CN104899965A - Multi-national paper money serial number identification method based on sorting machine - Google Patents
Multi-national paper money serial number identification method based on sorting machine Download PDFInfo
- Publication number
- CN104899965A CN104899965A CN201510251098.2A CN201510251098A CN104899965A CN 104899965 A CN104899965 A CN 104899965A CN 201510251098 A CN201510251098 A CN 201510251098A CN 104899965 A CN104899965 A CN 104899965A
- Authority
- CN
- China
- Prior art keywords
- character
- binaryzation
- sequence number
- matrix
- subtemplate
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000010606 normalization Methods 0.000 claims abstract description 24
- 239000011159 matrix material Substances 0.000 claims description 59
- 238000005070 sampling Methods 0.000 claims description 11
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 4
- 230000008878 coupling Effects 0.000 claims description 3
- 238000010168 coupling process Methods 0.000 claims description 3
- 238000005859 coupling reaction Methods 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 abstract 1
- 238000012216 screening Methods 0.000 abstract 1
- 239000000047 product Substances 0.000 description 14
- 230000011218 segmentation Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000001154 acute effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 239000007795 chemical reaction product Substances 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004900 laundering Methods 0.000 description 1
Landscapes
- Character Input (AREA)
Abstract
The embodiment of the invention discloses a multi-national paper money serial number identification method based on a sorting machine, and the method employs a character matching template to recognize a to-be-recognized character. The method comprises the steps: firstly enabling an obtained paper money serial number image to be converted into a gray scale image; secondly determining a character recognition region through initial location, binaryzation and relocation; thirdly enabling the size of a character region to be consistent with the size of a matching template through normalization; and finally screening out matched characters through dot product, and completing the recognition. The method not only avoids the influence from noisy points, and can save a filtering step. The method is suitable for a plurality of current mainstream currency types: RMB, dollar, Euro, Hongkong dollar, and yen.
Description
Technical field
The invention belongs to automatic identification technology field, particularly a kind of multinational paper money sequence number recognition methods based on cleaning-sorting machine, relate to the secondary splitting method combining bank note priori, and based on the character identifying method of template matches.
Background technology
In financial field, often opening bank note has a unique sequence number, is equivalent to the I.D. of bank note, and therefore sequence number becomes the important mark of bank note.Be no matter classification or false proof, sequence number all plays vital effect.And the caseload of some economic disputes at present and illegal Capital Flow increases day by day, also have in addition and hit the robbery of violence bank note and money laundering aspect, sequence number especially state apparatus, to the sharp weapon of banknote drifting management, is conducive to social stability.Branch and the subbranch of current domestic Ji Jia large bank are equipped with middle-size and small-size cleaning-sorting machine mostly, and add that other bank outlets and ATM are equipped with, the demand of cleaning-sorting machine just reached 100,000 in 2010, and in sustainable growth.In addition, economic globalization is development trend, and existing cleaning-sorting machine is also growing for the identification function demand of foreign currency, therefore needs to identify the sequence number of various countries' bank note, existing cleaning-sorting machine mainly achieves the sequence number identification to single Currency Type, limits range of application.
Current paper money sequence number identification mainly contains three steps, Image semantic classification, Character segmentation and character recognition.In the prior art, identify main only for 2005 new edition Renminbi, usable range is narrow, and new edition Renminbi background is single, and sequence number identification difficulty is relatively low.Employ secondary splitting method in Character segmentation step in prior art, dividing method is calculate based on trip point for the first time, is easy to be printed and pollution effect, and very responsive for the shading value of image; Second time is divided into the binaryzation of fixed threshold, and record edge trip point is segmentation result, responsive to light and shade, cannot process the bank note of background complexity, such as dollar.Use most neighbor interpolation algorithm at character normalization, each pixel all needs to recalculate gray scale, and the character comparatively fixed for size this method reduces efficiency.In identification, current method has two kinds, method one is gridding method, character is divided into several grid (number of grid is 20-40), each grid has 20-30 pixel, namely each character will process and at least will travel through statistics 500 pixels, although precision is higher, greatly sacrifices arithmetic speed; Method two is more excellent, it is the sorting technique supporting sequential machine based on SVM, its method sets up 0-9 and 26 digital 36 support sequential machines, error rate is lower, at about 2-3%, but processing speed is average slowly opens at 50-70ms/, due to note input, rotate correct, picked-up serial number image itself needs time of 40ms, the sequence number recognition speed of the A level cleaning-sorting machine required in " the GB 16999-2010 RMB-banknote discriminating device general technical specifications " that cannot reach in the industry requires >=700 (/min), the standard of average handling time 80ms.In addition, complicated for dollar bill background, the problems such as Euro number difference in size is very big, prior art does not all solve.Therefore for the above-mentioned defect existed in currently available technology, be necessary to study in fact, to provide a kind of scheme, solve the defect existed in prior art, avoid causing paper money sequence number identification Currency Type single, recognition speed is low, the problem that accuracy rate is not high.
Summary of the invention
For solving the problem, the object of the present invention is to provide a kind of multinational paper money sequence number recognition methods based on cleaning-sorting machine, adopt the secondary splitting method combining bank note priori, and based on the character identifying method of template matches, solve in existing paper money sequence number identification and can not be adapted to complex background and character boundary variation issue, recognition speed improves more than 8 times simultaneously.
For achieving the above object, technical scheme of the present invention is: a kind of multinational paper money sequence number recognition methods based on cleaning-sorting machine, and the method is: the image obtaining paper money sequence number, and image is converted to gray-scale map I, then carries out steps of processing to gray-scale map I:
Step 1: the bank note prior imformation obtaining different Currency Type, face amount, version information, described prior imformation is the position distribution of each character in sequence number; Utilize prior imformation to carry out character to gray-scale map just to locate, according to first positioning result, binary conversion treatment is carried out to gray-scale map, relocate after the image projection after binaryzation; According to relocating result, to the Image Segmentation Using after binaryzation, what obtain each character relocates binaryzation matrix.
Step 2, carries out character normalization to the binaryzation matrix that relocates that step 1 obtains, obtains normalization binaryzation Matrix C
n, C
nsize be A × B, identical with the size of each subtemplate in matching template set M; Described character normalization method is arest neighbors method of interpolation or bilinear interpolation method, and the Currency Type fixing for character shape adopts arest neighbors method of interpolation; Currency Type for character shape change adopts bilinear interpolation method;
Step 3: by C
neach mates subtemplate and carries out dot product respectively with matching template set M, and obtains each element summation of product matrix to dot product, obtains element summation r, if C
nwith character k mate subtemplate M
kduring dot product, element summation r obtains maximal value, namely
a=1,2 ..., A, b=1,2 ..., B; Then k is recognition result.
Further, the character described in step 1 is just located, and specifically comprises the following steps:
(1.1) the reference position m of the vertical direction of paper money sequence number character is determined
*, m
*=argmax
m(yS);
Wherein,
for gray-scale map I is at the projection vector of vertical direction,
represent the real number matrix of the capable q row of p; S=[S
1, S
2..., S
p]
t,
i=1,2 ..., p, S
ifor i-th element in projection vector S; Y is the projection signal of input picture at vertical direction, y=[y
1, y
2..., y
p]; y
i=1+s (i-m-h)-s (i-m), i=1,2 ..., p, s (i) are step signals,
H is the height of the highest character in sequence number, by the prior imformation determination value of bank note, is known terms; M ∈ [0, p-h] is the reference position variable of sequence number at vertical direction.
(1.2) the reference position l of the horizontal direction of paper money sequence number character is determined
*, l
*=arg max
l(xG);
Wherein,
for input picture projection vector in the horizontal direction, G=[G
1, G
2..., G
q]
t,
j=1,2 ..., q, G
jfor a jth element in projection vector G; X is input picture projection signal in the horizontal direction; X=[x
1, x
2..., x
q]; x
j=1+s (j-l-d
1-w
1)-s (j-l-d
1)+s (j-l-d
2-w
2)-s (j-l-d
2)+... + s (j-l-d
k-w
k)-s (j-l-d
k), j=1,2 ..., q, k are the numbers of sequence number character, w
1, w
2..., w
kbe respectively the 1st, 2 ..., the width of k character.With the lower left corner of first character in sequence number for initial point, d
1, d
2..., d
kbe on sequence number position, each character lower left corner relative to the distance of this initial point, d
1=0.K, w
1, w
2..., w
kand d
1, d
2..., d
kby the prior imformation determination value of bank note.L ∈ [0, q-w] is sequence number reference position variable in the horizontal direction.
(1.3) according to the first position location (l of first character
*, m
*), d
1, d
2..., d
kand w
1, w
2..., w
kobtain the first positioning result of other character on sequence number, be specially:
If the first position location of the n-th character is: horizontal direction (x
n1, x
n2), vertical direction (y
n1, y
n2).Wherein x
n1the reference position of this character horizontal direction, x
n1=l*+d
n; x
n2the final position of this character horizontal direction, x
n2=x
n1+ w
n; y
n1the reference position of this character vertical direction, y
n1=m*; y
n2the final position of this character vertical direction, y
n2=m
*+ h.
Further, the binary conversion treatment described in step 1, specifically comprises the following steps:
(1) determine binaryzation region, described binaryzation region is the horizontal reference position x of first character
11to the horizontal end position x of a kth character
k2, the vertical reference position y of first character
11to the vertical final position y of first character
12, be namely (x between binaryzation horizontal zone
11, x
k2), binaryzation vertically interval is (y
11, y
12);
(2) binaryzation is carried out to the character in binaryzation region, obtain the first location binaryzation matrix of each character, if the first location binaryzation matrix of the n-th character is N
n, its size is (y
n2-y
n1) × (x
n2-x
n1).The threshold value that binaryzation uses is calculated by Two-peak method.
Further, the character described in step 1 relocates, and specifically comprises the following steps:
(1) according to the first location binaryzation matrix N of character
ncarry out vertical direction projection to each character, obtain the discrete series of vertical direction, the vertical direction discrete series as obtained the n-th character is W
n.To W
ntravel through, meet continuous three elements be all greater than setting threshold value t
n1or an element is greater than setting threshold value t
n2first element position, as the starting position coordinates vertically relocated of character, be designated as y
n1', the position meeting last element of one of above two conditions, as the final position coordinate vertically relocated, is designated as y
n2'; Wherein t
n1=0.5w
n, t
n2=w
n-3.According to (y
n1', y
n2') upgrade the binaryzation matrix of each character, obtain new binaryzation matrix N
n', size is (y
n2'-y
n1') × (x
n2-x
n1).
(2) according to new binaryzation matrix N
n' horizontal direction projection is carried out to each character, obtain the discrete series of horizontal direction, the horizontal direction discrete series as obtained the n-th character is L
n.To L
ntravel through, meet continuous three elements be all greater than setting threshold value t
n1' or element be greater than setting threshold value t
n2' first element position, as the starting position coordinates vertically relocated of character, be designated as x
n1', the final position coordinate that the position meeting last element of one of above two conditions relocates as level, is designated as x
n2'; Wherein t
n1'=0.5 (y
n2'-y
n1'), t
n2'=(y
n2'-y
n1')-3.According to (x
n1', x
n2') upgrade the binaryzation matrix of each character, obtain relocating binaryzation matrix N
n", size is (y
n2'-y
n1') × (x
n2'-x
n1').
Further, the set of matching template described in step 3 M obtains in the following manner:
(1) sampling obtains U the normalization binaryzation matrix D of certain character k
k1, D
k2..., D
kU, this U normalization binaryzation matrix sampling is from the diverse location of different paper money sequence numbers or identical paper money sequence number.
Counting statistics matrix
u=1,2 ..., U, A
kin value minimum be 0, be U to the maximum.Preliminary subtemplate M is set up based on this
k', be specially: to A
kthe point of (i ', j ')>=0.8U (i ', j '), i.e. dense parts, M
k' (i ', j ') get 10; To A
kthe point of (i ', j ')≤0.2U (i ', j '), i.e. sparse part, M
k' (i ', j ') get-10; To 0.2U < A
kthe point of (i ', j ') < 0.8U (i ', j '), M
k' (i ', j ') get 0;
Repeat above-mentioned steps, corresponding preliminary subtemplate is set up to different characters, obtains preliminary matching template set M'.
(2) preliminary matching template set M' is mated with the normalization binaryzation matrix of sampling in step (1).According to the set of matching result adjustment matching template, be specially:
The normalization binaryzation matrix D of the sampling of character k
kucarry out dot product with each subtemplate in preliminary matching template set M', and each element summation of product matrix obtained to dot product, obtain element summation r,
a=1,2 ..., A, b=1,2 ..., B; K ' is arbitrary character; If during k '=k, M
k' be the coupling subtemplate of character k, M
k=M
k'; If k ≠ k ', then need to adjust preliminary subtemplate M
k', method is: for preliminary subtemplate M
k' and M
k '', for M
k' (i ", j ")=M
k '' (i ", and the point of j ") (i ", j "), then M
k(i ", j ")=M
k' (i ", j ")-1; For M
k(i ", j ") ≠ M
k '(i ", and the point of j ") (i ", j "), then M
k(i ", j ")=M
k' (i ", j ")+1; Thus obtain matching template M
k.
Repeat above-mentioned steps, corresponding preliminary subtemplate is adjusted to different characters, the matching template set after being adjusted.
The invention has the beneficial effects as follows:
(1) utilize the secondary splitting of the prior imformation of number of paper money relative position distribution to make segmentation result more accurate, and avoid the impact of noise, can filter step be omitted simultaneously;
(2) in identifying, set up template reduce 150-180 pixel, identify consuming time be before method 1/10, in addition accuracy comparatively support vector machine be significantly improved, method is flexible, the multiple Currency Type being applicable to current main flow comprise Renminbi, dollar, Euro, Hongkong dollar, yen etc.
(3) compared with the simple secondary splitting adopted with prior art and connected domain, support vector machine recognition methods, the present invention using pre-service as optional process, segmentation result is little by the change of background light and shade, properer character, can process the character that vertical setting of types, size and character ratio have acute variation; In feature extraction, reach character Euclidean distance maximum at present, greatly reduce template size, reduce computational complexity.
Accompanying drawing explanation
Fig. 1 is the flow chart of steps of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention;
Fig. 2 is the RMB sequence number code image be partitioned into of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention;
Fig. 3 is the dollar serial number image be partitioned into of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention;
Fig. 4 is Euro serial number image be partitioned into of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention;
Fig. 5 is the Hongkong dollar serial number image be partitioned into of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention;
Fig. 6 is the segmentation result schematic diagram of Hongkong dollar series number image after secondary splitting in Fig. 5;
Fig. 7 is the sparse part of the character of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention, and its sparse part is respectively upper, middle and lower;
Fig. 8 be the character " 3 " of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention relocate binaryzation Matrix C
3schematic diagram;
Fig. 9 is the schematic diagram of character " 3 " matching template of the multinational paper money sequence number recognition methods foundation based on cleaning-sorting machine of the embodiment of the present invention.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
On the contrary, the present invention is contained any by the substituting of making on marrow of the present invention and scope of defining of claim, amendment, equivalent method and scheme.Further, in order to make the public have a better understanding to the present invention, in hereafter details of the present invention being described, detailedly describe some specific detail sections.Do not have the description of these detail sections can understand the present invention completely for a person skilled in the art yet.
With reference to figure 1, be depicted as the flow chart of steps of the multinational paper money sequence number recognition methods based on cleaning-sorting machine of the embodiment of the present invention.Obtain the image of paper money sequence number, and image be converted to gray-scale map I, as shown in Fig. 2 to 5, be respectively Renminbi, dollar, Euro, the sequence number gray-scale map image of Hongkong dollar.Then steps of processing is carried out to I:
Step 1: the bank note prior imformation obtaining different Currency Type, face amount, version information, described prior imformation is the position distribution of each character in sequence number; Utilize prior imformation to carry out character to gray-scale map just to locate, according to first positioning result, binary conversion treatment is carried out to gray-scale map, relocate after the image projection after binaryzation; According to relocating result, to the Image Segmentation Using after binaryzation, what obtain each character relocates binaryzation matrix.What be illustrated in figure 8 character " 3 " relocates binaryzation Matrix C
3.
Step 2, carries out character normalization to the binaryzation matrix that relocates that step 1 obtains, obtains normalization binaryzation Matrix C
n, C
nthe size of size to be a × b, a × b be each subtemplate in matching template set M; Normalized object is: by the size adjustment one-tenth of gray-scale map and the in the same size of matching template that identify.Described character normalization method is arest neighbors method of interpolation or bilinear interpolation method, and the Currency Type fixing for character shape adopts arest neighbors method of interpolation; Currency Type for character shape change adopts bilinear interpolation method; The eigenmatrix C that normalization obtains
nlength and width all at 12-18;
Step 3: by C
ncarry out dot product with each subtemplate in matching template set M, and each element summation of product matrix is obtained to dot product, obtain element summation r, if C
nwith character k mate subtemplate M
kduring dot product, element summation r obtains maximal value, namely
A=1,2 ..., A, b=1,2 ..., B; Then k is recognition result.
In embody rule example, I can be 8 bit depth gray-scale maps, and size is best in 64*256 to 64*320 scope.
Character described in step 1 is just located, and specifically comprises the following steps:
(1) the reference position m of the vertical direction of paper money sequence number character is determined
*, m
*=arg max
m(yS);
Wherein,
for gray-scale map I is at the projection vector of vertical direction,
represent the real number matrix of the capable q row of p; S=[S
1, S
2..., S
p]
t,
i=1,2 ..., p, S
ifor i-th element in projection vector S;
In other application example, in order to ensure that data are not overflowed and save operation time, using each element of S all divided by 256 as end product, as shown in the formula,
Y is the projection signal of input picture at vertical direction, y=[y
1, y
2..., y
p]; y
i=1+s (i-m-h)-s (i-m), i=1,2 ..., p, s (i) are step signals,
H is the height of the highest character in sequence number, by the prior imformation determination value of bank note, is known terms; M ∈ [0, p-h] is the reference position variable of sequence number at vertical direction.
(2) the reference position l of the horizontal direction of paper money sequence number character is determined
*, l
*=arg max
l(xG);
Wherein,
for input picture projection vector in the horizontal direction, G=[G
1, G
2..., G
q]
t,
j=1,2 ..., q, G
jfor a jth element in projection vector G; X is input picture projection signal in the horizontal direction; X=[x
1, x
2..., x
q]; x
j=1+s (j-l-d
1-w
1)-s (j-l-d
1)+s (j-l-d
2-w
2)-s (j-l-d
2)+... + s (j-l-d
k-w
k)-s (j-l-d
k), j=1,2 ..., q, k are the numbers of sequence number character, w
1, w
2..., w
kbe respectively the 1st, 2 ..., the width of k character.With the lower left corner of first character in sequence number for initial point, d
1, d
2..., d
kbe on sequence number position, each character lower left corner relative to the distance of this initial point, d
1=0.K, w
1, w
2..., w
kand d
1, d
2..., d
kby the prior imformation determination value of bank note.L ∈ [0, q-w] is sequence number reference position variable in the horizontal direction.
(3) according to the first position location (l of first character
*, m
*), d
1, d
2..., d
kand w
1, w
2..., w
kobtain the first positioning result of other character on sequence number, be specially:
If the first position location of the n-th character is: horizontal direction (x
n1, x
n2), vertical direction (y
n1, y
n2).Wherein x
n1the reference position of this character horizontal direction, x
n1=l*+d
n; x
n2the final position of this character horizontal direction, x
n2=x
n1+ w
n; y
n1the reference position of this character vertical direction, y
n1=m*; y
n2the final position of this character vertical direction, y
n2=m
*+ h.
Binary conversion treatment described in step 1, specifically comprises the following steps:
(1) determine binaryzation region, described binaryzation region is the horizontal reference position x of first character
11to the horizontal end position x of a kth character
k2, the vertical reference position y of first character
11to the vertical final position y of first character
12, be namely (x between binaryzation horizontal zone
11, x
k2), binaryzation vertically interval is (y
11, y
12);
(2) binaryzation is carried out to the character in binaryzation region, obtain the first location binaryzation matrix of each character, if the first location binaryzation matrix of the n-th character is N
n, its size is (y
n2-y
n1) × (x
n2-x
n1).The threshold value that binaryzation uses is calculated by Two-peak method.
Character described in step 1 relocates, and specifically comprises the following steps:
(1) according to the first location binaryzation matrix N of character
ncarry out vertical direction projection to each character, obtain the discrete series of vertical direction, the vertical direction discrete series as obtained the n-th character is W
n.To W
ntravel through, meet continuous three elements be all greater than setting threshold value t
n1or an element is greater than setting threshold value t
n2first element position, as the starting position coordinates vertically relocated of character, be designated as y
n1', the position meeting last element of one of above two conditions, as the final position coordinate vertically relocated, is designated as y
n2'; Wherein t
n1=0.5w
n, t
n2=w
n-3.According to (y
n1', y
n2') upgrade the binaryzation matrix of each character, obtain new binaryzation matrix N
n', size is (y
n2'-y
n1') × (x
n2-x
n1).
(2) according to new binaryzation matrix N
n' horizontal direction projection is carried out to each character, obtain the discrete series of horizontal direction, the horizontal direction discrete series as obtained the n-th character is L
n.To L
ntravel through, meet continuous three elements be all greater than setting threshold value t
n1' or element be greater than setting threshold value t
n2' first element position, as the starting position coordinates vertically relocated of character, be designated as x
n1', the final position coordinate that the position meeting last element of one of above two conditions relocates as level, is designated as x
n2'; Wherein t
n1'=0.5 (y
n2'-y
n1'), t
n2'=(y
n2'-y
n1')-3.According to (x
n1', x
n2') upgrade the binaryzation matrix of each character, obtain relocating binaryzation matrix N
n", size is (y
n2'-y
n1') × (x
n2'-x
n1').Be illustrated in figure 6 relocating of Hongkong dollar, i.e. secondary splitting result figure.
The set of matching template described in step 3 M obtains in the following manner:
(1) sampling obtains U the normalization binaryzation matrix D of certain character k
k1, D
k2..., D
kU, the diverse location that this U normalization binaryzation matrix can be sampled from different paper money sequence numbers or identical paper money sequence number.
Counting statistics matrix
u=1,2 ..., U, A
kin value minimum be 0, be U to the maximum.Preliminary subtemplate M is set up based on this
k, be specially: to A
kthe point of (i ', j ')>=0.8U (i ', j '), i.e. dense parts, M
k(i ', j ') get 10; To A
kthe point of (i ', j ')≤0.2U (i ', j '), i.e. sparse part, as shown in Figure 7, M
k(i ', j ') get-10; To 0.2U < A
kthe point of (i ', j ') < 0.8U (i ', j '), M
k(i ', j ') get 0;
Repeat above-mentioned steps, corresponding preliminary subtemplate is set up to different characters, obtains preliminary matching template set M.Preliminary matching template set M contains 36 subtemplates, is respectively 0-9 ten numerals and A-Z 26 capitalizations.
(2) preliminary matching template set M is mated with the normalization binaryzation matrix of sampling in (1).Matching template set M is adjusted according to matching result.Be specially:
The normalization binaryzation matrix D of the sampling of character k
kucarry out dot product with each subtemplate in preliminary matching template set M, and each element summation of product matrix obtained to dot product, obtain element summation r,
a=1,2 ..., A, b=1,2 ..., B; K ' is arbitrary character; If during k '=k, M
k' be the coupling subtemplate of character k, M
k=M
k'; If k ≠ k ', then need to adjust preliminary subtemplate M
k', method is: for preliminary subtemplate M
k' and M
k '', for M
k' (i ", j ")=M
k '' (i ", and the point of j ") (i ", j "), then M
k(i ", j ")=M
k' (i ", j ")-1; For M
k(i ", j ") ≠ M
k '(i ", and the point of j ") (i ", j "), then M
k(i ", j ")=M
k' (i ", j ")+1; Thus obtain matching template M
k.
Repeat above-mentioned steps, corresponding preliminary subtemplate is adjusted to different characters, the matching template set after being adjusted.
Be illustrated in figure 9 character " 3 " adjust after final template.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any amendments done within the spirit and principles in the present invention, equivalent replacement and improvement etc., all should be included within protection scope of the present invention.
Claims (5)
1. based on a multinational paper money sequence number recognition methods for cleaning-sorting machine, it is characterized in that, the method is: the image obtaining paper money sequence number, and image is converted to gray-scale map I, then carries out steps of processing to gray-scale map I:
Step 1: the bank note prior imformation obtaining different Currency Type, face amount, version information, described prior imformation is the position distribution of each character in sequence number; Utilize prior imformation to carry out character to gray-scale map just to locate, according to first positioning result, binary conversion treatment is carried out to gray-scale map, relocate after the image projection after binaryzation; According to relocating result, to the Image Segmentation Using after binaryzation, what obtain each character relocates binaryzation matrix.
Step 2, carries out character normalization to the binaryzation matrix that relocates that step 1 obtains, obtains normalization binaryzation Matrix C
n, C
nsize be A × B, identical with the size of each subtemplate in matching template set M; Described character normalization method is arest neighbors method of interpolation or bilinear interpolation method, and the Currency Type fixing for character shape adopts arest neighbors method of interpolation; Currency Type for character shape change adopts bilinear interpolation method;
Step 3: by C
neach mates subtemplate and carries out dot product respectively with matching template set M, and obtains each element summation of product matrix to dot product, obtains element summation r, if C
nwith character k mate subtemplate M
kduring dot product, element summation r obtains maximal value, namely
a=1,2 ..., A, b=1,2 ..., B; Then k is recognition result.
2. method according to claim 1, is characterized in that, the character described in step 1 is just located, and specifically comprises the following steps:
(1) the reference position m of the vertical direction of paper money sequence number character is determined
*, m
*=argmax
m(yS); Wherein,
for gray-scale map I is at the projection vector of vertical direction,
represent the real number matrix of the capable q row of p; S=[S
1, S
2..., S
p]
t,
i=1,2 ..., p, S
ifor i-th element in projection vector S; Y is the projection signal of input picture at vertical direction, y=[y
1, y
2..., y
p];
Y
i=1+s (i-m-h)-s (i-m), i=1,2 ..., p, s (i) are step signals,
H is the height of the highest character in sequence number, by the prior imformation determination value of bank note, is known terms; M ∈ [0, p-h] is the reference position variable of sequence number at vertical direction.
(2) the reference position l of the horizontal direction of paper money sequence number character is determined
*, l
*=argmax
l(xG);
Wherein,
for input picture projection vector in the horizontal direction, G=[G
1, G
2..., G
q]
t,
j=1,2 ..., q, G
jfor a jth element in projection vector G; X is input picture projection signal in the horizontal direction; X=[x
1, x
2..., x
q]; x
j=1+s (j-l-d
1-w
1)-s (j-l-d
1)+s (j-l-d
2-w
2)-s (j-l-d
2)+... + s (j-l-d
k-w
k)-s (j-l-d
k), j=1,2 ..., q, k are the numbers of sequence number character, w
1, w
2..., w
kbe respectively the 1st, 2 ..., the width of k character.With the lower left corner of first character in sequence number for initial point, d
1, d
2..., d
kbe on sequence number position, each character lower left corner relative to the distance of this initial point, d
1=0.K, w
1, w
2..., w
kand d
1, d
2..., d
kby the prior imformation determination value of bank note.L ∈ [0, q-w] is sequence number reference position variable in the horizontal direction.
(3) according to the first position location (l of first character
*, m
*), d
1, d
2..., d
kand w
1, w
2..., w
kobtain the first positioning result of other character on sequence number, be specially:
If the first position location of the n-th character is: horizontal direction (x
n1, x
n2), vertical direction (y
n1, y
n2).Wherein x
n1the reference position of this character horizontal direction, x
n1=l*+d
n; x
n2the final position of this character horizontal direction, x
n2=x
n1+ w
n; y
n1the reference position of this character vertical direction, y
n1=m*; y
n2the final position of this character vertical direction, y
n2=m
*+ h.
3. method according to claim 1, is characterized in that, the binary conversion treatment described in step 1, specifically comprises the following steps:
(1) determine binaryzation region, described binaryzation region is the horizontal reference position x of first character
11to the horizontal end position x of a kth character
k2, the vertical reference position y of first character
11to the vertical final position y of first character
12, be namely (x between binaryzation horizontal zone
11, x
k2), binaryzation vertically interval is (y
11, y
12);
(2) binaryzation is carried out to the character in binaryzation region, obtain the first location binaryzation matrix of each character, if the first location binaryzation matrix of the n-th character is N
n, its size is (y
n2-y
n1) × (x
n2-x
n1).The threshold value that binaryzation uses is calculated by Two-peak method.
4. for method according to claim 1, it is characterized in that, the character described in step 1 relocates, and specifically comprises the following steps:
(1) according to the first location binaryzation matrix N of character
ncarry out vertical direction projection to each character, obtain the discrete series of vertical direction, the vertical direction discrete series as obtained the n-th character is W
n.To W
ntravel through, meet continuous three elements be all greater than setting threshold value t
n1or an element is greater than setting threshold value t
n2first element position, as the starting position coordinates vertically relocated of character, be designated as y
n1', the position meeting last element of one of above two conditions, as the final position coordinate vertically relocated, is designated as y
n2'; Wherein t
n1=0.5w
n, t
n2=w
n-3.According to (y
n1', y
n2') upgrade the binaryzation matrix of each character, obtain new binaryzation matrix N
n', size is (y
n2'-y
n1') × (x
n2-x
n1).
(2) according to new binaryzation matrix N
n' horizontal direction projection is carried out to each character, obtain the discrete series of horizontal direction, the horizontal direction discrete series as obtained the n-th character is L
n.To L
ntravel through, meet continuous three elements be all greater than setting threshold value t
n1' or element be greater than setting threshold value t
n2' first element position, as the starting position coordinates vertically relocated of character, be designated as x
n1', the final position coordinate that the position meeting last element of one of above two conditions relocates as level, is designated as x
n2'; Wherein t
n1'=0.5 (y
n2'-y
n1'), t
n2'=(y
n2'-y
n1')-3.According to (x
n1', x
n2') upgrade the binaryzation matrix of each character, obtain relocating binaryzation matrix N
n", size is (y
n2'-y
n1') × (x
n2'-x
n1').
5. method according to claim 1, is characterized in that, the set of matching template described in step 3 M obtains in the following manner:
(1) sampling obtains U the normalization binaryzation matrix D of certain character k
k1, D
k2..., D
kU, this U normalization binaryzation matrix sampling is from the diverse location of different paper money sequence numbers or identical paper money sequence number.
Counting statistics matrix
u=1,2 ..., U, A
kin value minimum be 0, be U to the maximum.Preliminary subtemplate M is set up based on this
k', be specially: to A
kthe point of (i ', j ')>=0.8U (i ', j '), i.e. dense parts, M
k' (i ', j ') get 10; To A
kthe point of (i ', j ')≤0.2U (i ', j '), i.e. sparse part, M
k' (i ', j ') get-10; To 0.2U < A
kthe point of (i ', j ') < 0.8U (i ', j '), M
k' (i ', j ') get 0;
Repeat above-mentioned steps, corresponding preliminary subtemplate is set up to different characters, obtains preliminary matching template set M '.
(2) preliminary matching template set M ' is mated with the normalization binaryzation matrix of sampling in step (1).According to the set of matching result adjustment matching template, be specially:
The normalization binaryzation matrix D of the sampling of character k
kucarry out dot product with each subtemplate in preliminary matching template set M ', and each element summation of product matrix obtained to dot product, obtain element summation r,
a=1,2 ..., A, b=1,2 ..., B; K ' is arbitrary character; If during k '=k, M
k' be the coupling subtemplate of character k, M
k=M
k'; If k ≠ k ', then need to adjust preliminary subtemplate M
k', method is: for preliminary subtemplate M
k' and M
k '', for M
k' (i ", j ")=M
k '' (i ", and the point of j ") (i ", j "), then M
k(i ", j ")=M
k' (i ", j ")-1; For M
k(i ", j ") ≠ M
k '(i ", and the point of j ") (i ", j "), then M
k(i ", j ")=M
k' (i ", j ")+1; Thus obtain matching template M
k.
Repeat above-mentioned steps, corresponding preliminary subtemplate is adjusted to different characters, the matching template set after being adjusted.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510251098.2A CN104899965B (en) | 2015-05-15 | 2015-05-15 | A kind of multinational paper money sequence number recognition methods based on cleaning-sorting machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510251098.2A CN104899965B (en) | 2015-05-15 | 2015-05-15 | A kind of multinational paper money sequence number recognition methods based on cleaning-sorting machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104899965A true CN104899965A (en) | 2015-09-09 |
| CN104899965B CN104899965B (en) | 2017-06-23 |
Family
ID=54032607
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510251098.2A Expired - Fee Related CN104899965B (en) | 2015-05-15 | 2015-05-15 | A kind of multinational paper money sequence number recognition methods based on cleaning-sorting machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104899965B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105336035A (en) * | 2015-10-28 | 2016-02-17 | 深圳怡化电脑股份有限公司 | Smudged serial number image classification method and system |
| CN106296969A (en) * | 2016-08-18 | 2017-01-04 | 深圳怡化电脑股份有限公司 | The recognition methods of bank note and system |
| CN106447905A (en) * | 2016-09-12 | 2017-02-22 | 深圳怡化电脑股份有限公司 | Banknote type identification method and device |
| CN106780961A (en) * | 2015-11-23 | 2017-05-31 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of Iranian bank note face amount and system |
| CN106780960A (en) * | 2015-11-18 | 2017-05-31 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of Iranian bank note currency type and system |
| CN107393124A (en) * | 2016-04-25 | 2017-11-24 | 澧达科技股份有限公司 | Judge the method for banknote kind of currencies and face amount and a variety of banknote acceptors |
| CN107665538A (en) * | 2017-09-28 | 2018-02-06 | 深圳怡化电脑股份有限公司 | A kind of bank note sorting technique, device, equipment and readable storage medium storing program for executing |
| CN107688813A (en) * | 2017-09-24 | 2018-02-13 | 中国航空工业集团公司洛阳电光设备研究所 | A kind of character identifying method |
| CN108665606A (en) * | 2017-03-30 | 2018-10-16 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of bank note Information and device |
| CN108831005A (en) * | 2018-04-16 | 2018-11-16 | 华中科技大学 | A kind of Euro version recognition methods and the system of the multiple features fusion based on image |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09212707A (en) * | 1996-01-30 | 1997-08-15 | Oki Electric Ind Co Ltd | Paper sheets, discriminating device |
| CN101329725A (en) * | 2008-07-30 | 2008-12-24 | 电子科技大学 | Method for dividing fingerprint image based on gradient projection and morphology |
| CN101923741A (en) * | 2010-08-11 | 2010-12-22 | 西安理工大学 | Paper currency number identification method based on currency detector |
| CN102194275A (en) * | 2010-03-15 | 2011-09-21 | 党力 | Automatic ticket checking method for train tickets |
| CN102222384A (en) * | 2011-05-27 | 2011-10-19 | 尤新革 | Analysis method of multispectral image of note |
| CN103456075A (en) * | 2013-09-06 | 2013-12-18 | 广州广电运通金融电子股份有限公司 | Paper money processing method and device |
-
2015
- 2015-05-15 CN CN201510251098.2A patent/CN104899965B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH09212707A (en) * | 1996-01-30 | 1997-08-15 | Oki Electric Ind Co Ltd | Paper sheets, discriminating device |
| CN101329725A (en) * | 2008-07-30 | 2008-12-24 | 电子科技大学 | Method for dividing fingerprint image based on gradient projection and morphology |
| CN102194275A (en) * | 2010-03-15 | 2011-09-21 | 党力 | Automatic ticket checking method for train tickets |
| CN101923741A (en) * | 2010-08-11 | 2010-12-22 | 西安理工大学 | Paper currency number identification method based on currency detector |
| CN102222384A (en) * | 2011-05-27 | 2011-10-19 | 尤新革 | Analysis method of multispectral image of note |
| CN103456075A (en) * | 2013-09-06 | 2013-12-18 | 广州广电运通金融电子股份有限公司 | Paper money processing method and device |
Non-Patent Citations (2)
| Title |
|---|
| 范春年: "汽车车牌自动识别技术研究", 《中国优秀博硕士学位论文全文数据库 (硕士) 信息科技辑》 * |
| 陈勇: "汽车牌照自动识别技术研究", 《中国优秀博硕士学位论文全文数据库 (硕士) 信息科技辑》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105336035A (en) * | 2015-10-28 | 2016-02-17 | 深圳怡化电脑股份有限公司 | Smudged serial number image classification method and system |
| CN105336035B (en) * | 2015-10-28 | 2019-02-01 | 深圳怡化电脑股份有限公司 | A kind of method and system of dirty crown word number image classification |
| CN106780960A (en) * | 2015-11-18 | 2017-05-31 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of Iranian bank note currency type and system |
| CN106780961A (en) * | 2015-11-23 | 2017-05-31 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of Iranian bank note face amount and system |
| CN107393124A (en) * | 2016-04-25 | 2017-11-24 | 澧达科技股份有限公司 | Judge the method for banknote kind of currencies and face amount and a variety of banknote acceptors |
| CN106296969A (en) * | 2016-08-18 | 2017-01-04 | 深圳怡化电脑股份有限公司 | The recognition methods of bank note and system |
| CN106296969B (en) * | 2016-08-18 | 2019-04-12 | 深圳怡化电脑股份有限公司 | The recognition methods and system of bank note |
| CN106447905A (en) * | 2016-09-12 | 2017-02-22 | 深圳怡化电脑股份有限公司 | Banknote type identification method and device |
| CN108665606A (en) * | 2017-03-30 | 2018-10-16 | 深圳怡化电脑股份有限公司 | A kind of recognition methods of bank note Information and device |
| CN107688813A (en) * | 2017-09-24 | 2018-02-13 | 中国航空工业集团公司洛阳电光设备研究所 | A kind of character identifying method |
| CN107665538A (en) * | 2017-09-28 | 2018-02-06 | 深圳怡化电脑股份有限公司 | A kind of bank note sorting technique, device, equipment and readable storage medium storing program for executing |
| CN108831005A (en) * | 2018-04-16 | 2018-11-16 | 华中科技大学 | A kind of Euro version recognition methods and the system of the multiple features fusion based on image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104899965B (en) | 2017-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104899965A (en) | Multi-national paper money serial number identification method based on sorting machine | |
| CN105303676B (en) | A kind of version recognition methods of bank note and system | |
| US10430681B2 (en) | Character segmentation and recognition method | |
| CN105184801B (en) | It is a kind of based on multi-level tactful optics and SAR image high-precision method for registering | |
| CN104408449B (en) | Intelligent mobile terminal scene literal processing method | |
| CN108197644A (en) | A kind of image-recognizing method and device | |
| CN109934181A (en) | Text recognition method, device, equipment and computer-readable medium | |
| CN102800148A (en) | RMB sequence number identification method | |
| CN104680161A (en) | Digit recognition method for identification cards | |
| CN101145196B (en) | Quick fingerprint identification method based on strange topology structure | |
| US8805056B2 (en) | Automatic detection and grouping of straight lines in images for personalization | |
| CN103034848A (en) | Identification method of form type | |
| CN106503694B (en) | Digit recognition method based on eight neighborhood feature | |
| CN105354578A (en) | Multi-target object image matching method | |
| CN105868759A (en) | Method and apparatus for segmenting image characters | |
| CN104658097A (en) | RMB paper currency denomination identification method based on histogram matching of images | |
| CN106845542A (en) | Paper money number intelligent identification Method based on DSP | |
| CN106846354B (en) | A kind of Book Inventory method on the frame converted based on image segmentation and random hough | |
| CN103279752A (en) | Eye locating method based on improved Adaboost algorithm and human face geometrical characteristics | |
| CN103473537A (en) | Method and device for representing contour feature of target image | |
| CN107578438A (en) | Circle recognition methods, device and electronic equipment | |
| CN105184225A (en) | Multinational paper money image identification method and apparatus | |
| CN108197624A (en) | The recognition methods of certificate image rectification and device, computer storage media | |
| CN105957237A (en) | Method and device for identifying version of banknote | |
| CN110826551B (en) | Intelligent screening method for VIN code topology printing mode image of motor vehicle |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170623 |