WO2016123903A1 - 一种褶皱票据鉴别方法及装置 - Google Patents

一种褶皱票据鉴别方法及装置 Download PDF

Info

Publication number
WO2016123903A1
WO2016123903A1 PCT/CN2015/083861 CN2015083861W WO2016123903A1 WO 2016123903 A1 WO2016123903 A1 WO 2016123903A1 CN 2015083861 W CN2015083861 W CN 2015083861W WO 2016123903 A1 WO2016123903 A1 WO 2016123903A1
Authority
WO
WIPO (PCT)
Prior art keywords
pass
image
cft
filtering
low
Prior art date
Application number
PCT/CN2015/083861
Other languages
English (en)
French (fr)
Inventor
刘光禄
陈健
肖助明
郑伟锐
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 EP15880858.4A priority Critical patent/EP3255617B1/en
Priority to US15/544,379 priority patent/US10319170B2/en
Priority to RU2017130019A priority patent/RU2673120C1/ru
Publication of WO2016123903A1 publication Critical patent/WO2016123903A1/zh
Priority to HK18105825.0A priority patent/HK1246483A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00—Testing 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
    • G07D7/181—Testing mechanical properties or condition, e.g. wear or tear
    • G07D7/183—Detecting folds or doubles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • G06V10/446—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering using Haar-like filters, e.g. using integral image techniques
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
    • G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
    • G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
    • G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00—Testing 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
    • G07D7/06—Testing 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
    • G07D7/12—Visible light, infrared or ultraviolet radiation
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00—Testing 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
    • G07D7/06—Testing 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
    • G07D7/12—Visible light, infrared or ultraviolet radiation
    • G07D7/121—Apparatus characterised by sensor details
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00—Testing 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
    • G07D7/20—Testing patterns thereon
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00—Testing 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
    • G07D7/20—Testing patterns thereon
    • G07D7/2008—Testing patterns thereon using pre-processing, e.g. de-blurring, averaging, normalisation or rotation
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D7/00—Testing 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
    • G07D7/20—Testing patterns thereon
    • G07D7/202—Testing patterns thereon using pattern matching
    • G07D7/2041—Matching statistical distributions, e.g. of particle sizes orientations
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07D—HANDLING OF COINS OR VALUABLE PAPERS, e.g. TESTING, SORTING BY DENOMINATIONS, COUNTING, DISPENSING, CHANGING OR DEPOSITING
    • G07D2207/00—Paper-money testing devices

Definitions

  • the invention relates to the technical field of financial self-service devices, and in particular to a method and a device for identifying wrinkle bills.
  • a pleated note which is one of the non-negotiable note types, is no longer suitable for recirculation, so when the pleated note is entered as an input into the authentication device, the authentication device needs to classify it as a non-negotiable note.
  • the signal obtained by the signal acquisition module of the self-service device is not robust enough, and thus the characteristics of the wrinkle bill in the signal are not significant enough, thereby increasing the pair.
  • the difficulty of identifying pleated notes due to the influence of existing imaging equipment, illumination changes, imaging environment and other factors, the signal obtained by the signal acquisition module of the self-service device is not robust enough, and thus the characteristics of the wrinkle bill in the signal are not significant enough, thereby increasing the pair. The difficulty of identifying pleated notes.
  • Feature description is the key premise of wrinkle ticket identification. Based on the existing signal, it is difficult to use the traditional simple gray-scale mean, or use the threshold to binarize the image, and count the single feature description method of the abnormal pixel.
  • the pleated notes are distinguished from the disturbed or contaminated notes. The underlying reason why traditional characterizations do not achieve good results is that there is no effective classification or pre-treatment of pleated notes and disturbed or contaminated notes.
  • the present invention provides a wrinkle bill discriminating method and apparatus, which adopts a feature classification method based on a high/low pass filter.
  • the characteristic area of the bill folds is effectively characterized and ultimately The discriminating performance of the discriminating device on the wrinkle bill is improved.
  • the invention provides a pleated bill discriminating device, comprising: a bill input port for accepting a bill or bill sample to be authenticated, and transporting the bill to a next module; and a signal collecting module for collecting the bill CIS image Obtaining an infrared transmission image T and an infrared reflection image F; a signal identification module for identifying whether the ticket to be identified has wrinkles; and a receiving/rejecting module for receiving or rejecting the authentication ticket; wherein
  • the signal discriminating module further comprises: a high-pass filtering first unit for filtering the infrared transmission image T to obtain a high-pass infrared transmission filtering image gT; and a low-pass filtering first unit for filtering the infrared transmission image T Low-pass infrared transmission filtering image dT; a high-pass filtering second unit for high-pass filtering of the infrared reflection image F according to the low-pass filtering condition of the infrared transmission image T, and obtaining high-pass infrared
  • the ticket classification decision model is: p 1 >T 1 , p 2 >T 2 , p 3 >T 3
  • the ticket to be authenticated is identified as a wrinkle note, and vice versa, it is identified as a non-pleated note, wherein p 1 , p 2 and p 3 are respectively the confidence that the note to be authenticated is determined to be a wrinkle note, T 1 , T 2 and T 3 are three confidence thresholds.
  • the ticket classification decision model can be further modified to:
  • T s is the threshold, and the general empirical value of T s is 0.5.
  • the invention also provides a method for identifying a wrinkle bill, comprising: step one, the bill input port accepts the bill to be authenticated, and delivers the bill to be authenticated to the signal acquisition module; and in step two, the signal acquisition module collects the CIS image of the bill to be authenticated
  • the signal obtains the infrared transmission image T s and the infrared reflection image F s ; in step 3, the first unit of the high-pass filtering filters the infrared transmission image T s to obtain a high-pass infrared transmission filtered image gT s ; and the fourth step, the low-pass filtering first unit
  • the infrared transmission image T s is filtered to obtain a low-pass infrared transmission filtered image dT s ; in step 5, the second unit of the high-pass filtering according to the low-pass filtering of the infrared transmission image T s , according to the geometric coordinate point-to-point mapping relationship to the infrared reflection map F s synchronous
  • Step 8 wherein extracting a first unit gT s for feature extraction calculates the average gray gT s of gT_G s value as the feature value; step 9, the second feature extraction unit dF s feature extraction calculation of dF s average gradation value as a characteristic value dF_G s; step 10, the third feature extraction unit cFT s feature extraction, cFT s calculated average gray value as a characteristic value cFT_G s; step 11, the feature value gT_G s, dF_G s feature value and feature value cFT_G s are substituted into pleated bill with three non-corrugated notes In the classification models y 1 , y 2 , y 3 ,
  • p 1 , p 2 and p 3 are the confidences of the decision to be identified as a wrinkle note, respectively. If p 1 >T 1 , p 2 >T 2 , p 3 >T 3 are simultaneously established, then the ticket to be authenticated is It is identified as a pleated note, and vice versa, identified as a non-pleated note, where T 1 , T 2 and T 3 are three confidence thresholds, and the empirical value is typically 0.5.
  • the method for obtaining the three classification models y 1 , y 2 , y 3 of the pleated and non-pleated tickets comprises: collecting a certain number of pleated and non-pleated sample samples, obtaining high-pass infrared transmission of each sample of the ticket
  • the average gray value gT_G characteristic value of the filter map gT, the average gray value dF_G feature value of the low-pass infrared reflection filter map dF, and the average gray value cFT_G feature value of the differential filter map cFT and respectively count the gT_G feature value, dF_G
  • the eigenvalue and the cFT_G eigenvalue are calculated, and the gT_G probability distribution map, the dF_G probability distribution map and the cFT_G probability distribution map corresponding to the pleated bill are calculated, and are respectively described by the following formula:
  • y 1 , y 2 , y 3 are the three classification models of the pleated and non-pleated notes, respectively.
  • the average gray value gT_G characteristic value of the high-pass infrared transmission filter map gT of each ticket sample, the average gray value dF_G characteristic value of the low-pass infrared reflection filter map dF, and the average gray value of the differential filter map cFT are obtained.
  • the method of the average gray value cFT_G s characteristic value of the figure cFT s is the same.
  • the steps 1 to 10 are not performed in sequence, wherein the third step and the fourth step are performed synchronously, the step 5 and the step 6 are performed simultaneously, the step 8 is performed after the step 3, and the step 9 is performed after the step 6.
  • step ten can be performed after step seven.
  • the method further includes: obtaining a high/low pass filter threshold for the image signals T s and F s : first obtaining an average gray value of T s :
  • T s is the corresponding pixel gray value
  • w is the width of the image signal T s
  • h is the height of the image signal T s
  • the calculation of obtaining gT s gT_G s average gray value model, the model obtaining model calculation obtaining the average gray value of dF s dF_G s cFT s and calculating the average gradation value and obtaining cFT_G s The average gray value of T s is obtained in the same model.
  • the features are effectively classified, which greatly improves the distinguishability of features.
  • different types of features correspond to different classifiers, and the classifiers have functions similar to Adaboost classifiers, which can ensure the identification of the present invention.
  • the reliability makes the authentication system more robust and compatible with external environment interference, the pollution of the bill itself, etc., and the entire pleated bill solution and device can effectively identify the wrinkle bill.
  • FIG. 1 is a schematic structural view of a pleated bill discriminating device according to a preferred embodiment of the present invention
  • FIG. 2 is a flow chart of a method for identifying a wrinkle ticket according to a preferred embodiment of the present invention.
  • the pleated bill discriminating device includes a bill input port 10, a signal acquisition module 20, a signal discriminating module 30, and a receiving/rejecting module 40.
  • the ticket input port 10 is configured to accept a ticket or ticket sample to be authenticated and deliver the ticket to the next module.
  • the signal acquisition module 20 is configured to collect the CIS image of the ticket to obtain an infrared transmission image T and an infrared reflection image F.
  • the signal authentication module 30 is configured to identify whether the ticket to be authenticated has wrinkles.
  • the receiving/rejecting module 40 is configured to receive or reject the ticket to be authenticated.
  • the signal discriminating module 30 further includes: a high pass filtering first unit, a low pass filtering first unit, a high pass filtering second unit, a low pass filtering second unit, a differential filtering unit, and a The feature extraction first unit, a feature extraction second unit, a feature extraction third unit, and an authentication decision unit.
  • the high-pass filtering first unit is configured to filter the infrared transmission image T to obtain a high-pass infrared transmission filtering image gT;
  • the low-pass filtering first unit is configured to filter the infrared transmission image T to obtain a low-pass infrared transmission filtering image.
  • the high-pass filtering second unit is configured to perform high-pass filtering on the infrared reflection map F according to the low-pass filtering condition of the infrared transmission image T according to the mapping relationship of the geometric coordinate points to the point, to obtain a high-pass infrared reflection filtering image gF;
  • the low-pass filtering The second unit is configured to perform low-pass filtering on the infrared reflection pattern F according to the mapping relationship between the geometric coordinate points and the points according to the high-pass filtering condition of the infrared transmission map T, to obtain a low-pass infrared reflection filtered image dF;
  • the differential filtering unit is used for The high-pass infrared reflection filtering image gF and the low-pass infrared transmission filtering image dT are differentially operated to obtain a differential filtering image cFT;
  • the feature extraction first unit is used for feature extraction of the high-pass infrared transmission filtering image gT, and the average gray value of the gT is
  • the discriminating decision unit is configured to calculate a classification model of the wrinkled bill and the non-pleated bill according to the gT_G feature value, the dF_G feature value and the cFT_G feature value of the bill sample, and determine whether the banknote to be identified has wrinkles according to the bill classification decision model, wherein the bill classification
  • the decision model is: p 1 >T 1 , p 2 >T 2 , and when p 3 >T 3 is established, the ticket to be identified is identified as a wrinkle note, and vice versa, as a non-pleated note, where p 1 , p
  • the bill classification decision model can be further amended to:
  • T s is the threshold, and the general empirical value of T s is 0.5.
  • the pleated ticket identification method includes the following steps: the ticket input port accepts the ticket to be authenticated, and the ticket to be authenticated is sent to the signal acquisition module; and in step 2, the signal acquisition module collects the CIS image signal of the ticket to be authenticated.
  • step 3 the first unit of the high-pass filtering filters the infrared transmission image T s to obtain a high-pass infrared transmission filtered image gT s ; and step 4, the first unit pair of the low-pass filtering
  • the infrared transmission image T s is filtered to obtain a low-pass infrared transmission filtered image dT s ;
  • step 5 the second unit of the high-pass filtering according to the low-pass filtering of the infrared transmission image T s , according to the geometric coordinate point-to-point mapping relationship to the infrared reflection map F s synchronous high-pass filtering to obtain high-pass infrared reflection filtered image gF s ;
  • step 6 low-pass filtering second unit according to the high-pass filtering of the infrared transmission map T s , according to the geometric coordinate point-to-point mapping relationship to the infrared reflection map
  • p 1 , p 2 and p 3 are respectively the confidences of the decision to be identified as a wrinkle note; and then according to the bill classification decision model, whether the banknote to be identified has wrinkles, wherein the bill classification decision model is: if p 1 >T 1 , p 2 >T 2 , p 3 >T 3 is established at the same time, the ticket to be identified is identified as a wrinkle note, and vice versa, identified as a non-pleated note, wherein T 1 , T 2 and T 3 are three confidence levels Threshold, the process ends.
  • ⁇ + ⁇ + ⁇ 1, ⁇ ⁇ 0, ⁇ ⁇ 0, ⁇ ⁇ 0, then exists
  • T s the bill classification decision model is as follows:
  • Steps 1 to 10 are not performed in sequence, wherein step 3 and step 4 can be performed simultaneously, step 5 and step 6 can be performed simultaneously, step 8 can be performed after step 3, step 9 can be performed after step 6, and Step 10 can be performed after step 7.
  • the method of obtaining the three classification models y 1 , y 2 , y 3 of the pleated and non-pleated notes comprises: collecting a number of pleated and non-pleated note samples to obtain high-pass infrared transmission filtering for each sample of the ticket Figure gT average gray value gT_G eigenvalue, low-pass infrared reflection filter dF average gray value dF_G eigenvalue and differential filter cFT average gray value cFT_G eigenvalue, and separately calculate the gT_G eigenvalue, dF_G feature
  • the value and the cFT_G eigenvalue are calculated, and the gT_G probability distribution map, the dF_G probability distribution map and the cFT_G probability distribution map corresponding to the pleated bill are calculated, and are respectively described by the following formula:
  • y 1 , y 2 , y 3 are the three classification models of the pleated and non-pleated notes, respectively.
  • the average gray value gT_G characteristic value of the high-pass infrared transmission filter map gT of each ticket sample the average gray value dF_G characteristic value of the low-pass infrared reflection filter map dF, and the average gray value of the differential filter map cFT are obtained.
  • the pleated bill A is placed in the self-service device receiving port
  • step 2 when the ticket A is mechanically transmitted through the signal acquisition module 20, the signal acquisition module 20 will perform signal acquisition on the ticket A, wherein the acquired CIS infrared transmission image signal is rT, infrared reflection image signal rF;
  • step 3 the high/low pass filter thresholds are first obtained for the image signals rT and rF:
  • pix(i) is the gray value corresponding to the pixel point on rT
  • w is the width of the rT image signal
  • h is the height of the rT image signal.
  • step 3 high-pass filtering is performed on the first unit by high-pass filtering to obtain a high-pass filter map GT;
  • step 4 using low-pass filtering first unit to low-pass filter rT to obtain a low-pass filter map DT;
  • step 5 using the mapping relationship of the geometric coordinates, performing high-pass filtering on the rF to obtain a high-pass filter map GF;
  • step 6 using the mapping relationship of the geometric coordinates, performing correlation low-pass filtering on the rF to obtain a low-pass filter map DF;
  • step 7 performing a difference operation on the high-pass filter map GF and the low-pass filter map DT, To the differential filter CFT;
  • step 10 obtaining the average gray value cAVG of the differential filter map CFT as the feature value, and obtaining the model is the same as formula (9);
  • step 9 obtaining the average gray value dAVG of the low-pass filter map DF as the feature value, and obtaining the model is the same as formula (9);
  • step 8 obtaining the average gray value gAVG of the high-pass filter map GT as the feature value, and obtaining the model is the same as formula (9);
  • step 11 inputting the CAVG, dAVG, gAVG into the multi-feature fusion decision unit, through the learned multi-feature classification probability distribution model f 1 (x 1 ), f 2 (x 2 ), f 3 ( x 3 ) is classified, if f 1 (cAVG)>T 1 , f 2 (dAVG)>T 2 , f 3 (gAVG)>T 3 , where T 1 , T 2 and T 3 are empirical thresholds, generally 0.5 When the decision unit output is true, the ticket A is identified as a wrinkle note, and when the decision unit output is false, the ticket A is identified as a non-pleated note, ending the identification function.
  • the method and apparatus for identifying wrinkle bills provided by this embodiment effectively classify features by adopting high/low pass filtering, which greatly improves the distinguishability of features.
  • different types of features correspond to different classifiers, and the classifiers have functions similar to Adaboost classifiers, which can ensure the authentication confidence of the present invention, making the authentication system more robust to external environment interference, the ticket itself.
  • Adaboost classifiers which can ensure the authentication confidence of the present invention, making the authentication system more robust to external environment interference, the ticket itself.
  • the entire pleated bill solution and apparatus can effectively identify wrinkle bills.

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)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Analytical Chemistry (AREA)
  • Immunology (AREA)
  • Biochemistry (AREA)
  • Pathology (AREA)
  • Chemical & Material Sciences (AREA)
  • Multimedia (AREA)
  • Inspection Of Paper Currency And Valuable Securities (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

一种褶皱票据鉴别方法及装置，包括：票据输入口（10），用于接受待鉴别票据或票据样本；信号采集模块（20），用于对票据CIS图像进行采集，得到红外透射图像T和红外反射图像F；信号鉴别模块（30），用于鉴别待鉴别票据是否有褶皱；以及接收/拒收模块（40），用于对待鉴别票据进行接收或拒收操作，整个装置能够有效地鉴别褶皱票据。

Description

一种褶皱票据鉴别方法及装置
本申请要求于2015年02月04日提交中国专利局、申请号为201510059223.X、发明名称为“一种褶皱票据鉴别方法及装置”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
本发明涉及金融自助设备技术领域，尤其涉及一种褶皱票据鉴别方法及装置。
背景技术
作为非流通票据类型之一的褶皱票据，已不适合再流通，因此，当褶皱票据作为输入、进入到鉴别装置时，鉴别装置需将其鉴别分类为非流通票据。
但受限于现有成像设备、光照变化、成像环境等因素的影响，使得自助设备的信号采集模块得到的信号不够鲁棒，进而影响褶皱票据在信号中的特征不够显著，从而加大了对褶皱票据的鉴别难度。
特征描述是褶皱票据鉴别的关键前提，基于现有的信号，假若采用传统简单的求灰度均值，或采用阈值对图像进行二值化，并统计异常像素点的单一特征描述方法，很难将褶皱票据与受干扰或受污染的票据区分开来。传统的特征描述之所以达不到好的效果的根本原因在于，未对褶皱票据与受干扰或受污染的票据进行有效的分类或预处理。
发明内容
为了解决现有技术中难以将褶皱票据与受干扰或受污染的票据区分开来的问题，本发明提出一种褶皱票据鉴别方法及装置，采用基于高/低通滤波器的特征分类方法，对票据褶皱的特征区域进行有效的特征描述，最终 提高鉴别装置对褶皱票据的鉴别性能。
本发明提供的一种褶皱票据鉴别装置，包括：一票据输入口，用于接受待鉴别票据或票据样本，并将票据输送至下一模块；一信号采集模块，用于对票据CIS图像进行采集，得到红外透射图像T和红外反射图像F；一信号鉴别模块，用于鉴别待鉴别票据是否有褶皱；以及一接收/拒收模块，用于对待鉴别票据进行接收或拒收操作；其中，该信号鉴别模块进一步包括：一高通滤波第一单元，用于对红外透射图像T进行滤波，得到高通红外透射滤波图像gT；一低通滤波第一单元，用于对红外透射图像T进行滤波，得到低通红外透射滤波图像dT；一高通滤波第二单元，用于根据红外透射图像T的低通滤波情况，按照几何坐标点对点的映射关系对红外反射图F同步进行高通滤波，得到高通红外反射滤波图像gF；一低通滤波第二单元，用于根据红外透射图T的高通滤波情况，按照几何坐标点对点的映射关系对红外反射图F同步进行低通滤波，得到低通红外反射滤波图像dF；一差分滤波图单元，用于对高通红外反射滤波图像gF与低通红外透射滤波图像dT进行差分运算，得到差分滤波图像cFT；一特征提取第一单元，用于对高通红外透射滤波图像gT进行特征提取，计算gT的平均灰度值gT_G作为其特征值；一特征提取第二单元，用于对低通红外反射滤波图像dF进行特征提取，计算dF的平均灰度值dF_G作为其特征值；一特征提取第三单元，用于对差分滤波图像cFT进行特征提取，计算cFT的平均灰度值cFT_G作为其特征值；以及一鉴别决策单元，用于根据票据样本的gT_G特征值、dF_G特征值及cFT_G特征值计算褶皱票据与非褶皱票据的分类模型，并依据票据分类决策模型决策待鉴别钞票是否有褶皱，其中该票据分类决策模型为：p1>T1,p2>T2,p3>T3同时成立时，则待鉴别票据被鉴别为褶皱票据，反之，则鉴别为非褶皱票据，其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，T1,T2及T3为三个置信度阈值。
优选的，该票据分类决策模型可进一步修正为：
Figure PCTCN2015083861-appb-000001
其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，α,β,γ是分别对p1,p2及p3给予的不同的权重值，且α+β+γ＝1,α≥0,β≥0,γ≥0，Ts为阈值，Ts的一般经验值为0.5。
本发明还提供一种褶皱票据鉴别方法，包括：步骤一，票据输入口接受待鉴别票据，并将该待鉴别票据输送至信号采集模块；步骤二，信号采集模块采集该待鉴别票据的CIS图像信号，得到红外透射图像Ts和红外反射图像Fs；步骤三，高通滤波第一单元对红外透射图像Ts进行滤波，得到高通红外透射滤波图像gTs；步骤四，低通滤波第一单元对红外透射图像Ts进行滤波，得到低通红外透射滤波图像dTs；步骤五，高通滤波第二单元根据红外透射图像Ts的低通滤波情况，按照几何坐标点对点的映射关系对红外反射图Fs同步进行高通滤波，得到高通红外反射滤波图像gFs；步骤六，低通滤波第二单元根据红外透射图Ts的高通滤波情况，按照几何坐标点对点的映射关系对红外反射图Fs同步进行低通滤波，得到低通红外反射滤波图像dFs；步骤七，差分滤波图单元对高通红外反射滤波图像gFs与低通红外透射滤波图像dTs进行差分运算，得到差分滤波图像cFTs；步骤八，特征提取第一单元对gTs进行特征提取，计算gTs的平均灰度值gT_Gs作为其特征值；步骤九，特征提取第二单元对dFs进行特征提取，计算dFs的平均灰度值dF_Gs作为其特征值；步骤十，特征提取第三单元对cFTs进行特征提取，计算cFTs的平均灰度值cFT_Gs作为其特征值；步骤十一，将该gT_Gs特征值、dF_Gs特征值及cFT_Gs特征值分别代入褶皱票据与非褶皱票据的三个分类模型y1、y2、y3中，
y1＝f1(gT_G)；
y2＝f2(dF_G)；
y3＝f3(cFT_G)；
得到
p1＝f1(gT_Gs)；
p2＝f2(dF_Gs)；
p3＝f3(cFT_Gs)；
其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，若p1>T1,p2>T2,p3>T3同时成立时，则待鉴别票据被鉴别为褶皱票据，反之，则鉴别为非褶皱票据，其中T1,T2及T3为三个置信度阈值，经验值一般为0.5。
优选的，步骤十一进一步包括对p1,p2及p3给予不同的权重值α,β,γ，其中α+β+γ＝1,α≥0,β≥0,γ≥0，则存在一个阈值Ts，票据分类决策模型如下：
Figure PCTCN2015083861-appb-000002
优选的，得到该褶皱票据与非褶皱票据的三个分类模型y1、y2、y3的方法包括：采集一定数量的褶皱票据与非褶皱票据样本，获得每一张票据样本的高通红外透射滤波图gT的平均灰度值gT_G特征值、低通红外反射滤波图dF的平均灰度值dF_G特征值及差分滤波图cFT的平均灰度值cFT_G特征值，并分别统计该gT_G特征值、dF_G特征值及cFT_G特征值，计算得出褶皱票据对应的gT_G概率分布图、dF_G概率分布图及cFT_G概率分布图，用如下公式分别描述：
y1＝f1(gT_G)；
y2＝f2(dF_G)；
y3＝f3(cFT_G)；
其中y1、y2、y3分别为该褶皱票据与非褶皱票据的三个分类模型。
具体的，获得每一张票据样本的高通红外透射滤波图gT的平均灰度值gT_G特征值、低通红外反射滤波图dF的平均灰度值dF_G特征值及差分滤波图cFT的平均灰度值cFT_G特征值的方法，与获得该待鉴别票据的高通红外透射滤波图gTs的平均灰度值gT_Gs特征值、低通红外反射滤波图dFs的平均灰度值dF_Gs特征值及差分滤波图cFTs的平均灰度值cFT_Gs特征值的方法相同。
优选的，步骤一至步骤十并非依次进行，其中步骤三与步骤四可同步进行，步骤五与步骤六可同步进行，步骤八在步骤三之后即可进行，步骤九在步骤六之后即可进行，以及步骤十在步骤七之后即可进行。
具体的，步骤三之前还包括对图像信号Ts和Fs求取高/低通滤波器阈值：先求取Ts的平均灰度值为：
Figure PCTCN2015083861-appb-000003
其中pix(i)为Ts上像素点对应的灰度值，w为Ts图像信号的宽度，h为Ts图像信号的高度，Ts对应的高通滤波器阈值则为T11＝j*avG,1≤j≤(255/avG)，低通滤波器阈值则为T22＝k*avG,0≤k≤1。
具体的，计算gTs的平均灰度值gT_Gs的求取模型、计算dFs的平均灰度值dF_Gs的求取模型以及计算cFTs的平均灰度值cFT_Gs的求取模型与求取Ts的平均灰度值的求取模型相同。
本发明的有益效果具体为：
由于采用了高/低通滤波的方式对特征进行了有效分类，很大程度上提升了特征的可区分性。特别是不同类型的特征对应不同的分类器，而分类器之间又具有类似Adaboost分类器的功能，能够保证本发明的鉴别置 信度，使得鉴别系统更为鲁棒地兼容外部环境干扰、票据本身污染等复杂情况，整个褶皱票据解决方法及装置能够有效地鉴别褶皱票据。
附图说明
图1是本发明一较佳实施例提供的褶皱票据鉴别装置结构模块示意图；
图2是本发明一较佳实施例提供的褶皱票据鉴别方法流程图。
具体实施方式
为进一步阐述本发明所提供的纸币褶皱识别方法和装置，本实施方式结合图示进行详细说明。
本实施例提供了一种褶皱票据鉴别装置，如图1所示，该褶皱票据鉴别装置包括一票据输入口10、一信号采集模块20、一信号鉴别模块30以及一接收/拒收模块40。
其中该票据输入口10用于接受待鉴别票据或票据样本，并将票据输送至下一模块。该信号采集模块20用于对票据CIS图像进行采集，得到红外透射图像T和红外反射图像F。该信号鉴别模块30用于鉴别待鉴别票据是否有褶皱。该接收/拒收模块40用于对待鉴别票据进行接收或拒收操作。
特别之处在于，该信号鉴别模块30进一步包括：一高通滤波第一单元、一低通滤波第一单元、一高通滤波第二单元、一低通滤波第二单元、一差分滤波图单元、一特征提取第一单元、一特征提取第二单元、一特征提取第三单元以及一鉴别决策单元。其中，该高通滤波第一单元用于对红外透射图像T进行滤波，得到高通红外透射滤波图像gT；该低通滤波第一单元用于对红外透射图像T进行滤波，得到低通红外透射滤波图像dT；该高通滤波第二单元用于根据红外透射图像T的低通滤波情况，按照几何坐标点对点的映射关系对红外反射图F同步进行高通滤波，得到高通红外反射滤波图像gF；该低通滤波第二单元用于根据红外透射图T的高通滤波情况， 按照几何坐标点对点的映射关系对红外反射图F同步进行低通滤波，得到低通红外反射滤波图像dF；该差分滤波图单元用于对高通红外反射滤波图像gF与低通红外透射滤波图像dT进行差分运算，得到差分滤波图像cFT；该特征提取第一单元用于对高通红外透射滤波图像gT进行特征提取，计算gT的平均灰度值gT_G作为其特征值；该特征提取第二单元用于对低通红外反射滤波图像dF进行特征提取，计算dF的平均灰度值dF_G作为其特征值；该特征提取第三单元用于对差分滤波图像cFT进行特征提取，计算cFT的平均灰度值cFT_G作为其特征值；该鉴别决策单元，用于根据票据样本的gT_G特征值、dF_G特征值及cFT_G特征值计算褶皱票据与非褶皱票据的分类模型，并依据票据分类决策模型决策待鉴别钞票是否有褶皱，其中该票据分类决策模型为：p1>T1,p2>T2,p3>T3同时成立时，则待鉴别票据被鉴别为褶皱票据，反之，则鉴别为非褶皱票据，其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，T1,T2及T3为三个置信度阈值。
其中，该票据分类决策模型可进一步修正为：
Figure PCTCN2015083861-appb-000004
其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，α,β,γ是分别对p1,p2及p3给予的不同的权重值，且α+β+γ＝1,α≥0,β≥0,γ≥0，Ts为阈值，Ts的一般经验值为0.5。
以下介绍该褶皱票据鉴别装置执行褶皱票据鉴别方法的具体流程：
如图2所示，该褶皱票据鉴别方法包括步骤1，票据输入口接受待鉴别票据，并将该待鉴别票据输送至信号采集模块；步骤2，信号采集模块采集该待鉴别票据的CIS图像信号，得到红外透射图像Ts和红外反射图像Fs；步骤3，高通滤波第一单元对红外透射图像Ts进行滤波，得到高通红外透射滤波图像gTs；步骤4，低通滤波第一单元对红外透射图像Ts进行滤波，得到低通红外透射滤波图像dTs；步骤5，高通滤波第二单元根据红外透射图像Ts的低通滤波情况，按照几何坐标点对点的映射关系对红外反 射图Fs同步进行高通滤波，得到高通红外反射滤波图像gFs；步骤6，低通滤波第二单元根据红外透射图Ts的高通滤波情况，按照几何坐标点对点的映射关系对红外反射图Fs同步进行低通滤波，得到低通红外反射滤波图像dFs；步骤7，差分滤波图单元对高通红外反射滤波图像gFs与低通红外透射滤波图像dTs进行差分运算，得到差分滤波图像cFTs；步骤8，特征提取第一单元对gTs进行特征提取，计算gTs的平均灰度值gT_Gs作为其特征值；步骤9，特征提取第二单元对dFs进行特征提取，计算dFs的平均灰度值dF_Gs作为其特征值；步骤10，特征提取第三单元对cFTs进行特征提取，计算cFTs的平均灰度值cFT_Gs作为其特征值；步骤11，将该gT_Gs特征值、dF_Gs特征值及cFT_Gs特征值分别代入褶皱票据与非褶皱票据的三个分类模型y1、y2、y3中，
y1＝f1(gT_G)；
y2＝f2(dF_G)；
y3＝f3(cFT_G)；
得到
p1＝f1(gT_Gs)；
p2＝f2(dF_Gs)；
p3＝f3(cFT_Gs)；
其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度；然后依据票据分类决策模型决策待鉴别钞票是否有褶皱，其中该票据分类决策模型为：若p1>T1,p2>T2,p3>T3同时成立时，则待鉴别票据被鉴别为褶皱票据，反之，则鉴别为非褶皱票据，其中T1,T2及T3为三个置信度阈值，流程结束。
优选的，步骤11进一步包括对p1,p2及p3给予不同的权重值α,β,γ，其中α+β+γ＝1,α≥0,β≥0,γ≥0，则存在一个阈值Ts，票据分类决策模型如 下：
Figure PCTCN2015083861-appb-000005
其中步骤1至步骤10并非依次进行，其中步骤3与步骤4可同步进行，步骤5与步骤6可同步进行，步骤8在步骤3之后即可进行，步骤9在步骤6之后即可进行，以及步骤10在步骤7之后即可进行。
另外，得到该褶皱票据与非褶皱票据的三个分类模型y1、y2、y3的方法包括：采集一定数量的褶皱票据与非褶皱票据样本，获得每一张票据样本的高通红外透射滤波图gT的平均灰度值gT_G特征值、低通红外反射滤波图dF的平均灰度值dF_G特征值及差分滤波图cFT的平均灰度值cFT_G特征值，并分别统计该gT_G特征值、dF_G特征值及cFT_G特征值，计算得出褶皱票据对应的gT_G概率分布图、dF_G概率分布图及cFT_G概率分布图，用如下公式分别描述：
y1＝f1(gT_G)；
y2＝f2(dF_G)；
y3＝f3(cFT_G)；
其中y1、y2、y3分别为该褶皱票据与非褶皱票据的三个分类模型。
具体的，获得每一张票据样本的高通红外透射滤波图gT的平均灰度值gT_G特征值、低通红外反射滤波图dF的平均灰度值dF_G特征值及差分滤波图cFT的平均灰度值cFT_G特征值的方法，与获得该待鉴别票据的高通红外透射滤波图gTs的平均灰度值gT_Gs特征值、低通红外反射滤波图dFs的平均灰度值dF_Gs特征值及差分滤波图cFTs的平均灰度值cFT_Gs特征值的方法相同，同步骤1至步骤10。
以下结合票据A介绍褶皱票据鉴别方法：
对应于步骤1的，将褶皱票据A放入到自助设备接受端口；
对应于步骤2的，当票据A通过机械传送经过信号采集模块20时，信号采集模块20将对票据A进行信号采集，其中采集到的CIS红外透射图像信号为rT，红外反射图像信号rF；
在步骤3之前，先对图像信号rT和rF求取高/低通滤波器阈值：
先求取rT的平均灰度值为：
Figure PCTCN2015083861-appb-000006
其中pix(i)为rT上像素点对应的灰度值，w为rT图像信号的宽度，h为rT图像信号的高度。
rT对应的高通滤波器阈值则为T11＝j*avG,1≤j≤(255/avG)，低通滤波器阈值则为T22＝k*avG,0≤k≤1。
对应于步骤3的，采用高通滤波第一单元对rT进行高通滤波，得到高通滤波图GT；
对应于步骤4的，采用低通滤波第一单元对rT进行低通滤波，得到低通滤波图DT；
对应于步骤5，利用几何坐标的映射关系，对rF进行相关高通滤波，得到高通滤波图GF；
对应于步骤6：利用几何坐标的映射关系，对rF进行相关低通滤波，得到低通滤波图DF；
对应于步骤7：对高通滤波图GF与低通滤波图DT进行差分运算，得 到差分滤波图CFT；
对应于步骤10：求取差分滤波图CFT的平均灰度值cAVG作为特征值，求取模型与公式(9)相同；
对应于步骤9：求取低通滤波图DF的平均灰度值dAVG作为特征值，求取模型与公式(9)相同；
对应于步骤8：求取高通滤波图GT的平均灰度值gAVG作为特征值，求取模型与公式(9)相同；
对应于步骤11：对求得cAVG、dAVG、gAVG，输入到多特征融合决策单元中，通过学习到的多特征分类概率分布模型f1(x1),f2(x2),f3(x3)进行分类，若f1(cAVG)>T1,f2(dAVG)>T2,f3(gAVG)>T3，其中T1,T2及T3为经验阈值，一般为0.5，当决策单元输出为真时，票据A则鉴别为褶皱票据，反之，当决策单元输出为假时，票据A则鉴别为非褶皱票据，结束本次鉴别功能。
本实施例提供的褶皱票据鉴别方法和装置由于采用了高/低通滤波的方式对特征进行了有效分类，很大程度上提升了特征的可区分性。特别是不同类型的特征对应不同的分类器，而分类器之间又具有类似Adaboost分类器的功能，能够保证本发明的鉴别置信度，使得鉴别系统更为鲁棒地兼容外部环境干扰、票据本身污染等复杂情况，整个褶皱票据解决方法及装置能够有效地鉴别褶皱票据。
以上仅是本发明的优选实施方式，应当指出的是，上述优选实施方式不应视为对本发明的限制，本发明的保护范围应当以权利要求所限定的范围为准。对于本技术领域的普通技术人员来说，在不脱离本发明的精神和 范围内，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。

Claims (9)

  1. 一种褶皱票据鉴别装置，包括：
    一票据输入口，用于接受待鉴别票据或票据样本，并将票据输送至下一模块；
    一信号采集模块，用于对票据CIS图像进行采集，得到红外透射图像T和红外反射图像F；
    一信号鉴别模块，用于鉴别待鉴别票据是否有褶皱；以及
    一接收/拒收模块，用于对待鉴别票据进行接收或拒收操作；
    其特征在于，该信号鉴别模块进一步包括：
    一高通滤波第一单元，用于对红外透射图像T进行滤波，得到高通红外透射滤波图像gT；
    一低通滤波第一单元，用于对红外透射图像T进行滤波，得到低通红外透射滤波图像dT；
    一高通滤波第二单元，用于根据红外透射图像T的低通滤波情况，按照几何坐标点对点的映射关系对红外反射图F同步进行高通滤波，得到高通红外反射滤波图像gF；
    一低通滤波第二单元，用于根据红外透射图T的高通滤波情况，按照几何坐标点对点的映射关系对红外反射图F同步进行低通滤波，得到低通红外反射滤波图像dF；
    一差分滤波图单元，用于对高通红外反射滤波图像gF与低通红外透射滤波图像dT进行差分运算，得到差分滤波图像cFT；
    一特征提取第一单元，用于对高通红外透射滤波图像gT进行特征提取，计算gT的平均灰度值gT_G作为其特征值；
    一特征提取第二单元，用于对低通红外反射滤波图像dF进行特征提取，计算dF的平均灰度值dF_G作为其特征值；
    一特征提取第三单元，用于对差分滤波图像cFT进行特征提取，计算cFT的平均灰度值cFT_G作为其特征值；以及
    一鉴别决策单元，用于根据票据样本的gT_G特征值、dF_G特征值及cFT_G特征值计算褶皱票据与非褶皱票据的分类模型，并依据票据分类决策模型决策待鉴别钞票是否有褶皱，其中该票据分类决策模型为：p1>T1,p2>T2,p3>T3同时成立时，则待鉴别票据被鉴别为褶皱票据，反之，则鉴别为非褶皱票据，其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，T1,T2及T3为三个置信度阈值。
  2. 如权利要求1所述的褶皱票据鉴别装置，其特征在于该票据分类决策模型进一步修正为：
    Figure PCTCN2015083861-appb-100001
    其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，α,β,γ是分别对p1,p2及p3给予的不同的权重值，且α+β+γ＝1,α≥0,β≥0,γ≥0，Ts为阈值，Ts的一般经验值为0.5。
  3. 一种褶皱票据鉴别方法，包括：
    步骤一，票据输入口接受待鉴别票据，并将该待鉴别票据输送至信号采集模块；
    步骤二，信号采集模块采集该待鉴别票据的CIS图像信号，得到红外透射图像Ts和红外反射图像Fs；
    步骤三，高通滤波第一单元对红外透射图像Ts进行滤波，得到高通红外透射滤波图像gTs；
    步骤四，低通滤波第一单元对红外透射图像Ts进行滤波，得到低通红外透射滤波图像dTs；
    步骤五，高通滤波第二单元根据红外透射图像Ts的低通滤波情况，按照几何坐标点对点的映射关系对红外反射图Fs同步进行高通滤波，得到高通红外反射滤波图像gFs；
    步骤六，低通滤波第二单元根据红外透射图Ts的高通滤波情况，按照几何坐标点对点的映射关系对红外反射图Fs同步进行低通滤波，得到低通红外反射滤波图像dFs；
    步骤七，差分滤波图单元对高通红外反射滤波图像gFs与低通红外透射滤波图像dTs进行差分运算，得到差分滤波图像cFTs；
    步骤八，特征提取第一单元对gTs进行特征提取，计算gTs的平均灰度值gT_Gs作为其特征值；
    步骤九，特征提取第二单元对dFs进行特征提取，计算dFs的平均灰度值dF_Gs作为其特征值；
    步骤十，特征提取第三单元对cFTs进行特征提取，计算cFTs的平均灰度值cFT_Gs作为其特征值；
    步骤十一，将该gT_Gs特征值、dF_Gs特征值及cFT_Gs特征值分别代入褶皱票据与非褶皱票据的三个分类模型y1、y2、y3中，
    y1＝f1(gT_G)；
    y2＝f2(dF_G)；
    y3＝f3(cFT_G)；
    得到
    p1＝f1(gT_Gs)；
    p2＝f2(dF_Gs)；
    p3＝f3(cFT_Gs)；
    其中p1,p2及p3分别为将待鉴别票据决策为褶皱票据的置信度，若p1>T1,p2>T2,p3>T3同时成立时，则待鉴别票据被鉴别为褶皱票据，反之，则鉴别为非褶皱票据，其中T1,T2及T3为三个置信度阈值。
  4. 如权利要求3所述的褶皱票据鉴别方法，其特征在于，步骤十一进一步包括对p1,p2及p3给予不同的权重值α,β,γ，其中
    α+β+γ＝1,α≥0,β≥0,γ≥0，则存在一个阈值Ts，票据分类决策模型如下：
    Figure PCTCN2015083861-appb-100002
  5. 如权利要求3或4所述的褶皱票据鉴别方法。其特征在于，得到该褶皱票据与非褶皱票据的三个分类模型y1、y2、y3的方法包括：采集一定数量的褶皱票据与非褶皱票据样本，获得每一张票据样本的高通红外透射滤波图gT的平均灰度值gT_G特征值、低通红外反射滤波图dF的平均灰度值dF_G特征值及差分滤波图cFT的平均灰度值cFT_G特征值，并分别统计该gT_G特征值、dF_G特征值及cFT_G特征值，计算得出褶皱票据对应的gT_G概率分布图、dF_G概率分布图及cFT_G概率分布图，用如下公式分别描述：
    y1＝f1(gT_G)；
    y2＝f2(dF_G)；
    y3＝f3(cFT_G)；
    其中y1、y2、y3分别为该褶皱票据与非褶皱票据的三个分类模型。
  6. 如权利要求5所述的褶皱票据鉴别方法，其特征在于，获得每一张票据样本的高通红外透射滤波图gT的平均灰度值gT_G特征值、低通红外反射滤波图dF的平均灰度值dF_G特征值及差分滤波图cFT的平均灰度值cFT_G特征值的方法，与权利要求1中步骤一至步骤十中，获得该待鉴别票据的高通红外透射滤波图gTs的平均灰度值gT_Gs特征值、低通红外反射滤波图dFs的平均灰度值dF_Gs特征值及差分滤波图cFTs的平均灰度值cFT_Gs特征值的方法相同。
  7. 如权利要求3所述的褶皱票据鉴别方法，其特征在于，步骤一至步骤十 并非依次进行，其中步骤三与步骤四可同步进行，步骤五与步骤六可同步进行，步骤八在步骤三之后即可进行，步骤九在步骤六之后即可进行，以及步骤十在步骤七之后即可进行。
  8. 如权利要求3所述的褶皱票据鉴别方法，其特征在于，步骤三之前还包括对图像信号Ts和Fs求取高/低通滤波器阈值：
    先求取Ts的平均灰度值为：
    Figure PCTCN2015083861-appb-100003
    其中pix(i)为Ts上像素点对应的灰度值，w为Ts图像信号的宽度，h为Ts图像信号的高度，Ts对应的高通滤波器阈值则为T11＝j*avG,1≤j≤(255/avG)，低通滤波器阈值则为T22＝k*avG,0≤k≤1。
  9. 如权利要求8所述的褶皱票据鉴别方法，其特征在于，计算gTs的平均灰度值gT_Gs的求取模型、计算dFs的平均灰度值dF_Gs的求取模型以及计算cFTs的平均灰度值cFT_Gs的求取模型与求取Ts的平均灰度值的求取模型相同。
PCT/CN2015/083861 2015-02-04 2015-07-13 一种褶皱票据鉴别方法及装置 WO2016123903A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP15880858.4A EP3255617B1 (en) 2015-02-04 2015-07-13 Folded bill identification method and device
US15/544,379 US10319170B2 (en) 2015-02-04 2015-07-13 Folded bill identification method and device
RU2017130019A RU2673120C1 (ru) 2015-02-04 2015-07-13 Способ и устройство для идентификации сложенных банкнот
HK18105825.0A HK1246483A1 (zh) 2015-02-04 2018-05-07 一種褶皺票據鑒別方法及裝置

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510059223.XA CN104573700B (zh) 2015-02-04 2015-02-04 一种褶皱票据鉴别方法及装置
CN201510059223.X 2015-02-04

Publications (1)

Publication Number Publication Date
WO2016123903A1 true WO2016123903A1 (zh) 2016-08-11

Family

ID=53089722

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/083861 WO2016123903A1 (zh) 2015-02-04 2015-07-13 一种褶皱票据鉴别方法及装置

Country Status (6)

Country Link
US (1) US10319170B2 (zh)
EP (1) EP3255617B1 (zh)
CN (1) CN104573700B (zh)
HK (1) HK1246483A1 (zh)
RU (1) RU2673120C1 (zh)
WO (1) WO2016123903A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110717516A (zh) * 2019-09-06 2020-01-21 中国平安财产保险股份有限公司 票据图像分类方法、装置及计算机可读存储介质

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104573700B (zh) * 2015-02-04 2017-12-22 广州广电运通金融电子股份有限公司 一种褶皱票据鉴别方法及装置
CN105046807B (zh) * 2015-07-09 2017-12-26 中山大学 一种基于智能手机的伪钞识别方法及系统
CN105551133B (zh) * 2015-11-16 2018-11-23 新达通科技股份有限公司 一种纸币拼接缝或折痕的识别方法及系统
CN105551134B (zh) * 2015-12-17 2018-06-19 深圳怡化电脑股份有限公司 一种纸币褶皱的识别的方法及系统
JP6615014B2 (ja) * 2016-03-15 2019-12-04 グローリー株式会社 紙葉類識別装置および紙葉類識別方法
CN107356834A (zh) * 2017-08-21 2017-11-17 赵崇征 电子配件鉴别方法及装置
CN112825142A (zh) * 2019-11-20 2021-05-21 深圳怡化电脑股份有限公司 一种票据检测方法、装置、终端和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010117803A (ja) * 2008-11-12 2010-05-27 Hitachi Omron Terminal Solutions Corp 紙葉類処理装置及び紙葉類処理プログラム
CN102236897A (zh) * 2011-05-09 2011-11-09 哈尔滨工业大学 基于纸币退化能量函数的清分机纸币新旧和残缺检测方法
CN102881074A (zh) * 2012-09-12 2013-01-16 上海古鳌电子科技股份有限公司 一种纸币清分机
CN203812312U (zh) * 2014-05-08 2014-09-03 中国工商银行股份有限公司 一种纸钞处理设备
CN104573700A (zh) * 2015-02-04 2015-04-29 广州广电运通金融电子股份有限公司 一种褶皱票据鉴别方法及装置

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4984280A (en) * 1988-06-08 1991-01-08 Laurel Bank Machines Co., Ltd. Bill discriminating apparatus
US6040584A (en) * 1998-05-22 2000-03-21 Mti Corporation Method and for system for detecting damaged bills
MXPA01011472A (es) * 1999-05-11 2002-06-04 Diebold Inc Detector de hoja dobhle para maquina de transaccion automatica.
DE10210687A1 (de) * 2002-03-12 2003-10-09 Giesecke & Devrient Gmbh Vorrichtung zur Bearbeitung von Banknoten
AU2006202180A1 (en) * 2005-06-17 2007-01-11 Aruze Corp. Bill handling device
CN102236926A (zh) * 2006-11-23 2011-11-09 锡克拜控股有限公司 使用指纹识别设备进行片状物品的验证
JP4966117B2 (ja) * 2007-07-09 2012-07-04 日立オムロンターミナルソリューションズ株式会社 紙葉類集積装置
JP4673393B2 (ja) * 2008-06-05 2011-04-20 日立オムロンターミナルソリューションズ株式会社 紙葉類取扱装置及び方法
WO2010151305A2 (en) * 2009-06-26 2010-12-29 Venkatesh Rajagopalan Chari Compact assistive reading device
DE102010021803A1 (de) * 2010-05-27 2011-12-01 Giesecke & Devrient Gmbh Vorrichtung zur Echtheitsprüfung von Wertdokumenten
KR101883425B1 (ko) 2011-08-01 2018-07-31 삼성전자주식회사 휴대 단말기를 이용하는 위폐 감별법

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010117803A (ja) * 2008-11-12 2010-05-27 Hitachi Omron Terminal Solutions Corp 紙葉類処理装置及び紙葉類処理プログラム
CN102236897A (zh) * 2011-05-09 2011-11-09 哈尔滨工业大学 基于纸币退化能量函数的清分机纸币新旧和残缺检测方法
CN102881074A (zh) * 2012-09-12 2013-01-16 上海古鳌电子科技股份有限公司 一种纸币清分机
CN203812312U (zh) * 2014-05-08 2014-09-03 中国工商银行股份有限公司 一种纸钞处理设备
CN104573700A (zh) * 2015-02-04 2015-04-29 广州广电运通金融电子股份有限公司 一种褶皱票据鉴别方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3255617A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110717516A (zh) * 2019-09-06 2020-01-21 中国平安财产保险股份有限公司 票据图像分类方法、装置及计算机可读存储介质
CN110717516B (zh) * 2019-09-06 2023-08-25 中国平安财产保险股份有限公司 票据图像分类方法、装置及计算机可读存储介质

Also Published As

Publication number Publication date
US20180268634A1 (en) 2018-09-20
EP3255617A1 (en) 2017-12-13
EP3255617B1 (en) 2020-06-17
HK1246483A1 (zh) 2018-09-07
CN104573700B (zh) 2017-12-22
US10319170B2 (en) 2019-06-11
EP3255617A4 (en) 2018-03-28
CN104573700A (zh) 2015-04-29
RU2673120C1 (ru) 2018-11-22

Similar Documents

Publication Publication Date Title
WO2016123903A1 (zh) 一种褶皱票据鉴别方法及装置
US9633269B2 (en) Image-based liveness detection for ultrasonic fingerprints
WO2017197884A1 (zh) 一种纸币管理方法及其系统
CN107085883B (zh) 一种纸币识别的方法和装置
CN104464079B (zh) 基于模板特征点及其拓扑结构的多币种面值识别方法
CN104298989B (zh) 基于斑马线红外图像特征的鉴伪方法及其系统
CN106709450A (zh) 一种指纹图像识别方法及系统
CN105956572A (zh) 一种基于卷积神经网络的活体人脸检测方法
CN102324134A (zh) 有价文件鉴别方法及其鉴别装置
WO2012016484A1 (zh) 有价文件识别方法及其识别系统、装置
CN102005078A (zh) 一种纸币、票券识别方法和装置
WO2016078455A1 (zh) 纸币褶皱的识别装置和方法
CN108009472A (zh) 一种基于卷积神经网络和贝叶斯分类器的指背关节纹识别方法
CN104969268A (zh) 安全文件鉴别及执行鉴别的移动装置
CN106056597A (zh) 物体视觉检测方法及装置
CN103886309A (zh) 人像识别鉴别美元面额的方法
CN106991448A (zh) 一种人像比对处理方法
CN102306415A (zh) 便携式有价文件识别装置
WO2007105890A1 (en) Recognizing the denomination of a note using wavelet transform
Hendryli et al. Automatic license plate recognition for parking system using convolutional neural networks
CN104680142B (zh) 一种基于特征点集分割和rst不变特征的四联指比对方法
CN106683261B (zh) 一种纸币的鉴伪方法和装置
WO2018107574A1 (zh) 一种对印防伪特征的检测方法及装置
CN107134048A (zh) 一种智能识别水印特征的票据防伪鉴别方法
WO2016011640A1 (zh) 基于手部图像纹理的身份识别方法

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: 15880858

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15544379

Country of ref document: US

REEP Request for entry into the european phase

Ref document number: 2015880858

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2017130019

Country of ref document: RU

Kind code of ref document: A