WO2017092209A1 - 有价文件自适应识别方法和装置 - Google Patents

有价文件自适应识别方法和装置 Download PDF

Info

Publication number
WO2017092209A1
WO2017092209A1 PCT/CN2016/078506 CN2016078506W WO2017092209A1 WO 2017092209 A1 WO2017092209 A1 WO 2017092209A1 CN 2016078506 W CN2016078506 W CN 2016078506W WO 2017092209 A1 WO2017092209 A1 WO 2017092209A1
Authority
WO
WIPO (PCT)
Prior art keywords
correction amount
photoelectric signal
value
value document
component
Prior art date
Application number
PCT/CN2016/078506
Other languages
English (en)
French (fr)
Inventor
刘梦涛
王荣秋
王卫锋
Original Assignee
广州广电运通金融电子股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 广州广电运通金融电子股份有限公司 filed Critical 广州广电运通金融电子股份有限公司
Priority to EP16869523.7A priority Critical patent/EP3385923A4/en
Priority to RU2018118168A priority patent/RU2690716C1/ru
Priority to US15/774,560 priority patent/US10529163B2/en
Publication of WO2017092209A1 publication Critical patent/WO2017092209A1/zh
Priority to ZA2018/02811A priority patent/ZA201802811B/en

Links

Images

Classifications

    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/06Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency using wave or particle radiation
    • G07D7/12Visible light, infrared or ultraviolet radiation
    • G07D7/128Viewing devices
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/06Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency using wave or particle radiation
    • G07D7/12Visible light, infrared or ultraviolet radiation
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/20Testing patterns thereon
    • G07D7/2008Testing patterns thereon using pre-processing, e.g. de-blurring, averaging, normalisation or rotation
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/20Testing patterns thereon
    • G07D7/2016Testing patterns thereon using feature extraction, e.g. segmentation, edge detection or Hough-transformation
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07DHANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00Testing specially adapted to determine the identity or genuineness of valuable papers or for segregating those which are unacceptable, e.g. banknotes that are alien to a currency
    • G07D7/20Testing patterns thereon
    • G07D7/2075Setting acceptance levels or parameters
    • G07D7/2083Learning

Definitions

  • the present invention relates to the field of finance, and in particular to a method and apparatus for adaptive identification of value documents.
  • the banking system is a bit of a currency machine, a sorting machine, an ATM, etc.
  • the retail industry has vending machines
  • the intelligent transportation industry has automatic ticket vending machines.
  • a common feature of these devices is the need to rely on the identification device to complete the detection and identification of the ticket, and the photosensitive sensor and recognition algorithm are an important part of all identification devices.
  • the common practice of each product in the photosensitive sensor is to correct the sensor output signal by the principle of white reference film and photoelectric signal feedback compensation;
  • the common method in the recognition algorithm is to use the large sample training method according to the physical object to be processed. Find the appropriate threshold and then solidify it into the algorithm in the form of parameters to meet specific product requirements.
  • the CIS In the process of acquiring the target image by using CIS, due to factors such as optical unevenness, response difference of the photosensitive unit itself, dark current and offset, an image with uneven intensity may be output for a target with uniform gray scale. This will be detrimental to target recognition and measurement in subsequent image processing. Therefore, before and after the target image is acquired by CIS, the CIS needs to be calibrated in black and white. Currently, in the well-known CIS non-uniformity correction algorithm, the two-point method is used to correct the CIS.
  • the feedback system is mainly composed of proportional link, integral link and differential link.
  • the traditional white-based photoelectric signal feedback correction method only uses the proportional link, and the deviation of the feedback signal is multiplied by the scaling factor. This method has certain effect on the real-time deviation correction of the sensor itself, but the sensor is added and recognized.
  • the cumulative error of the entire system formed by the algorithm cannot be handled, that is, the integral feedback link is lacking.
  • the sensor cannot actively predict the change of the processing object. Therefore, after the processing object changes, the traditional method cannot be perceived, and the early correction cannot be made, that is, the differential feedback link is lacking.
  • the embodiment of the invention provides a method and a device for adaptively identifying a value document, which can solve the problem of system cumulative error and early correction.
  • the recognition result is output.
  • correction formula for digital compensation is as follows:
  • p is an arbitrary point of the photoelectric signal gradation value
  • M 0 is the photoelectric signal correction amount
  • calculating a cumulative component and a differential error of the value document according to the feature information The difference specifically includes:
  • Calculating the cumulative component of the value document m i is the characteristic component values of M n at time i;
  • a differential error M w M n -M 1 of the value document is calculated.
  • calculating the total correction amount of the photoelectric signal according to the cumulative component and the differential error specifically includes:
  • Calculating the third correction amount M 3 of the photoelectric signal according to the differential error specifically comprising: if
  • ⁇ w, the third correction amount M 3 ⁇ M w ; if
  • updating the photoelectric signal correction amount and the acquisition parameter according to the total correction amount includes:
  • Updating the value of the photoelectric signal correction amount M 0 is equal to the total correction amount M;
  • the method before the digital compensation of the photoelectric signal according to the photoelectric signal correction amount, the method further includes:
  • x is the photoelectric signal value before any point correction
  • y is the corresponding corrected photoelectric signal value
  • a is the first correction coefficient
  • b is the second correction coefficient.
  • the preset initialization value of the acquisition parameter is acquired, and an initialization value of the photoelectric signal correction amount is obtained, and the initialization value of the photoelectric signal correction amount is 0.
  • the photoelectric signal acquisition module is configured to acquire an acquisition parameter, and collect the photoelectric signal of the value document according to the collection parameter;
  • a digital compensation module configured to acquire a photoelectric signal correction amount, and perform digital compensation on the photoelectric signal according to the photoelectric signal correction amount
  • a feature extraction module configured to perform feature extraction on the digitally compensated photoelectric signal to obtain a feature vector
  • An identification module configured to send the feature vector into a preset classifier to obtain a recognition result of the value document
  • a specific area obtaining module configured to acquire a specific area corresponding to the value document according to the identification result
  • a feature information acquiring module configured to acquire feature information of the photoelectric signal of the value document according to the specific area
  • a cumulative component and differential error calculation module configured to calculate a cumulative component and a differential error of the value document according to the feature information
  • a total correction amount calculation module configured to calculate a total correction amount of the photoelectric signal according to the cumulative component and the differential error
  • an update module configured to update the photoelectric signal correction amount and the acquisition parameter according to the total correction amount
  • a recognition result output module is configured to output the recognition result.
  • the cumulative component and differential error calculation module specifically includes:
  • a cumulative component calculation unit for calculating a cumulative component of the value document m i is the characteristic component values of M n at time i;
  • the total correction amount calculation module specifically includes:
  • a third correction amount calculation unit configured to calculate a third correction amount M 3 of the photoelectric signal according to the differential error, specifically comprising: if
  • ⁇ w, the third correction amount M 3 ⁇ M w If
  • the update module specifically includes:
  • the photoelectric signal correction amount updating unit is configured to update the photoelectric signal correction amount M 0 to be equal to the total correction amount M;
  • the device further includes:
  • the acquisition parameter initialization value acquisition module is configured to acquire a preset initialization value of the acquisition parameter when the photoelectric signal of the value document is collected for the first time;
  • the correction amount initialization value acquisition module is configured to acquire an initialization value of the photoelectric signal correction amount when the photoelectric signal of the value document is first collected, and the initialization value of the photoelectric signal correction amount is 0.
  • the present invention first, acquiring an acquisition parameter, collecting a photoelectric signal of the value document according to the collection parameter, acquiring a photoelectric signal correction amount, and performing digital compensation on the photoelectric signal according to the photoelectric signal correction amount;
  • the digitally compensated photoelectric signal is subjected to feature extraction to obtain a feature vector;
  • the feature vector is sent to a preset classifier for identification to obtain a recognition result of the value document; and then, according to the recognition result, Corresponding specific area on the price file; acquiring characteristic information of the photoelectric signal of the value document according to the specific area; calculating a cumulative component and a differential error of the value document according to the characteristic information; according to the cumulative component And differential
  • the error is calculated to obtain the total correction amount of the photoelectric signal; finally, the photoelectric signal correction amount and the acquisition parameter are updated according to the total correction amount; and the recognition result is output. Therefore, the cumulative feedback adaptive and differential feedback control adaptation of the value document identification process can be realized, and the cumulative error problem and the differential error problem of the system can be
  • FIG. 1 is a flow chart of an embodiment of a method for adaptively identifying a value document according to an embodiment of the present invention
  • FIG. 2 is a flow chart of another embodiment of a method for adaptively identifying a value document according to an embodiment of the present invention
  • 3 is a stable rectangular area selected by the white light transmission crown image of the present invention.
  • Figure 5 is a schematic diagram of cumulative components of the present invention.
  • Figure 6 is a schematic diagram of differential error of the present invention.
  • FIG. 7 is a structural diagram of an embodiment of a value document adaptive identification apparatus according to an embodiment of the present invention.
  • FIG. 8 is a structural diagram of another embodiment of a value document adaptive identification apparatus according to an embodiment of the present invention.
  • Embodiments of the present invention provide a method and apparatus for adaptive identification of value documents for solving problems of system cumulative error and early correction.
  • an embodiment of an adaptive identification method for a value document includes:
  • the acquisition parameter Before collecting the photoelectric signal of the value document, the acquisition parameter may be acquired, and the photoelectric signal of the value document is collected according to the acquisition parameter.
  • the photoelectric signal correction amount may be acquired, and the photoelectric signal is digitally compensated according to the photoelectric signal correction amount.
  • the digitally compensated photoelectric signal can be subjected to feature extraction to obtain a feature vector.
  • the feature vector is sent to a preset classifier for identification, and the recognition result of the value document is obtained;
  • the feature vector may be sent to a preset classifier for identification to obtain a recognition result of the value document.
  • the specific area corresponding to the value document can be obtained according to the recognition result.
  • the feature information of the photoelectric signal of the value document may be acquired according to the specific area.
  • the cumulative component and the differential error of the value document may be calculated according to the feature information.
  • the total correction amount of the photoelectric signal can be calculated according to the cumulative component and the differential error.
  • the photoelectric signal correction amount and the acquisition parameter may be updated according to the total correction amount.
  • the recognition result may be output.
  • the photoelectric signal correction amount and the acquisition parameter are updated according to the total correction amount; and the recognition result is output. Therefore, the cumulative feedback adaptive and differential feedback control adaptation of the value document identification process can be realized, and the cumulative error problem and the differential error problem of the system can be solved.
  • FIG. 2 another embodiment of the method for adaptively identifying a value document in the embodiment of the present invention includes:
  • the acquisition parameters can be acquired, and the photoelectric signals of the value documents are collected according to the acquisition parameters.
  • the photoelectric signal of the value document is collected for the first time, the preset initialization value of the acquisition parameter is acquired, and the initialization value of the photoelectric signal correction amount is obtained, and the initialization value of the photoelectric signal correction amount is 0.
  • the photoelectric signal can be signal compensated before the photoelectric signal is digitally compensated. It is necessary to acquire a preset first correction coefficient and a second correction coefficient.
  • the photoelectric signal may be compensated according to the first correction coefficient and the second correction coefficient.
  • This embodiment is exemplified by a two-point method, and the specific compensation correction formula is as follows :
  • x is the photoelectric signal value before any point correction
  • y is the corresponding corrected photoelectric signal value
  • a is The first correction coefficient
  • b is the second correction coefficient.
  • the photoelectric signal correction amount After performing signal compensation on the photoelectric signal according to the first correction coefficient and the second correction coefficient, the photoelectric signal correction amount may be acquired, and the photoelectric signal is digitally compensated according to the photoelectric signal correction amount, and the specific correction formula is as follows:
  • p is the gray value of any point of the photoelectric signal
  • p' is the correction value of p
  • M 0 is the photoelectric signal correction amount. Obviously, the corrected value is exactly the same as expected.
  • M 0 >0 the gray value is increased, and when M 0 ⁇ 0, the gray value is lowered.
  • the digitally compensated photoelectric signal may be subjected to feature extraction to obtain a feature vector.
  • the feature vector is sent to a preset classifier for identification, and the recognition result of the value document is obtained.
  • the feature vector can be sent to a preset classifier for identification, and the recognition result of the value document can be obtained.
  • the classifier can be, but is not limited to, a neural network or a support vector machine.
  • the specific area corresponding to the value document can be obtained according to the recognition result.
  • the feature information of the photoelectric signal of the value document may be acquired according to the specific area.
  • the cumulative component of the value document can be calculated m i is the characteristic component values of M n at time i.
  • the differential error of the value document can be calculated.
  • a is a photoelectric characteristic curve of a value document
  • b is a standard curve
  • M 1 is a cumulative component
  • a is a cumulative component curve
  • b is a standard curve
  • a cumulative error is calculated.
  • the signal correction amount M 2 k 2 * (M * - M l )
  • M * is a preset standard information
  • k 2 is The empirical value
  • c is the cumulative error curve
  • the differential error M w M n - M l of the value document photoelectric signal.
  • the third correction amount M 3 of the photoelectric signal may be calculated according to the differential error, specifically including: if
  • ⁇ w, the third correction amount M 3 ⁇ M w ; if
  • update the photoelectric signal correction amount M 0 value is equal to the total correction amount M;
  • the value of the photoelectric signal correction amount M 0 may be updated to be equal to the total correction amount M.
  • indicates that in the normal light-emitting range of the CIS sensor, it is necessary to increase the photoelectric intensity by increasing one gray value on average.
  • the recognition result may be output.
  • the linear feedback analysis module uses the proportional link to correct the feedback photoelectric signal. There is a large error, and the crown number area may be overexposed or too dark to affect the recognition effect. .
  • the value document adaptive identification method it is possible to solve the problem that the recognition device aging causes the crown number recognition effect to be poor.
  • the value identification device When the value identification device is aging or when a new banknote is entered. If the white light transmission image is too dark or too bright, based on the cumulative feedback adaptive method, that is, the system enhances or weakens the photoelectric energy, and when the banknote is re-entered, the problem that the white light transmission image is too dark or overexposed can be avoided.
  • the photoelectric signal correction amount and the acquisition parameter are updated according to the total correction amount; and the recognition result is output. Therefore, the cumulative feedback adaptive and differential feedback control adaptation of the value document identification process can be realized, and the cumulative error problem and the differential error problem of the system can be solved.
  • the photoelectric signal collecting module 701 is configured to acquire an acquisition parameter, and collect the valuable according to the collection parameter.
  • the photoelectric signal of the document is configured to acquire an acquisition parameter, and collect the valuable according to the collection parameter.
  • the digital compensation module 702 is configured to acquire a photoelectric signal correction amount, and perform digital compensation on the photoelectric signal according to the photoelectric signal correction amount;
  • the feature extraction module 703 is configured to perform feature extraction on the digitally compensated photoelectric signal to obtain a feature vector
  • the identification module 704 is configured to send the feature vector into a preset classifier to obtain a recognition result of the value document;
  • the specific area obtaining module 705 is configured to obtain a specific area corresponding to the value document according to the recognition result
  • the feature information obtaining module 706 is configured to acquire feature information of the photoelectric signal of the value document according to the specific area;
  • the cumulative component and differential error calculation module 707 is configured to calculate a cumulative component and a differential error of the value document according to the feature information
  • the total correction amount calculation module 708 is configured to calculate a total correction amount of the photoelectric signal according to the cumulative component and the differential error;
  • the update module 709 is configured to update the photoelectric signal correction amount and the acquisition parameter according to the total correction amount
  • the recognition result output module 710 is configured to output the recognition result.
  • the photoelectric signal acquisition module 701 acquires the acquisition parameter, and collects the photoelectric signal of the value document according to the acquisition parameter; the digital compensation module 702 acquires the photoelectric signal correction amount, and performs digitalization on the photoelectric signal according to the photoelectric signal correction amount.
  • the feature extraction module 703 performs feature extraction on the digitally compensated optical signal to obtain a feature vector; the identification module 704 sends the feature vector to a preset classifier for identification, to obtain a recognition result of the value document; Then, the specific area obtaining module 705 acquires a specific area corresponding to the value document according to the recognition result; the feature information acquiring module 706 acquires the feature information of the photoelectric signal of the value document according to the specific area; and then, the accumulated component and the differential error
  • the calculation module 707 calculates the cumulative component and the differential error of the value document according to the feature information; the total correction amount calculation module 708 calculates the total correction amount of the photoelectric signal according to the cumulative component and the differential error; the update module 709 is based on the total correction The amount of the photoelectric signal correction amount and the acquisition parameter are updated; After the recognition result output module 710 outputs the recognition result. Therefore, the cumulative feedback adaptive and differential feedback control adaptation of the value document identification process can be realized, and the cumulative error problem and the differential
  • FIG. 8 another embodiment of the value document adaptive identification device in the embodiment of the present invention includes:
  • the photoelectric signal collecting module 801 is configured to acquire an acquisition parameter, and collect the photoelectric signal of the value document according to the collection parameter;
  • the digital compensation module 802 is configured to acquire a photoelectric signal correction amount, and perform digital compensation on the photoelectric signal according to the photoelectric signal correction amount;
  • the feature extraction module 803 is configured to perform feature extraction on the digitally compensated photoelectric signal to obtain a feature vector
  • the identification module 804 is configured to send the feature vector into a preset classifier to obtain a recognition result of the value document;
  • the specific area obtaining module 805 is configured to obtain a specific area corresponding to the value document according to the recognition result
  • the feature information obtaining module 806 is configured to acquire feature information of the photoelectric signal of the value document according to the specific region;
  • the cumulative component and differential error calculation module 807 is configured to calculate a cumulative component and a differential error of the value document according to the feature information
  • the total correction amount calculation module 808 is configured to calculate a total correction amount of the photoelectric signal according to the cumulative component and the differential error;
  • the update module 809 is configured to update the photoelectric signal correction amount and the acquisition parameter according to the total correction amount
  • the recognition result output module 810 is configured to output the recognition result.
  • the cumulative component and differential error calculation module 807 in this embodiment may specifically include:
  • a cumulative component calculation unit 8072 for calculating a cumulative component of the value document m i is the characteristic component values of M n at time i;
  • the total correction amount calculation module 808 in this embodiment may specifically include:
  • the third correction amount calculation unit 8082 is configured to calculate the third correction amount M 3 of the photoelectric signal according to the differential error, and specifically includes: if
  • ⁇ w, the third correction amount M 3 ⁇ M w ; If
  • the update module 809 in this embodiment may specifically include:
  • the photoelectric signal correction amount updating unit 8091 is configured to update the photoelectric signal correction amount M 0 to be equal to the total correction amount M;
  • the acquisition parameter initialization value obtaining module 811 is configured to acquire a preset initialization value of the acquisition parameter when the photoelectric signal of the value document is collected for the first time;
  • the correction amount initialization value acquisition module 812 is configured to acquire an initialization value of the photoelectric signal correction amount when the photoelectric signal of the value document is first collected, and the initialization value of the photoelectric signal correction amount is 0.
  • the disclosed system, apparatus, and method may be implemented in other manners.
  • the device embodiments described above are merely illustrative.
  • the division of the unit is only a logical function division, and may be implemented in actual implementation.
  • There are additional ways of dividing for example, multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, and may be in an electrical, mechanical or other form.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.
  • each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
  • the above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.
  • the integrated unit if implemented in the form of a software functional unit and sold or used as a standalone product, can be stored in a computer readable storage medium.
  • the technical solution of the present invention which is essential or contributes to the prior art, or all or part of the technical solution, may be embodied in the form of a software product stored in a storage medium.
  • a number of instructions are included to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method of various embodiments of the present invention.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Toxicology (AREA)
  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Input (AREA)
  • Facsimile Scanning Arrangements (AREA)
  • Character Discrimination (AREA)

Abstract

一种有价文件自适应识别方法和装置,用于解决系统累积误差和提前修正的问题。识别方法包括:获取采集参数,根据采集参数采集有价文件的光电信号(101);获取光电信号修正量,根据光电信号修正量对光电信号进行数字补偿(102);对数字补偿后的光电信号进行特征提取,得到特征向量(103);将特征向量送入预设的分类器中识别,得到有价文件的识别结果(104);根据识别结果获取有价文件上对应的特定区域(105);根据特定区域获取有价文件的光电信号的特征信息(106);根据特征信息计算得到有价文件的累积分量和微分误差(107);根据累积分量和微分误差计算得到光电信号的总修正量(108);根据总修正量更新光电信号修正量和采集参数(109);输出识别结果(110)。

Description

有价文件自适应识别方法和装置
本申请要求于2015年12月02日提交中国专利局、申请号为201510874880.X、发明名称为“有价文件自适应识别方法和装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本发明涉及金融领域,尤其涉及有价文件自适应识别方法和装置。
背景技术
全球范围内现金流通使用到大量的票据识别和处理设备,银行系统有点钞机、清分机、ATM等,零售行业有自动售卖机,智能交通行业有自动售票机等。这些设备的共同特点是需要依靠识别装置完成对票据的检测和识别,而光敏传感器和识别算法则是所有识别装置的重要组成部分。
各种识别装置因为应用行业不同,需要适应不同的需求和使用环境,这就需要光敏传感器和识别算法具有一定的自动适应能力。比如:传感器需要适应不同温度、湿度的变化,保证信号输出的稳定性和一致性;识别算法需要适应不同新旧程度、不同面值、不同版本的票据,保证识别效果的稳定性和一致性。
目前在光敏传感器方面各产品的通用做法是利用白基准片和光电信号反馈补偿的原理对传感器输出信号进行修正;在识别算法方面的通用做法是根据要处理票据的实物,用大样本训练的方式找到合适的阈值,然后用参数的形式固化到算法里以满足具体的产品要求。
在利用CIS采集目标图像过程中,由于受到诸如光学不均匀性、光敏单元本身响应差异、暗电流及偏置等因素的影响,对一个灰度均匀的目标,可能会输出强度不均匀的图像,这将对后续图像处理中的目标识别及测量不利,因此,在利用CIS采集目标图像之前,需要对CIS进行黑白校准,目前,公知的CIS非均匀校正算法中,两点法是用来校正CIS非均匀性的一种有效方法,它成立的假设条件是每个光敏单元的响应是线性的,只需对直线上的两点进行定标测量,便可求出光敏单元的响应直线,从而对非均匀性进行校正。然而,因年月变迁造成的发光体及受光原件的变化带来的有价文件光电信号精度变化亦会影响相关设备的识别精度。
基于过程控制中的反馈控制原理,反馈系统主要由比例环节、积分环节、微分环节构成。传统的基于白基准的光电信号反馈修正方法只利用了其中的比例环节,对反馈信号的偏差乘上比例因子进行修正,这种方法对传感器本身的实时偏差修正有一定效果,但对于传感器加识别算法构成的整个系统的累积误差却无法处理,也就是缺乏积分反馈环节。另外传感器作为被动器件,无法主动预知处理对象的变化,所以处理对象发生变化后,传统方法无法感知,也无法做出提前修正,也就是缺乏了微分反馈环节。
因此,需要改进整个反馈控制系统的设计,引入积分和微分反馈控制环节,解决系统累积误差和提前修正的问题。
发明内容
本发明实施例提供了有价文件自适应识别方法和装置,能够解决系统累积误差和提前修正的问题。
本发明实施例提供的一种有价文件自适应识别方法,包括:
获取采集参数,根据所述采集参数采集有价文件的光电信号;
获取光电信号修正量,根据所述光电信号修正量对所述光电信号进行数字补偿;
对数字补偿后的所述光电信号进行特征提取,得到特征向量;
将所述特征向量送入预设的分类器中识别,得到所述有价文件的识别结果;
根据所述识别结果获取有价文件上对应的特定区域;
根据所述特定区域获取所述有价文件的光电信号的特征信息;
根据所述特征信息计算得到所述有价文件的累积分量和微分误差;
根据所述累积分量和微分误差计算得到所述光电信号的总修正量;
根据所述总修正量更新所述光电信号修正量和所述采集参数;
输出所述识别结果。
可选地,数字补偿的修正公式如下:
p'=p+M0
p是所述光电信号任意一点的灰度值,p'为p的修正值,M0为所述光电信号修正量。
可选地,根据所述特征信息计算得到所述有价文件的累积分量和微分误 差具体包括:
根据所述特征信息计算所述光电信号的特征分量,所述特征分量Mn的表达式为:
Figure PCTCN2016078506-appb-000001
θi为特征信息,i=1,2,…,t;
计算所述有价文件的累积分量
Figure PCTCN2016078506-appb-000002
mi为特征分量Mn在i时刻的值;
计算所述有价文件的微分误差Mw=Mn-M1
可选地,根据所述累积分量和微分误差计算得到所述光电信号的总修正量具体包括:
根据所述累积分量计算所述光电信号的第二修正量M2=k2*(M*-Ml),M*为预设的标准信息,k2为预设的第二系数;
根据所述微分误差计算得到所述光电信号的第三修正量M3,具体包括:若|Mw|<w,则第三修正量M3=-Mw;若|Mw|≥w,当满足|Mw|≥w的所述光电信号样本数为n,且
Figure PCTCN2016078506-appb-000003
时,则第三修正量M3=0,否则,第三修正量M3=-k3*Mw,N为所述光电信号的样本总数,k3为预设的第三系数;
根据所述累积分量、所述第二修正量和所述第三修正量得到总修正量,总修正量M=M1+M2+M3
可选地,根据所述总修正量更新所述光电信号修正量和所述采集参数包括:
更新所述光电信号修正量M0的值等于所述总修正量M;
初始化所述采集参数,更新所述采集参数E0=E0+λ·M0,E0的初始化值为预设值,λ为预设的修正系数。
可选地,根据所述光电信号修正量对所述光电信号进行数字补偿之前还包括:
获取预设的第一校正系数和第二校正系数;
根据所述第一校正系数和第二校正系数对所述光电信号进行信号补偿,补偿修正公式如下:
y=a·x+b;
x为任一点修正前的光电信号值,y为对应的修正后的光电信号值,a为第一校正系数,b为第二校正系数。
可选地,当第一次采集有价文件的光电信号时,获取采集参数的预设初始化值,获取光电信号修正量的初始化值,所述光电信号修正量的初始化值为0。
本发明实施例提供的一种有价文件自适应识别装置,包括:
光电信号采集模块,用于获取采集参数,根据所述采集参数采集有价文件的光电信号;
数字补偿模块,用于获取光电信号修正量,根据所述光电信号修正量对所述光电信号进行数字补偿;
特征提取模块,用于对数字补偿后的所述光电信号进行特征提取,得到特征向量;
识别模块,用于将所述特征向量送入预设的分类器中识别,得到所述有价文件的识别结果;
特定区域获取模块,用于根据所述识别结果获取有价文件上对应的特定区域;
特征信息获取模块,用于根据所述特定区域获取所述有价文件的光电信号的特征信息;
累积分量和微分误差计算模块,用于根据所述特征信息计算得到所述有价文件的累积分量和微分误差;
总修正量计算模块,用于根据所述累积分量和微分误差计算得到所述光电信号的总修正量;
更新模块,用于根据所述总修正量更新所述光电信号修正量和所述采集参数;
识别结果输出模块,用于输出所述识别结果。
可选地,
所述累积分量和微分误差计算模块具体包括:
特征分量计算单元,用于根据所述特征信息计算所述光电信号的特征分量,所述特征分量Mn的表达式为:
Figure PCTCN2016078506-appb-000004
θi为特征信息,i=1,2,…,t;
累积分量计算单元,用于计算所述有价文件的累积分量
Figure PCTCN2016078506-appb-000005
mi为特征分量Mn在i时刻的值;
微分误差计算单元,用于计算所述有价文件的微分误差Mw=Mn-M1
所述总修正量计算模块具体包括:
第二修正量计算单元,用于根据所述累积分量计算所述光电信号的第二修正量M2=k2*(M*-Ml),M*为预设的标准信息,k2为预设的第二系数;
第三修正量计算单元,用于根据所述微分误差计算得到所述光电信号的第三修正量M3,具体包括:若|Mw|<w,则第三修正量M3=-Mw;若|Mw|≥w,当满足|Mw|≥w的所述光电信号样本数为n,且
Figure PCTCN2016078506-appb-000006
时,则第三修正量M3=0,否则,第三修正量M3=-k3*Mw,N为所述光电信号的样本总数,k3为预设的第三系数;
总修正量计算单元,用于根据所述累积分量、所述第二修正量和所述第三修正量得到总修正量,总修正量M=M1+M2+M3
所述更新模块具体包括:
光电信号修正量更新单元,用于更新所述光电信号修正量M0的值等于所述总修正量M;
采集参数更新单元,用于初始化所述采集参数,更新所述采集参数E0=E0+λ·M0,E0的初始化值为预设值,λ为预设的修正系数。
可选地,所述装置还包括:
采集参数初始化值获取模块,用于当第一次采集有价文件的光电信号时,获取采集参数的预设初始化值;
修正量初始化值获取模块,用于当第一次采集有价文件的光电信号时,获取光电信号修正量的初始化值,所述光电信号修正量的初始化值为0。
从以上技术方案可以看出,本发明实施例具有以下优点:
本发明实施例中,首先,获取采集参数,根据所述采集参数采集有价文件的光电信号;获取光电信号修正量,根据所述光电信号修正量对所述光电信号进行数字补偿;然后,对数字补偿后的所述光电信号进行特征提取,得到特征向量;将所述特征向量送入预设的分类器中识别,得到所述有价文件的识别结果;接着,根据所述识别结果获取有价文件上对应的特定区域;根据所述特定区域获取所述有价文件的光电信号的特征信息;根据所述特征信息计算得到所述有价文件的累积分量和微分误差;根据所述累积分量和微分 误差计算得到所述光电信号的总修正量;最后,根据所述总修正量更新所述光电信号修正量和所述采集参数;输出所述识别结果。从而,可以实现有价文件识别过程的累积反馈自适应及微分反馈控制自适应,解决系统的累积误差问题和微分误差问题。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其他的附图。
图1为本发明实施例中有价文件自适应识别方法一个实施例流程图;
图2为本发明实施例中有价文件自适应识别方法另一个实施例流程图;
图3是本发明白光透射冠字号图像选取稳定矩形区域;
图4是本发明有价文件特征信息输出示意图;
图5是本发明累积分量示意图;
图6是本发明微分误差示意图;
图7为本发明实施例中有价文件自适应识别装置一个实施例结构图;
图8为本发明实施例中有价文件自适应识别装置另一个实施例结构图。
具体实施方式
本发明实施例提供了有价文件自适应识别方法和装置,用于解决系统累积误差和提前修正的问题。
为使得本发明的发明目的、特征、优点能够更加的明显和易懂,下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述,显然,下面所描述的实施例仅仅是本发明一部分实施例,而非全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例,都属于本发明保护的范围。
请参阅图1,本发明实施例中一种有价文件自适应识别方法一个实施例包括:
101、获取采集参数,根据该采集参数采集有价文件的光电信号;
在采集有价文件的光电信号之前,可以获取采集参数,根据该采集参数采集有价文件的光电信号。
102、获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿;
在根据该采集参数采集有价文件的光电信号之后,可以获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿。
103、对数字补偿后的该光电信号进行特征提取,得到特征向量;
在根据该光电信号修正量对该光电信号进行数字补偿之后,可以对数字补偿后的该光电信号进行特征提取,得到特征向量。
104、将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果;
在得到特征向量之后,可以将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果。
105、根据该识别结果获取有价文件上对应的特定区域;
在得到该有价文件的识别结果之后,可以根据该识别结果获取有价文件上对应的特定区域。
106、根据该特定区域获取该有价文件的光电信号的特征信息;
在根据该识别结果获取有价文件上对应的特定区域之后,可以根据该特定区域获取该有价文件的光电信号的特征信息。
107、根据该特征信息计算得到该有价文件的累积分量和微分误差;
在根据该特定区域获取该有价文件的光电信号的特征信息之后,可以根据该特征信息计算得到该有价文件的累积分量和微分误差。
108、根据该累积分量和微分误差计算得到该光电信号的总修正量;
在根据该特征信息计算得到该有价文件的累积分量和微分误差之后,可以根据该累积分量和微分误差计算得到该光电信号的总修正量。
109、根据该总修正量更新该光电信号修正量和该采集参数;
在根据该累积分量和微分误差计算得到该光电信号的总修正量之后,可以根据该总修正量更新该光电信号修正量和该采集参数。
110、输出该识别结果。
在根据该总修正量更新该光电信号修正量和该采集参数之后,可以输出该识别结果。
本实施例中,首先,获取采集参数,根据该采集参数采集有价文件的光电信号;获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿;然后,对数字补偿后的该光电信号进行特征提取,得到特征向量;将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果;接着,根据该识别结果获取有价文件上对应的特定区域;根据该特定区域获取该有价文件的光电信号的特征信息;根据该特征信息计算得到该有价文件的累积分量和微分误差;根据该累积分量和微分误差计算得到该光电信号的总修正量;最后,根据该总修正量更新该光电信号修正量和该采集参数;输出该识别结果。从而,可以实现有价文件识别过程的累积反馈自适应及微分反馈控制自适应,解决系统的累积误差问题和微分误差问题。
为便于理解,下面对本发明实施例中的一种有价文件自适应识别方法进行详细描述,请参阅图2,本发明实施例中一种有价文件自适应识别方法另一个实施例包括:
201、获取采集参数,根据该采集参数采集有价文件的光电信号;
首先,可以获取采集参数,根据该采集参数采集有价文件的光电信号。
需要说明的是,当第一次采集有价文件的光电信号时,获取采集参数的预设初始化值,获取光电信号修正量的初始化值,该光电信号修正量的初始化值为0。
202、获取预设的第一校正系数和第二校正系数;
在对该光电信号进行数字补偿之前,可以对该光电信号进行信号补偿。需要获取预设的第一校正系数和第二校正系数。需要说明的是,该第一校正系数和第二校正系数可以预先计算得到,具体获取方法可以为:通过白样张和黑样张求出光敏单元的响应直线并代入修正公式y=a·x+b(两点法信号补偿),求出校正系数(第一校正系数)和暗电流的校正量(第二校正系数)。
203、根据该第一校正系数和第二校正系数对该光电信号进行信号补偿;
在获取预设的第一校正系数和第二校正系数之后,可以根据该第一校正系数和第二校正系数对该光电信号进行信号补偿,本实施例以两点法举例,具体补偿修正公式如下:
y=a·x+b;
x为任一点修正前的光电信号值,y为对应的修正后的光电信号值,a为 第一校正系数,b为第二校正系数。
204、获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿;
在根据该第一校正系数和第二校正系数对该光电信号进行信号补偿之后,可以获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿,具体修正公式如下:
p'=p+M0
其中,p是该光电信号任意一点的灰度值,p'为p的修正值,M0为该光电信号修正量。显然修正后的值与期望完全相同,其中M0>0时,调高灰度值,M0<0时,调低灰度值。
205、对数字补偿后的该光电信号进行特征提取,得到特征向量;
在获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿之后,可以对数字补偿后的该光电信号进行特征提取,得到特征向量。该特征向量可以为β=(ε12,…,εt)。
206、将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果;
在得到特征向量之后,可以将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果,该分类器可以是但不仅限于神经网络或支持向量机。
207、根据该识别结果获取有价文件上对应的特定区域;
在得到该有价文件的识别结果之后,可以根据该识别结果获取有价文件上对应的特定区域。
208、根据该特定区域获取该有价文件的光电信号的特征信息;
在根据该识别结果获取有价文件上对应的特定区域之后,可以根据该特定区域获取该有价文件的光电信号的特征信息。
209、根据该特征信息计算该光电信号的特征分量;
在根据该特定区域获取该有价文件的光电信号的特征信息之后,可以根据该特征信息计算该光电信号的特征分量,该特征分量Mn的表达式为:
Figure PCTCN2016078506-appb-000007
θi为特征信息,i=1,2,…,t。
下面通过具体应用场景详细描述步骤207~209:如图3,基于步骤4中有价文件类别识别结果,获取预先设定矩形区域1,矩形区域2,矩形区域3的亮度或色度或饱和度或对比度特征信息θi,i=1,2,3,得到有价文件光电信号特征
Figure PCTCN2016078506-appb-000008
其中,n为有价文件的放入序号。
210、计算该有价文件的累积分量;
在根据该特征信息计算该光电信号的特征分量之后,可以计算该有价文件的累积分量
Figure PCTCN2016078506-appb-000009
mi为特征分量Mn在i时刻的值。
211、计算该有价文件的微分误差;
在计算该有价文件的累积分量之后,可以计算该有价文件的微分误差。如图4,a为有价文件光电特征曲线,b为标准曲线,M1为累积分量,其中
Figure PCTCN2016078506-appb-000010
如图5,a为累积分量曲线,b为标准曲线,计算累积误差,既信号修正量M2=k2*(M*-Ml),M*为预先设定的标准信息,k2为经验值,如图6,c为累积误差曲线,有价文件光电信号的微分误差Mw=Mn-Ml
212、根据该累积分量计算该光电信号的第二修正量;
在计算该有价文件的累积分量之后,可以根据该累积分量计算该光电信号的第二修正量M2=k2*(M*-Ml),M*为预设的标准信息,k2为预设的第二系数。
213、根据该微分误差计算得到该光电信号的第三修正量;
在计算该有价文件的微分误差之后,可以根据该微分误差计算得到该光电信号的第三修正量M3,具体包括:若|Mw|<w,则第三修正量M3=-Mw;若|Mw|≥w,当满足|Mw|≥w的该光电信号样本数为n,且
Figure PCTCN2016078506-appb-000011
时,则第三修正量M3=0,否则,第三修正量M3=-k3*Mw,N为该光电信号的样本总数,k3为预设的第三系数。
214、根据该累积分量、该第二修正量和该第三修正量得到总修正量;
在得到该累积分量、该第二修正量和该第三修正量之后,可以根据该累积分量、该第二修正量和该第三修正量得到总修正量,总修正量M=M1+M2+M3
215、更新该光电信号修正量M0的值等于该总修正量M;
在根据该累积分量、该第二修正量和该第三修正量得到总修正量之后,可以更新该光电信号修正量M0的值等于该总修正量M。
216、初始化该采集参数,更新该采集参数;
在更新该光电信号修正量M0之后,还可以初始化该采集参数,更新该采集参数E0=E0+λ·M0,E0的初始化值为预设值,λ为预设的修正系数。其中,λ表示在CIS传感器的正常打光范围内,平均提高一个灰度值需要增强光电强度。
217、输出该识别结果。
在更新该光电信号修正量和该采集参数之后,可以输出该识别结果。
当识别装置出现老化时,线性度失效,“线性反馈分析模块”利用比例环节对所反馈光电信号进行修正的方式存在较大的误差,可能出现冠字号码区域过曝或者过暗从而影响识别效果。本实施例中,通过有价文件自适应识别方法,能够解决识别装置老化导致冠字号码识别效果不好的问题。
有价识别装置出现老化或输入崭新钞票时。若出现白光透射图像过暗或过亮时,基于累积反馈自适应方法,即系统增强或减弱光电能量,再次输入钞票时,便可避免白光透射图像过暗或过曝的问题。
本实施例中,首先,获取采集参数,根据该采集参数采集有价文件的光电信号;获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿;然后,对数字补偿后的该光电信号进行特征提取,得到特征向量;将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果;接着,根据该识别结果获取有价文件上对应的特定区域;根据该特定区域获取该有价文件的光电信号的特征信息;根据该特征信息计算得到该有价文件的累积分量和微分误差;根据该累积分量和微分误差计算得到该光电信号的总修正量;最后,根据该总修正量更新该光电信号修正量和该采集参数;输出该识别结果。从而,可以实现有价文件识别过程的累积反馈自适应及微分反馈控制自适应,解决系统的累积误差问题和微分误差问题。
上面主要对一种有价文件自适应识别方法进行描述,下面将详细描述一种有价文件自适应识别装置,请参阅图7,本发明实施例中一种有价文件自适应识别装置一个实施例包括:
光电信号采集模块701,用于获取采集参数,根据该采集参数采集有价 文件的光电信号;
数字补偿模块702,用于获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿;
特征提取模块703,用于对数字补偿后的该光电信号进行特征提取,得到特征向量;
识别模块704,用于将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果;
特定区域获取模块705,用于根据该识别结果获取有价文件上对应的特定区域;
特征信息获取模块706,用于根据该特定区域获取该有价文件的光电信号的特征信息;
累积分量和微分误差计算模块707,用于根据该特征信息计算得到该有价文件的累积分量和微分误差;
总修正量计算模块708,用于根据该累积分量和微分误差计算得到该光电信号的总修正量;
更新模块709,用于根据该总修正量更新该光电信号修正量和该采集参数;
识别结果输出模块710,用于输出该识别结果。
本实施例中,首先,光电信号采集模块701获取采集参数,根据该采集参数采集有价文件的光电信号;数字补偿模块702获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿;然后,特征提取模块703对数字补偿后的该光电信号进行特征提取,得到特征向量;识别模块704将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果;接着,特定区域获取模块705根据该识别结果获取有价文件上对应的特定区域;特征信息获取模块706根据该特定区域获取该有价文件的光电信号的特征信息;再之,累积分量和微分误差计算模块707根据该特征信息计算得到该有价文件的累积分量和微分误差;总修正量计算模块708根据该累积分量和微分误差计算得到该光电信号的总修正量;更新模块709根据该总修正量更新该光电信号修正量和该采集参数;最后,识别结果输出模块710输出该识别 结果。从而,可以实现有价文件识别过程的累积反馈自适应及微分反馈控制自适应,解决系统的累积误差问题和微分误差问题。
为便于理解,下面对本发明实施例中的一种有价文件自适应识别装置进行详细描述,请参阅图8,本发明实施例中一种有价文件自适应识别装置另一个实施例包括:
光电信号采集模块801,用于获取采集参数,根据该采集参数采集有价文件的光电信号;
数字补偿模块802,用于获取光电信号修正量,根据该光电信号修正量对该光电信号进行数字补偿;
特征提取模块803,用于对数字补偿后的该光电信号进行特征提取,得到特征向量;
识别模块804,用于将该特征向量送入预设的分类器中识别,得到该有价文件的识别结果;
特定区域获取模块805,用于根据该识别结果获取有价文件上对应的特定区域;
特征信息获取模块806,用于根据该特定区域获取该有价文件的光电信号的特征信息;
累积分量和微分误差计算模块807,用于根据该特征信息计算得到该有价文件的累积分量和微分误差;
总修正量计算模块808,用于根据该累积分量和微分误差计算得到该光电信号的总修正量;
更新模块809,用于根据该总修正量更新该光电信号修正量和该采集参数;
识别结果输出模块810,用于输出该识别结果。
本实施例中该累积分量和微分误差计算模块807具体可以包括:
特征分量计算单元8071,用于根据该特征信息计算该光电信号的特征分量,该特征分量Mn的表达式为:
Figure PCTCN2016078506-appb-000012
θi为特征信息,i=1,2,…,t;
累积分量计算单元8072,用于计算该有价文件的累积分量
Figure PCTCN2016078506-appb-000013
mi为特征分量Mn在i时刻的值;
微分误差计算单元8073,用于计算该有价文件的微分误差Mw=Mn-M1
本实施例中该总修正量计算模块808具体可以包括:
第二修正量计算单元8081,用于根据该累积分量计算该光电信号的第二修正量M2=k2*(M*-Ml),M*为预设的标准信息,k2为预设的第二系数;
第三修正量计算单元8082,用于根据该微分误差计算得到该光电信号的第三修正量M3,具体包括:若|Mw|<w,则第三修正量M3=-Mw;若|Mw|≥w,当满足|Mw|≥w的该光电信号样本数为n,且
Figure PCTCN2016078506-appb-000014
时,则第三修正量M3=0,否则,第三修正量M3=-k3*Mw,N为该光电信号的样本总数,k3为预设的第三系数;
总修正量计算单元8083,用于根据该累积分量、该第二修正量和该第三修正量得到总修正量,总修正量M=M1+M2+M3
本实施例中该更新模块809具体可以包括:
光电信号修正量更新单元8091,用于更新该光电信号修正量M0的值等于该总修正量M;
采集参数更新单元8092,用于初始化该采集参数,更新该采集参数E0=E0+λ·M0,E0的初始化值为预设值,λ为预设的修正系数。
本实施例中该装置还可以包括:
采集参数初始化值获取模块811,用于当第一次采集有价文件的光电信号时,获取采集参数的预设初始化值;
修正量初始化值获取模块812,用于当第一次采集有价文件的光电信号时,获取光电信号修正量的初始化值,该光电信号修正量的初始化值为0。
所属领域的技术人员可以清楚地了解到,为描述的方便和简洁,上述描述的系统,装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
在本申请所提供的几个实施例中,应该理解到,所揭露的系统,装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,该单元的划分,仅仅为一种逻辑功能划分,实际实现时可以 有另外的划分方式,例如多个单元或组件可以结合或者可以集成到另一个系统,或一些特征可以忽略,或不执行。另一点,所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口,装置或单元的间接耦合或通信连接,可以是电性,机械或其它的形式。
该作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外,在本发明各个实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用软件功能单元的形式实现。
该集成的单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中。基于这样的理解,本发明的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行本发明各个实施例该方法的全部或部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(ROM,Read-Only Memory)、随机存取存储器(RAM,Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
以上该,以上实施例仅用以说明本发明的技术方案,而非对其限制;尽管参照前述实施例对本发明进行了详细的说明,本领域的普通技术人员应当理解:其依然可以对前述各实施例所记载的技术方案进行修改,或者对其中部分技术特征进行等同替换;而这些修改或者替换,并不使相应技术方案的本质脱离本发明各实施例技术方案的精神和范围。

Claims (10)

  1. 一种有价文件自适应识别方法,其特征在于,包括:
    获取采集参数,根据所述采集参数采集有价文件的光电信号;
    获取光电信号修正量,根据所述光电信号修正量对所述光电信号进行数字补偿;
    对数字补偿后的所述光电信号进行特征提取,得到特征向量;
    将所述特征向量送入预设的分类器中识别,得到所述有价文件的识别结果;
    根据所述识别结果获取有价文件上对应的特定区域;
    根据所述特定区域获取所述有价文件的光电信号的特征信息;
    根据所述特征信息计算得到所述有价文件的累积分量和微分误差;
    根据所述累积分量和微分误差计算得到所述光电信号的总修正量;
    根据所述总修正量更新所述光电信号修正量和所述采集参数;
    输出所述识别结果。
  2. 根据权利要求1所述的方法,其特征在于,数字补偿的修正公式如下:
    p'=p+M0
    p是所述光电信号任意一点的灰度值,p'为p的修正值,M0为所述光电信号修正量。
  3. 根据权利要求1所述的方法,其特征在于,根据所述特征信息计算得到所述有价文件的累积分量和微分误差具体包括:
    根据所述特征信息计算所述光电信号的特征分量,所述特征分量Mn的表达式为:
    Figure PCTCN2016078506-appb-100001
    θi为特征信息,i=1,2,…,t;
    计算所述有价文件的累积分量
    Figure PCTCN2016078506-appb-100002
    mi为特征分量Mn在i时刻的值;
    计算所述有价文件的微分误差Mw=Mn-M1
  4. 根据权利要求3所述的方法,其特征在于,根据所述累积分量和微分误差计算得到所述光电信号的总修正量具体包括:
    根据所述累积分量计算所述光电信号的第二修正量M2=k2*(M*-Ml),M*为预设的标准信息,k2为预设的第二系数;
    根据所述微分误差计算得到所述光电信号的第三修正量M3,具体包括:若|Mw|<w,则第三修正量M3=-Mw;若|Mw|≥w,当满足|Mw|≥w的所述光电信号样本数为n,且
    Figure PCTCN2016078506-appb-100003
    时,则第三修正量M3=0,否则,第三修正量M3=-k3*Mw,N为所述光电信号的样本总数,k3为预设的第三系数;
    根据所述累积分量、所述第二修正量和所述第三修正量得到总修正量,总修正量M=M1+M2+M3
  5. 根据权利要求4所述的方法,其特征在于,根据所述总修正量更新所述光电信号修正量和所述采集参数包括:
    更新所述光电信号修正量M0的值等于所述总修正量M;
    初始化所述采集参数,更新所述采集参数E0=E0+λ·M0,E0的初始化值为预设值,λ为预设的修正系数。
  6. 根据权利要求1所述的方法,其特征在于,根据所述光电信号修正量对所述光电信号进行数字补偿之前还包括:
    获取预设的第一校正系数和第二校正系数;
    根据所述第一校正系数和第二校正系数对所述光电信号进行信号补偿,补偿修正公式如下:
    y=a·x+b;
    x为任一点修正前的光电信号值,y为对应的修正后的光电信号值,a为第一校正系数,b为第二校正系数。
  7. 根据权利要求1所述的方法,其特征在于,当第一次采集有价文件的光电信号时,获取采集参数的预设初始化值,获取光电信号修正量的初始化值,所述光电信号修正量的初始化值为0。
  8. 一种有价文件自适应识别装置,其特征在于,包括:
    光电信号采集模块,用于获取采集参数,根据所述采集参数采集有价文件的光电信号;
    数字补偿模块,用于获取光电信号修正量,根据所述光电信号修正量对所述光电信号进行数字补偿;
    特征提取模块,用于对数字补偿后的所述光电信号进行特征提取,得到特征向量;
    识别模块,用于将所述特征向量送入预设的分类器中识别,得到所述有 价文件的识别结果;
    特定区域获取模块,用于根据所述识别结果获取有价文件上对应的特定区域;
    特征信息获取模块,用于根据所述特定区域获取所述有价文件的光电信号的特征信息;
    累积分量和微分误差计算模块,用于根据所述特征信息计算得到所述有价文件的累积分量和微分误差;
    总修正量计算模块,用于根据所述累积分量和微分误差计算得到所述光电信号的总修正量;
    更新模块,用于根据所述总修正量更新所述光电信号修正量和所述采集参数;
    识别结果输出模块,用于输出所述识别结果。
  9. 根据权利要求8所述的装置,其特征在于,
    所述累积分量和微分误差计算模块具体包括:
    特征分量计算单元,用于根据所述特征信息计算所述光电信号的特征分量,所述特征分量Mn的表达式为:
    Figure PCTCN2016078506-appb-100004
    θi为特征信息,i=1,2,…,t;
    累积分量计算单元,用于计算所述有价文件的累积分量
    Figure PCTCN2016078506-appb-100005
    mi为特征分量Mn在i时刻的值;
    微分误差计算单元,用于计算所述有价文件的微分误差Mw=Mn-M1
    所述总修正量计算模块具体包括:
    第二修正量计算单元,用于根据所述累积分量计算所述光电信号的第二修正量M2=k2*(M*-Ml),M*为预设的标准信息,k2为预设的第二系数;
    第三修正量计算单元,用于根据所述微分误差计算得到所述光电信号的第三修正量M3,具体包括:若|Mw|<w,则第三修正量M3=-Mw;若|Mw|≥w,当满足|Mw|≥w的所述光电信号样本数为n,且
    Figure PCTCN2016078506-appb-100006
    时,则第三修正量M3=0,否则,第三修正量M3=-k3*Mw,N为所述光电信号的样本总数,k3为预设的第三系数;
    总修正量计算单元,用于根据所述累积分量、所述第二修正量和所述第 三修正量得到总修正量,总修正量M=M1+M2+M3
    所述更新模块具体包括:
    光电信号修正量更新单元,用于更新所述光电信号修正量M0的值等于所述总修正量M;
    采集参数更新单元,用于初始化所述采集参数,更新所述采集参数E0=E0+λ·M0,E0的初始化值为预设值,λ为预设的修正系数。
  10. 根据权利要求8所述的装置,其特征在于,所述装置还包括:
    采集参数初始化值获取模块,用于当第一次采集有价文件的光电信号时,获取采集参数的预设初始化值;
    修正量初始化值获取模块,用于当第一次采集有价文件的光电信号时,获取光电信号修正量的初始化值,所述光电信号修正量的初始化值为0。
PCT/CN2016/078506 2015-12-02 2016-04-06 有价文件自适应识别方法和装置 WO2017092209A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP16869523.7A EP3385923A4 (en) 2015-12-02 2016-04-06 Self-adaptive identification method of identifying negotiable instrument and device
RU2018118168A RU2690716C1 (ru) 2015-12-02 2016-04-06 Способ и устройство самоадаптивной идентификации для идентифицирования защищенного документа
US15/774,560 US10529163B2 (en) 2015-12-02 2016-04-06 Self-adaptive identification method of identifying negotiable instrument and device
ZA2018/02811A ZA201802811B (en) 2015-12-02 2018-04-26 Self-adaptive identification method of identifying negotiable instrument and device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510874880.XA CN105528825B (zh) 2015-12-02 2015-12-02 有价文件自适应识别方法和装置
CN201510874880.X 2015-12-02

Publications (1)

Publication Number Publication Date
WO2017092209A1 true WO2017092209A1 (zh) 2017-06-08

Family

ID=55771025

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/078506 WO2017092209A1 (zh) 2015-12-02 2016-04-06 有价文件自适应识别方法和装置

Country Status (7)

Country Link
US (1) US10529163B2 (zh)
EP (1) EP3385923A4 (zh)
CN (1) CN105528825B (zh)
CL (1) CL2018001162A1 (zh)
RU (1) RU2690716C1 (zh)
WO (1) WO2017092209A1 (zh)
ZA (1) ZA201802811B (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106408744B (zh) * 2016-08-24 2019-03-15 中国人民银行印制科学技术研究所 票据特征值的检测系统和票据特征值的检测方法
CN109658586B (zh) * 2018-11-20 2021-04-23 广州广电运通金融电子股份有限公司 薄片类可堆叠物体的数量校准方法、装置和计算机设备
CN114544638A (zh) * 2020-11-25 2022-05-27 南宁富联富桂精密工业有限公司 空气湿度量测方法及电子装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101681532A (zh) * 2007-03-30 2010-03-24 德国捷德有限公司 检查有价文件的方法和设备
CN101968903A (zh) * 2010-09-21 2011-02-09 广州广电运通金融电子股份有限公司 有价文件识别方法及系统
CN102890841A (zh) * 2012-10-08 2013-01-23 广州广电运通金融电子股份有限公司 一种有价文件鉴别方法及装置

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6731785B1 (en) * 1999-07-26 2004-05-04 Cummins-Allison Corp. Currency handling system employing an infrared authenticating system
DE10151854A1 (de) 2001-10-24 2003-05-08 Giesecke & Devrient Gmbh Wertdokumentbearbeitungsvorrichtung, Verfahren zur Funktionsprüfung, Justierung und/oder Kalibrierung der Wertdokumentbearbeitungsvorrichtung und Testmedien zur Durchführung des Verfahrens
EP1908027B1 (en) * 2005-07-27 2010-09-29 Ingenia Holdings Limited Verification of authenticity
EP1868166A3 (en) 2006-05-31 2007-12-26 MEI, Inc. Method and apparatus for validating banknotes
GB0612856D0 (en) * 2006-06-28 2006-08-09 Rue De Int Ltd Document handling apparatus
JP2009059177A (ja) * 2007-08-31 2009-03-19 Kalbas Japan Ltd 紙葉類の真贋判定システム及び真贋判定方法
CN101458836B (zh) * 2007-12-14 2010-11-03 广州广电运通金融电子股份有限公司 有价票据真伪鉴别方法及系统
RU2543931C2 (ru) * 2009-08-11 2015-03-10 КБА-НотаСис СА Проверка подлинности защищенных документов, в частности банкнот
CN101908241B (zh) 2010-08-03 2012-05-16 广州广电运通金融电子股份有限公司 有价文件识别方法及其识别系统
JP2014186373A (ja) 2013-03-21 2014-10-02 Toshiba Corp 紙葉類処理方法、紙葉類処理装置、及び紙葉類処理システム
CN103310528B (zh) 2013-07-08 2016-08-17 广州广电运通金融电子股份有限公司 图像补偿修正方法及识别验钞装置
CN105095895B (zh) * 2015-04-23 2018-09-25 广州广电运通金融电子股份有限公司 有价文件识别装置自修正识别方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101681532A (zh) * 2007-03-30 2010-03-24 德国捷德有限公司 检查有价文件的方法和设备
CN101968903A (zh) * 2010-09-21 2011-02-09 广州广电运通金融电子股份有限公司 有价文件识别方法及系统
CN102890841A (zh) * 2012-10-08 2013-01-23 广州广电运通金融电子股份有限公司 一种有价文件鉴别方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3385923A4 *

Also Published As

Publication number Publication date
CL2018001162A1 (es) 2018-09-14
US10529163B2 (en) 2020-01-07
US20190340862A1 (en) 2019-11-07
CN105528825B (zh) 2018-08-31
RU2690716C1 (ru) 2019-06-05
CN105528825A (zh) 2016-04-27
EP3385923A4 (en) 2018-12-26
ZA201802811B (en) 2019-01-30
EP3385923A1 (en) 2018-10-10

Similar Documents

Publication Publication Date Title
US10366309B2 (en) Image quality assessment and improvement for performing optical character recognition
EP1600893B1 (en) Radiometric calibration from a single image
WO2017092209A1 (zh) 有价文件自适应识别方法和装置
Brusa et al. Increasing the precision of canopy closure estimates from hemispherical photography: Blue channel analysis and under-exposure
CN108734108B (zh) 一种基于ssd网络的裂纹舌识别方法
US8149300B2 (en) Radiometric calibration from noise distributions
CN103997590B (zh) 一种图像校准方法及装置
Nixon et al. Accurate device-independent colorimetric measurements using smartphones
CN103197444B (zh) 一种液晶显示器响应时间的测量方法及系统
CN108055530A (zh) 图像传感器的校正方法、装置、atm设备及存储介质
CN102110323A (zh) 一种验钞方法及验钞装置
CN111291778B (zh) 深度分类模型的训练方法、曝光异常检测方法以及装置
Sintorn et al. Gradient based intensity normalization
RU2678491C1 (ru) Способ контроля качества при проверке подлинности банкнот и система для этого
WO2015085817A1 (zh) 一种多光谱券类质量检测方法和系统
US9934434B2 (en) Determining image forensics using an estimated camera response function
CN101151882A (zh) 处理扫描数据的系统和方法
WO2020201623A1 (en) Determining tooth color shade based on image obtained with a mobile device
CN104568639A (zh) 一种水果糖度的确定方法和装置
Li et al. A novel method for seed cotton color measurement based on machine vision technology
CN115292722A (zh) 基于不同色彩空间的模型安全检测方法和装置
CN108346213B (zh) 纸币图像的特征识别方法及装置
Ullah et al. Gray-level image transformation of paved road cracks with metaphorical and computational analysis
WO2018053977A1 (zh) 一种图像补偿方法和装置
CN109300138A (zh) 一种基于两步纳什均衡改进c-v模型的医学图像分割方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16869523

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 2018118168

Country of ref document: RU