acquired as needed and is further analyzed.
For example, in an alternative embodiment, above-mentioned target image includes obtaining institute by contact-type image sensorState the target RGB color image of bank note to be identified, the front of bank note as to be detected towards, reverse side towards feux rouges image, greenLight image, blue light images add up to 6 images.
For another example in another optional embodiment, above-mentioned target image includes being obtained by contact-type image sensorTake the Infrared Targets transmission image of the bank note to be identified；That is, acquiring bank note to be identified by infrared image sensorInfrared transmission image as target image, also, infrared transmission image need not from front towards and reverse side towards respectivelyIt is acquired, it is only necessary to acquire one.
Step S104：The target signature data of the target image are extracted, the target signature data include denomination featureAt least one of data, time characteristic and/or distribution row characteristic.
Step S106：The target signature data are identified, determine target corresponding with the target signature dataBank note classification information.
In the present embodiment, it for the target image of bank note to be identified, needs further to extract its corresponding characteristicAccording in order to carry out further specific feature recognition work.
Since it is desired that the denomination of bank note to be identified, version time and distribution row are identified respectively, therefore, acquisitionCharacteristic is also required to include corresponding characteristic, such as denomination characteristic, is carried out for the denomination to bank note to be identifiedIdentification；For another example time characteristic, is identified for the version time to bank note to be identified；For another example distribution row featureData, for emission of banknotes bank to be identified to be identified.
After determining target signature data to be identified, different target signature data are identified respectively, toCorresponding bank note classification information is determined, for example, the denomination size of bank note to be identified is determined according to denomination characteristic, according to the timeCharacteristic determines the version time of bank note to be identified, and emission of banknotes bank to be identified is determined according to distribution row characteristic.
Specifically, as shown in Fig. 2, determining that the process of the denomination size of bank note to be identified walks as follows according to denomination characteristicRapid S201-S203：
S201：The target RGB color image is converted into target HSV images；
S202：The H component datas of the target HSV images are obtained as the denomination characteristic；
S204：It is searched and the denomination characteristic in the H component informations for each denomination that the feature templates includeMatched H component informations determine denomination corresponding with the matched H component informations as target basis as target signature templateCoin type information；
Wherein, target image includes the target RGB coloured silks that the bank note to be identified is obtained by contact-type image sensorColor image.
That is, target image include bank note to be identified based on R component, based on G components and based on B componentRGB color image.Also, in order to which aspect is analyzed and is identified, in the present embodiment, it is also necessary to convert RGB color imageAt the image under hsv color space, that is to say, that obtain the number of the corresponding H components of each pixel, S components and V componentAccording to.
In the present embodiment, the target HSV images under the hsv color space being converted into for target RGB color image, are obtainedThe corresponding H component datas of each pixel for including in target HSV images are taken, and as denomination characteristic.ExampleSuch as, count the data such as numeric distribution or the mean value of the corresponding H components of each pixel, and by its with count each in advanceThe color threshold range of the bank note of denomination is compared, and determines the color threshold range belonging to it, so that it is determined that target HSV imagesThe matched color threshold range of H component datas (i.e. target signature template), each target signature template corresponded to a faceThe corresponding color threshold of bank note of volume size according to the target signature template conversely, can determine the corresponding denomination of bank note to be identifiedSize.
In another specific embodiment, it can determine that the time version of bank note to be identified is believed by time characteristicBreath.Specifically, above-mentioned target image include obtained by contact-type image sensor the bank note to be identified Infrared Targets it is saturatingPenetrate image；Target signature data by lifting Infrared Targets transmission image determine the time version information of bank note to be identified.
As shown in figure 3, determining that the process of the time version information of bank note to be identified includes as follows according to time characteristicStep S301-S303：
Step S301：The quantity of safety line in the Infrared Targets transmission image is obtained as target signature data；
Step S302：According to the quantity of safety line in the target signature data；
Step S303：Determine time version information corresponding with the bank note to be detected as target bank note type information.
Infrared transmission image can show the features such as the structure of the different zones of bank note to be identified, material, such as, if havingFor the safety line of anti-counterfeit recognition, for example, the marking etc. of ink printing.For example, in application scenarios as shown in Figure 4 and Figure 5,Infrared transmission image can show the quantity of the printing marking of the ink in bank note and safety line, also, application shown in Fig. 4In scene, the quantity of safety line is one, and in application scenarios shown in Fig. 5, the quantity of safety line is two.
In specific implementation, after getting the Infrared Targets transmission image of bank note to be detected, determined by image recognitionThe quantity of the quantity of safety line wherein included, the safety line is target signature data.
For example, in Hongkong dollar bank note, 2 safety lines are contained in the bank note got off the ground in 2003, but 2010 openThe quantity for originating the safety line for including in capable new edition bank note is one, therefore, bank note can be determined according to the quantity of safety lineCorresponding time version, namely determine target bank note type information.
In another optional embodiment, becoming pattern by the 3D light in the upper right corner on the reversed face of bank note can be with areaPoint different banks of issue, for example, in 2010 editions bank note of Standard Chartered Bank's distribution, reversely towards the 3D light in the upper right corner becomeCorresponding image is the corresponding number of denomination, for another example in 2010 editions bank note of Master Card, reversely towards the right sideThe corresponding 3D light variation images at upper angle are large meatball, for another example in 2010 editions bank note of Bank of China (Hong Kong) distribution, it is anti-To towards the upper right corner 3D light variation images it is corresponding be bauhinia.That is, by bank note it is reversed towards imageThe image of predeterminated position in the upper right corner be identified, it may be determined that the corresponding bank of issue.
Specifically, as shown in fig. 6, determining the process packet of emission of banknotes bank to be identified according to distribution row characteristicInclude following steps S401-S403：
Step S401：Determine the topography of the predeterminated position for the bank note reversed image that the target image includes；
Step S402：Obtain binary image corresponding with the topography；
Step S403：What it is by binary image input training in advance includes the capable characteristic image information of various distributionGrader in, export the emission of banknotes row parameter to be detected as target bank note type information.
Specifically, the 3D light that the present invention is preset upper right Angle Position in the reversed image based on bank note to be identified becomes patternInformation judges emission of banknotes bank to be identified, thus in needing to obtain the target image of bank note to be identified first with 3D lightBecome the image information of the topography of the corresponding predeterminated position of pattern.Topography is intercepted into squarely or rectangle, is conducive to essenceTrue further determines the area image.
Further, in the present embodiment, in order to improve accuracy of identification, in specific identification process, office is usedThe binary image of portion's image.That is, using fixed threshold value, binary conversion treatment is carried out to the topography, toTo binary image corresponding with topography, and the identification by the binary image for the subsequent bank of issue.
Specifically, in order to ensure that the corresponding different sorting parameter of the different banks of issue can in advance training processPin-point accuracy corresponds to corresponding grader, is the sorting parameter of the bank note by preserving the trained different banks of issue.According toThe reversed bank note of the different banks of issue has different sorting parameters, can correspond to obtain the classification of the different banks of issue.
Above example can be seen that scheme versatility provided by the invention is good, in the training stage of parameter, does not needMore changes, you can quickly and accurately identify emission of banknotes bank.
Fig. 7 is the specific implementation of the recognition methods for the emission of banknotes bank to be identified that another optional embodiment providesFlow chart.It is shown in Figure 7, above-mentioned steps S403：What it is by binary image input training in advance includes various distributionFurther include step S4030 before step in the grader of capable characteristic image information：It is instructed using convolutional neural networks graderGet include it is various distribution row bank note predeterminated position image information, and by it is described it is various distribution row bank note predeterminated positionsImage information is stored to the grader.
Specifically, the convolutional neural networks grader includes：Multiple feature figure layers, at least one in multiple feature figure layersAt least one of a feature figure layer characteristic pattern is divided into multiple regions；And multiple convolution masks, multiple convolution masks withMultiple regions correspond to respectively, and each convolution mask is used to obtain the response of the neuron in corresponding region.By by various hairsThe binary image of the localized area image information of the bank note of row bank is stored in respectively in multiple convolution masks, to distinguish notThe same bank of issue.
Further, described to obtain various including various bank of issue's bank note using convolutional neural networks classifier trainingLocalized area image information binary image, and the binary image of various bank of issue's bank note is stored to describedGrader can specifically include：
The banknote image to be identified for obtaining the various banks of issue extracts the bank note figure to be identified according to the predeterminated positionAs the binary image of upper topography；The pattern corresponding to the various banks of issue is identified respectively；Each bank of issue selectsIt takes the sample of preset data to repeat above-mentioned two step, obtains sample data；Nerve is rolled into sample data inputIt is trained in network classifier, obtains the binary image of various bank of issue's bank note.
Further, the binary image of above-mentioned topography is the histogram feature under binaryzation pattern.
Step S108：The target category of the bank note to be identified is determined according to the target bank note classification information.
As previously mentioned, after each target bank note classification information of bank note to be identified is determined in step s 106, you canThe classification of the bank note to be identified is determined, to complete the classification to bank note to be identified.
It should be noted that may be in the present embodiment, between different bank note sizes it is different, therefore, in order toStandardization during calculating, in the present embodiment, it is also necessary to which the target image of bank note to be identified is converted into unificationIt is further processed after image size.
That is, before the step of target signature data of target image described in said extracted, further include：To the meshLogo image is normalized, and the target image after the normalized is substituted to the target figure of the bank note to be identifiedThe image size of picture, the target image after the normalized meets preset image size parameter.
In addition, to solve the method (in particular for Hongkong dollar) that traditional technology classifies for paper money recognition, there are classification effectivenessesTechnical problem low, accuracy is low, in one embodiment, it is also proposed that a kind of device of bank note classification.
Specifically, as shown in figure 8, the device of above-mentioned bank note classification includes following module：
Target image acquisition module 101, the target image for obtaining bank note to be identified；
Characteristic extraction module 102, the target signature data for extracting the target image, the target signature numberAccording to including at least one of denomination characteristic, time characteristic and/or distribution row characteristic；
Bank note classification information determination module 103 determines and the target for the target signature data to be identifiedThe corresponding target bank note classification information of characteristic；
Bank note classification identification module 104, for determining the bank note to be identified according to the target bank note classification informationTarget category.
Optionally, in one embodiment, the target image includes by being waited for described in contact-type image sensor acquisitionIdentify the target RGB color image of bank note；The characteristic extraction module 102 is additionally operable to the target RGB color imageTarget HSV images are converted into, obtain the H component datas of the target HSV images as the denomination characteristic；The bank noteClassification information determining module 103 be additionally operable in the H component informations for each denomination that the feature templates include search with it is describedThe matched H component informations of denomination characteristic determine face corresponding with the matched H component informations as target signature templateFor volume as target bank note type information, the feature templates include the H component informations of each denomination.
Optionally, in one embodiment, the target image includes by being waited for described in contact-type image sensor acquisitionIdentify the Infrared Targets transmission image of bank note；Characteristic extraction module 102 is additionally operable to obtain the Infrared Targets transmission imageThe quantity of middle safety line is as target signature data；Bank note classification information determination module 103 is additionally operable to according to the target signatureThe quantity of safety line in data determines time version information corresponding with the bank note to be detected as target bank note type letterBreath.
Optionally, in one embodiment, characteristic extraction module 102 is additionally operable to determine that the target image includesThe topography of the predeterminated position of bank note reversed image obtains binary image corresponding with the topography；Bank note classificationInformation determination module 103 is additionally operable to binary image input training in advance include the capable characteristic image of various distributionIn the grader of information, the emission of banknotes row parameter to be detected is exported as target bank note type information.
Optionally, in one embodiment, it as shown in figure 8, above-mentioned apparatus further includes classifier training module 105, is used forIt obtains including the various predeterminated position image informations for issuing row bank note using convolutional neural networks classifier training, and will be describedThe predeterminated position image information of various distribution row bank note is stored to the grader.
Optionally, in one embodiment, as shown in figure 8, above-mentioned apparatus image normalization processing module 106, for pairThe target image is normalized, and the target image after the normalized is substituted the bank note to be identifiedThe image size of target image, the target image after the normalized meets preset image size parameter.
Implement the embodiment of the present invention, will have the advantages that：
After the method and apparatus for using above-mentioned bank note classification, in the process that the classification of the bank note such as Hongkong dollar is identifiedIn, corresponding characteristic in the image of bank note to be identified can be obtained according to the feature that be based on of bank note classification identification, then rootCorresponding classification information is identified respectively according to each characteristic, so that it is determined that the classification where bank note to be identified.Especially needleFor the Hongkong dollar bank note more to classification, from 3 dimensions such as denomination size, time version, bank of issue respectively to bank note classificationIt is identified, effectively the classification of Hongkong dollar bank note can be identified, and improve the accuracy of paper money recognition.
In the above-described embodiments, all or part of reality can be come by software, hardware, firmware or its arbitrary combinationIt is existing.When being realized using software program, can entirely or partly realize in the form of a computer program product.The computerProgram product includes one or more computer instructions.When loading on computers and executing the computer program instructions, entirelyPortion is partly generated according to the flow or function described in the embodiment of the present invention.The computer can be all-purpose computer, speciallyWith computer, computer network or other programmable devices.The computer instruction can be stored in computer-readable storageIn medium, or from a computer readable storage medium to the transmission of another computer readable storage medium, for example, the meterThe instruction of calculation machine can pass through wired (such as coaxial cable, light from a web-site, computer, server or data centerFine, Digital Subscriber Line (DSL)) or wireless (such as infrared, wireless, microwave etc.) mode to another web-site, computer, clothesBusiness device or data center are transmitted.It is any available can be that computer can access for the computer readable storage mediumMedium is either comprising data storage devices such as one or more usable mediums integrated server, data centers.It is described to useMedium can be magnetic medium, (for example, floppy disk, hard disk, tape), optical medium (for example, DVD) or semiconductor medium (such as it is solidState hard disk Solid State Disk (SSD)) etc..
In one embodiment, as shown in figure 9, Fig. 9 illustrate a kind of method of operation above-mentioned bank note classification based on FengThe terminal of the computer system of Nuo Yiman systems.The computer system can be smart mobile phone, tablet computer, palm PC, penRemember the terminal devices such as this computer or PC.Specifically, may include the outer input interface 1001 connected by system bus,Processor 1002, memory 1003 and output interface 1004.Wherein, outer input interface 1001 can optionally include at least networkInterface 10012.Memory 1003 may include external memory 10032 (such as hard disk, CD or floppy disk etc.) and built-in storage10034.Output interface 1004 can include at least the equipment such as display screen 10042.
In the present embodiment, the operation of this method is based on computer program, and the program file of the computer program is stored inIn the external memory 10032 of the aforementioned computer system based on von Neumann system, it is loaded into built-in storage at runtimeIt in 10034, is then compiled as being transferred in processor 1002 after machine code executing, so that being based on von Neumann systemComputer system in form target image acquisition module 101 in logic, characteristic extraction module 102, bank note classification letterCease determining module 103, bank note classification identification module 104, classifier training module 105, image normalization processing module 106.AndIn the method implementation procedure of above-mentioned bank note classification, the parameter of input is received by outer input interface 1001, and is transferred toIt is cached in memory 1003, is then input in processor 1002 and is handled, the result data of processing or be cached in memoryIt is subsequently handled in 1003, or is passed to output interface 1004 and is exported.
The above disclosure is only the preferred embodiments of the present invention, cannot limit the right model of the present invention with this certainlyIt encloses, therefore equivalent changes made in accordance with the claims of the present invention, is still within the scope of the present invention.