CN100447815C - Method for compressing fingerprint direction quantized diagram to embedded system - Google Patents
Method for compressing fingerprint direction quantized diagram to embedded system Download PDFInfo
- Publication number
- CN100447815C CN100447815C CNB2005101081350A CN200510108135A CN100447815C CN 100447815 C CN100447815 C CN 100447815C CN B2005101081350 A CNB2005101081350 A CN B2005101081350A CN 200510108135 A CN200510108135 A CN 200510108135A CN 100447815 C CN100447815 C CN 100447815C
- Authority
- CN
- China
- Prior art keywords
- data
- field
- difference
- storage
- statistics
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000010586 diagram Methods 0.000 title claims description 23
- 238000003860 storage Methods 0.000 claims abstract description 35
- 238000011002 quantification Methods 0.000 claims abstract description 22
- 238000007906 compression Methods 0.000 claims description 22
- 230000006835 compression Effects 0.000 claims description 22
- 239000012634 fragment Substances 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 5
- 238000013144 data compression Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
Images
Abstract
This invention is to open a quantized compressing method for finger-print directional graph in a embedding system. It is includes: 1. quantize level and sectional storage parameter selection; 2. directional graph quantification; 3. getting forward difference of the quantized directional graph; 4. substitute the original data by the sum of the repeating times of forward difference and the original data; sectional store the directional data and the forward difference; read the compressed data by the method stated in procedure 4.
Description
Technical field
The invention belongs to embedded system technology and digital image processing techniques field, be particularly related to digital image compression coding and embedded system data memory technology, be specially the fingerprint orientation quantification and the compression storing data and the method that reads that are applicable to embedded system.
Background technology
Fingerprint recognition or checking generally are to realize that by the characteristic of fingerprint is compared and mated this characteristic is called template.Template and nonprimitive fingerprint image, but to the numeral of key feature in the fingerprint image.After the input fingerprint image extracts key feature, form template, can save the storage space of fingerprint recognition system, reduce the capacity of Network Transmission channel, be convenient to realize fingerprint strange land coupling.Feature commonly used in the Automated Fingerprint Identification System comprises minutiae feature and field of direction feature etc.Minutiae point refers to the bifurcation or the end points of fingerprint ridge, and the characteristic of extraction comprises the minutiae point coordinate, the direction of the local streakline at minutiae point place and curvature, and minutiae point type information such as (end points or bifurcations).The above-mentioned information of a minutiae point of storage generally needs 5 bytes, it is limited that true details on one width of cloth fingerprint image is counted, usually can be above 100, therefore the template data of storing the minutiae feature of one piece of fingerprint generally can not surpass 500 bytes, can satisfy the requirement for restriction of most of Automated Fingerprint Identification Systems to the template base memory space.
Yet what be different from minutiae feature is that field of direction feature needs relatively large storage space.The directional diagram of fingerprint is to represent original streakline with the direction of streakline, and it is a kind of key character of fingerprint image, can describe the one-piece construction of fingerprint image.Except being used for fingerprint matching, important use is also arranged at aspects such as fingerprint image preprocessing, feature extraction, central point detection, classification and fingerprint image synthesize.There are a large amount of directional diagram computing method at present, general algorithm for recognizing fingerprint adopts block directed graph, the source fingerprint image is divided into the fritter (generally getting w=8) of w * w, uses field of direction computing method, calculate the crestal line direction of each fritter based on the local gray level gradient.All the direction of block image has constituted the directional diagram of fingerprint image.Can find,, be difficult to satisfy the requirement of embedded Automated Fingerprint Identification System even storage block field of direction data only still need bigger storage space.Fortunately, the field of direction feature of fingerprint exists very strong correlativity, and this just provides important basis for we compress the direction diagram data.
Summary of the invention
The objective of the invention is to propose a kind of embedded system that is applicable to, neither can influence fingerprint recognition performance, the fingerprint orientation that can improve compression efficiency again quantizes the method for compression.
Fingerprint orientation towards embedded system quantizes compression method, and comprise step: direction of passage figure quantizes in piece field of direction data normalization to an integer range; On the directional diagram that quantizes, to ask for the forward direction difference of the piece field of direction, and add up the number of times that the forward direction difference data repeats continuously, the number of times that repeats with field of direction data forward direction difference adds that data itself replace raw data; Distinguish the storage section, be used for the statistics of storage direction field data and field of direction forward direction difference respectively.
Also comprise step: according to the precision needs of algorithm and the capacity requirement of template, choose quantification gradation and fragment bit stored parameter, between quantified precision and compression efficiency, obtain balance by same parameter.
Also comprise step: according to the quantification gradation parameter of choosing, the direction of passage diagram data quantizes the angle of the piece field of direction is normalized in the integer range.
Also comprise step: on the directional diagram that quantizes, with the piecemeal is unit, ask for the forward direction difference of the piece field of direction line by line, and add up the number of times that the forward direction difference data repeats continuously, the number of times that repeats with field of direction data forward direction difference adds that data itself replace raw data.
Also comprise step: fragment bit compression storage---according to quantification gradation of choosing and fragment bit stored parameter, the byte or the field of 8 bits are divided into two sections, a section is used for the storage direction field data, the statistics of another section storage direction field forward direction difference.
Also comprise step: packed data reads---at first read the statistics position, and then read direction field data position, and according to statistics, direction field data block-by-block is compensated with poor summation.
Description of drawings
Fig. 1 is the fingerprint orientation quantification compression Stored Procedure figure towards embedded system.
Fig. 2 is fingerprint orientation and source fingerprint image comparison diagram after quantizing.
Fig. 3 is the embodiment synoptic diagram of fragment bit storage direction field data and difference statistics.
Embodiment
Further specify the operating process that quantizes compression method towards the fingerprint orientation of embedded system below by embodiment.As shown in Figure 1, the specific implementation step of directional diagram quantification compression storage is as follows:
(1-1) determine quantification gradation and fragment bit stored parameter N.If being used for the maximal bit figure place note of storage direction field data makes N, the quantification gradation of field of direction data just is 2 so
NAccording to the precision needs of algorithm and the capacity requirement of template, we need seek one and trade off between quantified precision and compression efficiency, usually, can choose N and equal 3,4,5 or 6.
(1-2) directional diagram data-measuring.Normalize in the integer range by the angle of quantification the piece field of direction.The angle θ size of the field of direction is generally at 0~180 °, perhaps-90 °~90 ° between, if the latter need transform in the former scope.We carry out 2 with 180 ° angular range
NFive equilibrium, the field of direction data after the position of the by stages such as dropping on according to field of direction angle is determined to quantize.Concrete computing method are as follows: establish the field of direction data fingerprint effective coverage in and be θ (i, j), wherein, i, the position coordinates of j presentation video piece, the field of direction data after quantizing so are: [θ (i, j) * 32 ÷ 180], [] represented wherein numerical value is rounded.Can know that the numerical range of asking for is 0~2
NBetween-1.What deserves to be explained is that the interested field of direction numerical value of fingerprint recognition is generally relative data, therefore, only need in the fingerprint image template, store the field of direction data after quantizing,, also can calculate according to the data back after quantizing as the needs raw data.In addition, field of direction data are stored continuously according to from top to bottom order from left to right, therefore do not need stored position information.Fig. 2 (b) is depicted as the fingerprint orientation after the quantification, and quantification gradation is 2
4, Fig. 2 (a) is the source fingerprint image, can find that by contrast the directional diagram of quantification has kept the field of direction feature of former fingerprint image.
(1-3) ask for the forward direction difference.On the directional diagram that quantizes,, ask for difference successively with respect to the field of direction data of previous piecemeal according to from top to bottom order from left to right.
(1-4) the continuous number of times that repeats of statistics difference.For the forward direction difference of continuous repetition, to calculate its number of times that occurs continuously and deduct 1, note is made iCount.To ask for the forward direction difference the same for same step (1-3) during the continuous multiplicity of statistics difference, is unit with the piecemeal of directional diagram, carries out line by line.In fact, the principle of image run-length encoding is: the neighbor that color in the delegation is identical replaces with a count value and this color value.Thought based on run-length encoding, we add that with the number of times that field of direction data forward direction difference repeats data itself replace raw data, the method of this data compression is not only than the method efficient height of fragment bit storage direction field data, and be better than image stroke Methods for Coding, this is because we have made full use of the characteristics of fingerprint orientation---generally, the piece field of direction data of fingerprint image vicinity are more approaching, generally can not suddenly change, and of paramount importance is that the field of direction of contiguous block often changes along same direction.
(1-5) distinguish the storage section.The byte of 8 bits (bit) position is divided into two sections,, at first a word (word) is divided into most-significant byte and two fields of least-significant byte, and then be divided into two sections respectively for 16 processor.A section is used for the storage direction field data, the statistics of another section storage direction field forward direction difference, being used for the number of bits N of storage direction field data is exactly definite quantification gradation of step (1-1) and fragment bit stored parameter, and the bit K of storage statistical data then equals 8-N.
(1-6) fragment bit storage direction field and difference statistics.The statistics that starting block or forward direction difference are unduplicated is all remembered work 0.The difference statistics maximum occurrences that can store is 2
K-1, if statistics iCount is greater than 2
K-1, then must handle several times.Can find that maximum compression ratio is in the ideal case
For 16 processor, owing to divide high and low 8 storages, then maximum compression ratio can reach in theory
With 16 bit processors, quantification gradation and fragment bit stored parameter N equal 5 and are example, and the fragment bit data storing method is described below.Octet is divided into 5 and 3 two sections, and preceding 5 field of direction data of depositing quantification are deposited the image block statistical number with identical forward direction difference for back 3.Particularly, be followed successively by 7,9,11,13,15,17 as field of direction data, 19,21,23,25,27,29,28,27,26,25,24,23, the field of direction data of 7 corresponding starting blocks wherein, then corresponding forward direction difference is followed successively by 2,2,2,2,2,2,2,2,2,2,2 ,-1 ,-1 ,-1 ,-1,-1 ,-1, so the result of fragment bit compression storage is: (0x07|0x00), (0x09|0x07), (0x19|0x02), (0x1C|0x05).For 16 embedded system, as shown in Figure 3, can represent that ratio of compression reaches with two words (WORD)
The method that above-mentioned compression storage has been arranged also is easy to during reading of data so.At first read the low K position of an octet or field, obtain statistics iCount, if iCount is 0x00, then direct numerical value with high N position deposits the directional diagram buffer zone in; If iCount is greater than 0x00, then at first deposit the field of direction numerical value of high N position in buffer zone, poor with the previous field of direction numerical value of buffer zone then, with this same difference, repeat iCount time to direction Flow Field Numerical block-by-block summation compensation, deposit buffer zone successively in.
The method of above-mentioned fingerprint orientation data compression is simple fast effectively, the technology of the present invention successful Application we the exploitation the Automated Fingerprint Identification System module in, this system module was once repeatedly exhibition in various technology exhibitions, and, fed back to gratifying result by client's test that the investor brings Europe.In addition, we are developing is used for the product of communicating by letter with long-range host computer, and data compression is even more important, and the present invention has solved this technical need preferably.
Claims (7)
1. the fingerprint orientation towards embedded system quantizes compression method, and comprise step: direction of passage figure quantizes in piece field of direction data normalization to an integer range; On the directional diagram that quantizes, to ask for the forward direction difference of the piece field of direction, and add up the number of times that the forward direction difference data repeats continuously, the number of times that repeats with field of direction data forward direction difference adds that data itself replace raw data; Distinguish the storage section, be used for the statistics of storage direction field data and field of direction forward direction difference respectively.
2. by the described method of claim 1, it is characterized in that, also comprise step: according to the precision needs of algorithm and the capacity requirement of template, choose quantification gradation and fragment bit stored parameter, quantification gradation and fragment bit stored parameter are determined by same numerical value, obtain balance between quantified precision and compression efficiencies.
3. by the described method of claim 1, it is characterized in that, also comprise step: according to the quantification gradation parameter of choosing, the direction of passage diagram data quantizes the angle of the piece field of direction is normalized in the integer range.
4. by the described method of claim 1, it is characterized in that, also comprise step: on the directional diagram that quantizes, with the piecemeal is unit, ask for the forward direction difference of the piece field of direction line by line, and add up the number of times that the forward direction difference data repeats continuously, the number of times that repeats with field of direction data forward direction difference adds that data itself replace raw data.
5. by the described method of claim 1, it is characterized in that, also comprise step: fragment bit compression storage---according to quantification gradation of choosing and fragment bit stored parameter, the byte or the field of 8 bits are divided into two sections, a section is used for the storage direction field data, the statistics of another section storage direction field forward direction difference.
6. by the described method of claim 1, it is characterized in that also comprise step: packed data reads---at first read the statistics position, and then read direction field data position, and according to statistics, direction field data block-by-block is compensated with poor summation.
7. by the described method of claim 1, the concrete steps that directional diagram quantizes the compression storage are as follows:
(1-1) determine quantification gradation and fragment bit stored parameter, fragment bit stored parameter N is used for the maximal bit figure place of storage direction field data, and the quantification gradation of field of direction data just is 2 so
N, according to the precision needs of algorithm and the capacity requirement of template, we need seek one and trade off between quantified precision and compression efficiency, choose N and equal 3,4,5 or 6;
(1-2) directional diagram data-measuring, the angle of the piece field of direction is normalized in the integer range by quantizing, the angle θ size of the field of direction is at 0~180 °, perhaps-90 °~90 ° between, if the latter need transform in the former scope, we carry out 2 with 180 ° angular range
NFive equilibrium, field of direction data after the position of the by stages such as dropping on according to field of direction angle is determined to quantize, concrete computing method are as follows: the field of direction data of establishing in the fingerprint effective coverage are θ (i, j), wherein, i, the position coordinates of j presentation video piece, field of direction data after quantizing so are: [θ (i, j) * 2
N÷ 180], (N=3,4,5,6), [] expression rounds numerical value wherein;
(1-3) ask for the forward direction difference, on the directional diagram that quantizes,, ask for difference successively with respect to the field of direction data of previous piecemeal according to from top to bottom order from left to right;
(1-4) the continuous number of times that repeats of statistics difference, forward direction difference for continuous repetition, calculate its number of times that occurs continuously and deduct 1, note is made iCount, to ask for the forward direction difference the same for same step (1-3) during the continuous multiplicity of statistics difference, is unit with the piecemeal of directional diagram, carries out line by line, based on the thought of run-length encoding, the number of times that repeats with field of direction data forward direction difference adds that data itself replace raw data;
(1-5) distinguish the storage section, the byte of 8 bits is divided into two sections, processor for 16, at first a block is divided into most-significant byte and two fields of least-significant byte, and then being divided into two sections respectively, a section is used for the storage direction field data, the statistics of another section storage direction field forward direction difference, being used for the number of bits of storage direction field data is exactly the definite fragment bit stored parameter N of step (1-1), and the bit K of storage statistical data then equals 8-N;
(1-6) fragment bit storage direction field and difference statistics, the statistics that starting block or forward direction difference are unduplicated are all remembered work 0, and the difference statistics maximum occurrences that can store is 2
K-1, if statistics iCount is greater than 2
K-1, then must handle several times.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2005101081350A CN100447815C (en) | 2005-09-29 | 2005-09-29 | Method for compressing fingerprint direction quantized diagram to embedded system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNB2005101081350A CN100447815C (en) | 2005-09-29 | 2005-09-29 | Method for compressing fingerprint direction quantized diagram to embedded system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1940995A CN1940995A (en) | 2007-04-04 |
CN100447815C true CN100447815C (en) | 2008-12-31 |
Family
ID=37959144
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNB2005101081350A Expired - Fee Related CN100447815C (en) | 2005-09-29 | 2005-09-29 | Method for compressing fingerprint direction quantized diagram to embedded system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN100447815C (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102262208B (en) * | 2010-05-31 | 2015-11-25 | 无锡中星微电子有限公司 | A kind of method and system of chip testing |
CN104537099A (en) * | 2015-01-09 | 2015-04-22 | 北京道通天下信息科技有限责任公司 | Storage method and device and electronic equipment |
WO2018126368A1 (en) * | 2017-01-05 | 2018-07-12 | 深圳市汇顶科技股份有限公司 | Touch control device and method for determining capacitive sensing amount of touch control device |
CN107515931B (en) * | 2017-08-28 | 2023-04-25 | 华中科技大学 | Repeated data detection method based on clustering |
CN110677156A (en) * | 2019-09-19 | 2020-01-10 | 南京国电南自电网自动化有限公司 | Compression algorithm and decompression method for black and white dot matrix data in power system protection device |
CN112199049B (en) * | 2020-10-22 | 2023-10-20 | Tcl通讯(宁波)有限公司 | Fingerprint storage method, fingerprint storage device and terminal |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1021857C (en) * | 1989-11-23 | 1993-08-18 | 北京大学 | Circuit for extracting discriminating features of finger print |
US5995642A (en) * | 1997-06-30 | 1999-11-30 | Aetex Biometric Corporation | Method for automatic fingerprint classification |
CN1617161A (en) * | 2003-11-10 | 2005-05-18 | 北京握奇数据系统有限公司 | Finger print characteristic matching method based on inter information |
-
2005
- 2005-09-29 CN CNB2005101081350A patent/CN100447815C/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1021857C (en) * | 1989-11-23 | 1993-08-18 | 北京大学 | Circuit for extracting discriminating features of finger print |
US5995642A (en) * | 1997-06-30 | 1999-11-30 | Aetex Biometric Corporation | Method for automatic fingerprint classification |
CN1617161A (en) * | 2003-11-10 | 2005-05-18 | 北京握奇数据系统有限公司 | Finger print characteristic matching method based on inter information |
Non-Patent Citations (1)
Title |
---|
基于块方向信息的指纹图像无损压缩编码. 唐良瑞,蔡安妮,孙景鳌.计算机工程,第28卷第11期. 2002 * |
Also Published As
Publication number | Publication date |
---|---|
CN1940995A (en) | 2007-04-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Duan et al. | Overview of the MPEG-CDVS standard | |
Chen et al. | Tree histogram coding for mobile image matching | |
Ramadan et al. | Face recognition using particle swarm optimization-based selected features | |
CN100447815C (en) | Method for compressing fingerprint direction quantized diagram to embedded system | |
US7174044B2 (en) | Method for character recognition based on gabor filters | |
KR101565265B1 (en) | Coding of feature location information | |
US6438268B1 (en) | Vector quantization codebook generation method | |
US8891878B2 (en) | Method for representing images using quantized embeddings of scale-invariant image features | |
Liaw et al. | Fast exact k nearest neighbors search using an orthogonal search tree | |
Khashman et al. | Image compression using neural networks and Haar wavelet | |
KR101912748B1 (en) | Scalable Feature Descriptor Extraction and Matching method and system | |
CN103218427A (en) | Local descriptor extracting method, image searching method and image matching method | |
CN104838653A (en) | Lossless image compression using differential transfer | |
CN105374054A (en) | Hyperspectral image compression method based on spatial spectrum characteristics | |
CN108520265B (en) | Method for converting image descriptors and related image processing device | |
CN104115162A (en) | Image analysis | |
Baroffio et al. | Coding local and global binary visual features extracted from video sequences | |
Li et al. | Quantized embeddings of scale-invariant image features for mobile augmented reality | |
CN112395503A (en) | Face recognition-based sharing platform intelligent recommendation method and system and readable storage medium | |
Duan et al. | Fast MPEG-CDVS encoder with GPU-CPU hybrid computing | |
Wang et al. | A fast algorithm for mining association rules in image | |
CN108536772B (en) | Image retrieval method based on multi-feature fusion and diffusion process reordering | |
CN108712655A (en) | A kind of group's image encoding method merged for similar image collection | |
Levenets | The Basic principles and methods of the system approach to compression of telemetry data | |
Merrouche et al. | Accuracy analysis of lossless and lossy disparity map compression |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20081231 Termination date: 20150929 |
|
EXPY | Termination of patent right or utility model |