CN111428064B - Small-area fingerprint image fast indexing method, device, equipment and storage medium - Google Patents
Small-area fingerprint image fast indexing method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN111428064B CN111428064B CN202010526646.9A CN202010526646A CN111428064B CN 111428064 B CN111428064 B CN 111428064B CN 202010526646 A CN202010526646 A CN 202010526646A CN 111428064 B CN111428064 B CN 111428064B
- Authority
- CN
- China
- Prior art keywords
- fingerprint image
- matching
- grid
- point
- small
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
- G06F16/535—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/757—Matching configurations of points or features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention relates to the field of image comparison, and discloses a method, a device, equipment and a readable storage medium for quickly indexing a small-area fingerprint image, wherein the method comprises the following steps: the description area is determined by taking the feature points as the center, the features of the description area are represented by the difference value of the feature data, so that the descriptors have more stable expression capacity, in addition, the description information is encoded into binary data and then subjected to Hash mapping, the robustness of the encoded information can be ensured, furthermore, the descriptors of different feature points are subjected to Hash mapping to form index values through Hash mapping, the whole data set can be preliminarily screened during comparison, a matching point set similar to the feature points is extracted, the distance calculation with all the points of the whole data set during query is avoided, and the comparison efficiency is greatly improved.
Description
Technical Field
The invention relates to the field of image retrieval, in particular to a method, a device, equipment and a storage medium for quickly indexing a small-area fingerprint image.
Background
There are three key modules in a fingerprint image recognition system: the fingerprint image processing device comprises a fingerprint image preprocessing module, a feature extraction module and a feature comparison module. The fingerprint image preprocessing module is used for carrying out a series of processing on an original image, different fingerprint identification systems have different mechanisms of selecting preprocessing methods, and operations which may include fingerprint image filtering, gradient field and direction field calculation, image foreground and background separation, image normalization, image enhancement, binaryzation, thinning and the like. The fingerprint image feature extraction module is mainly used for extracting global feature or local feature information capable of expressing features of fingerprint images, and methods which may be selected by different fingerprint identification systems include traditional minutiae extraction, feature topological structures or codes based on minutiae methods, streak line structure features, machine vision feature codes, image correlation and the like. For a fingerprint feature comparison module, different systems can make comparison strategies according to feature extraction and organization methods of the fingerprint feature comparison module, most systems only pay attention to how to improve matching accuracy and neglect matching time overhead, and in fact, the rapid indexing can be considered in the design of feature extraction and feature comparison, and the fingerprint feature comparison module is suitable for the rapid indexing function of a database. The existing fingerprint identification system considering a feature extraction method and a quick indexing mechanism simultaneously has the following strategies:
(1) and constructing a fingerprint index based on the global features and the ridge features. Most of fingerprint images are divided into a plurality of classes by using the characteristics related to the direction and frequency of fingerprint lines to construct indexes, so that the matching range can be reduced during identification, and the method has limited index efficiency;
(2) and constructing a fingerprint image index based on the geometric structure information or the construction descriptor among the minutiae points. For example, Liang et al construct an index by using the structural information of the minutiae and by constructing a Delaunay triangulation method; MatteoFerrara et al propose an index is established for fingerprint feature data based on Minutia-Cylinder-code (MCC), and because the MCC descriptor is of a fixed length, LSH (local Sensitive Hash) can be used for picture indexing, and the efficiency of fingerprint identification under mass data can be improved to a greater extent by using the method. The strategy of the type has better effect on large-area fingerprint images (with sufficient detail feature points), has higher memory and calculation speed, is more intuitive and easy to understand in feature extraction and comparison, but is not feasible for small-area fingerprint images, and does not have enough detail points to construct geometric relations or MCC codes and the like;
(3) and constructing a fingerprint image index based on SIFT or SURF characteristics. Usually, a scale space is constructed by utilizing Gaussian difference, key points are detected by using Hessian matrix and other modes, then descriptors are constructed for the key points, and indexes are formulated by comparing the descriptors, the strategy has large calculation and memory overhead, and the size of an extracted feature template of a single fingerprint image is in the order of hundred Kb;
in summary, the conventional fingerprint identification strategy cannot identify a small-area fingerprint image, and has a large memory required for calculation during indexing, which is not efficient.
Disclosure of Invention
The invention aims to provide a method for quickly indexing a small-area fingerprint image, which has the characteristics of improving the indexing efficiency of the small-area fingerprint image and reducing the memory resources required by calculation.
The above object of the present invention is achieved by the following technical solutions:
when the indexing mechanism is established:
extracting a preset number of matching points from the reference fingerprint image, wherein the extracted matching points at least comprise coordinate information and reference direction information;
calculating to obtain the description information of the matching points;
compiling the description information into binary data according to a preset compiling rule to obtain a first descriptor;
performing hash calculation on the first descriptor through a hash function to obtain a hash value of the first descriptor, and taking the hash value as a first index value of the first descriptor;
when the indexing mechanism is applied:
extracting a characteristic point of the fingerprint image to be compared, and building the characteristic point according to an index mechanism to obtain a second descriptor and a second index value;
screening out feature points corresponding to the first index value with the Hamming distance within a preset threshold value by calculating the Hamming distance between the first index value and the second index value to obtain a similar point set of the feature points;
judging whether the matching point corresponding to the closest Hamming distance and the feature point are matching point pairs or not according to the Hamming distance of the descriptor between the feature point and each matching point in the similar point set;
continuously extracting n characteristic points from the fingerprint image to be compared, traversing each characteristic point of the image to be compared to execute the indexing mechanism application, and counting the logarithm of the matching points;
judging whether the logarithm of the matching points is greater than a preset logarithm threshold value or not;
if yes, the fingerprint image to be compared is judged to be similar to the reference fingerprint image.
By adopting the technical scheme, the description area is determined by taking the feature points as the center, the features of the description area are represented by the difference value of the feature data, so that the descriptors have more stable expression capability, in addition, the description information is encoded into binary data, the robustness of the encoded information can be ensured by performing hash mapping, furthermore, the descriptors of different feature points are subjected to hash mapping to form index values by the hash mapping, the whole data set can be preliminarily screened during comparison, the matching point set similar to the feature points is extracted, the distance calculation with all the points of the whole data set is avoided during query, and the comparison efficiency is greatly improved.
The present invention in a preferred example may be further configured to:
determining a description area according to the size of a preset area by taking the matching point as a center;
performing multi-level blocking on the description area to obtain grid small blocks of the description area;
respectively calculating the characteristic data of each grid small block;
pairing the grid small blocks pairwise according to a preset pairing rule to obtain a grid block pair;
and extracting the grid block pair with a preset logarithm to calculate the characteristic data difference value of 2 grid small blocks in the grid block pair to obtain the description information.
By adopting the technical scheme, the characteristic data difference value between the 2 grid small blocks in the grid block pair is calculated, and the characteristic data difference value of the grid block pair is used as description information, so that the characteristics of the description area are more stably expressed.
The present invention in a preferred example may be further configured to:
the characteristic data comprises a gray level mean value, an X-direction gradient mean value, a Y-direction gradient mean value and a curvature mean value of the grid small blocks.
By adopting the technical scheme, the accuracy of the description information on the description area is improved.
The present invention in a preferred example may be further configured to:
presetting a variation range of the description information near 0;
extracting a maximum value and a minimum value from the description information, and obtaining different value ranges of the description information by combining the change range;
and compiling the description information into corresponding binary data according to the preset description information corresponding to different binary data in different value ranges.
By adopting the technical scheme, the comparison speed in the later indexing process is increased, and the memory occupation of a computer is saved.
The present invention in a preferred example may be further configured to:
sequencing all grid small blocks to obtain the serial number of each grid small block;
and according to the preset matching sequence number, matching the 2 grid small blocks corresponding to the matching sequence number to obtain a grid block pair.
By adopting the technical scheme, the specific grid small block object to be differentiated is determined for calculating the difference value of the characteristic data.
The present invention in a preferred example may be further configured to:
extracting corresponding logarithm grid block pairs according to the preset logarithm of extracted grid block pairs;
carrying out differential processing on the characteristic data of 2 grid small blocks in each pair of grid blocks to obtain a difference value sequence of the characteristic data of each pair of grid small blocks;
and arranging and combining the difference value sequences to obtain the description information.
By adopting the technical scheme, the characteristics of the description area are more stably expressed.
The present invention in a preferred example may be further configured to:
calculating the Hamming distance between the first descriptor and the second descriptor corresponding to the matching points in the similar point set, and screening out the nearest Hamming distance and the next nearest Hamming distance;
judging whether the ratio of the nearest Hamming distance to the next nearest Hamming distance is within a preset threshold value;
if yes, judging that the matching point corresponding to the closest Hamming distance and the feature point are a pair of matching point pairs.
By adopting the technical scheme, the matching points which are the best and excellent with the characteristic points are screened out.
The invention also aims to provide a small-area fingerprint image fast indexing device which has the characteristics of improving the image comparison speed and reducing the memory occupation of a computer.
The second aim of the invention is realized by the following technical scheme:
a small-area fingerprint image fast indexing device comprises:
the first extraction module is used for extracting a preset number of matching points from the reference fingerprint image, and the extracted matching points at least comprise coordinate information and reference direction information;
the calculation module is used for calculating and obtaining the description information of the matching points;
the compiling module is used for compiling the description information into binary data according to a preset compiling rule to obtain a first descriptor;
the processing module is used for carrying out hash calculation on the first descriptor through a hash function to obtain a hash value of the first descriptor, and the hash value is used as a first index value of the first descriptor;
the second extraction module is used for extracting a characteristic point of the fingerprint image to be compared, and obtaining a second descriptor and a second index value through the processing of the calculation module, the compiling module and the processing module; the screening module is used for screening out feature points corresponding to the first index value with the Hamming distance within a preset distance threshold value by calculating the Hamming distance between the first index value and the second index value to obtain a similar point set of the feature points;
the first judgment module is used for judging whether the matching point corresponding to the closest Hamming distance and the feature point are matching point pairs or not according to the Hamming distance of the descriptor between the feature point and each matching point in the similar point set;
the statistical module is used for continuously extracting n characteristic points from the fingerprint image to be compared, traversing each characteristic point of the fingerprint image to be compared, and performing statistics on the logarithm of the matching point after the characteristic points are processed by the calculation module, the compiling module, the processing module, the screening module and the first judgment module;
the second judgment module is used for judging whether the logarithm of the matching points is greater than a preset logarithm threshold value;
and the judging module is used for judging that the fingerprint image to be compared is similar to the reference fingerprint image if the matching point number is greater than the preset logarithm threshold value.
Preferably, the calculation module comprises:
the determining unit is used for determining a description area according to the size of a preset area by taking the matching point as a center;
the partitioning unit is used for partitioning the description area in multiple stages to obtain grid small blocks of the description area;
the first calculating unit is used for respectively calculating the characteristic data of each grid small block;
the matching unit is used for matching the grid small blocks pairwise according to a preset matching rule to obtain a grid block pair;
and the second calculation unit is used for extracting the characteristic data difference values of the 2 grid small blocks in the grid block pair calculation grid block pair with the preset logarithm to obtain the description information.
Preferably, the pairing unit includes:
the sorting subunit is used for sorting all the grid small blocks to obtain the serial number of each grid small block;
and the matching subunit is used for matching the 2 grid small blocks corresponding to the matching serial numbers according to the preset matching serial numbers to obtain the grid block pairs.
Preferably, the second calculation unit includes:
the extraction subunit is used for extracting the corresponding logarithm of the grid blocks according to the preset logarithm of the extracted grid block pairs;
the difference subunit is used for carrying out difference processing on the characteristic data of 2 grid small blocks in each pair of grid blocks to obtain a difference sequence of the characteristic data of each pair of grid small blocks;
and the combination unit is used for arranging and combining the difference value sequences to obtain the description information.
The invention aims to provide small-area fingerprint image fast indexing equipment which has the characteristics of fast comparison of fingerprint images and capability of running on an embedded platform.
The third object of the invention is realized by the following technical scheme:
a small-area fingerprint image fast indexing device comprises a memory and a processor, wherein the memory is stored with a computer program which can be loaded by the processor and can execute the small-area fingerprint image fast indexing method.
The fourth purpose of the invention is to provide a computer storage medium which can store corresponding programs and has the characteristic of being convenient for realizing the quick indexing of small-area fingerprint images.
The fourth object of the invention is realized by the following technical scheme:
a computer storage medium storing a computer program that can be loaded by a processor and that can perform any of the above-described small-area fingerprint image fast indexing methods.
In summary, the invention has the following beneficial technical effects: the description area is determined by taking the feature points as the center, the features of the description area are represented by the difference value of the feature data, so that the descriptors have more stable expression capacity, in addition, the description information is encoded into binary data, the robustness of the encoded information can be ensured by performing hash mapping, furthermore, the descriptors of different feature points are subjected to hash mapping to form index values through the hash mapping, the whole data set can be preliminarily screened during comparison, a matching point set similar to the feature points is extracted, the distance calculation with all the points of the whole data set during query is avoided, and the comparison efficiency is greatly improved.
Drawings
FIG. 1 is a flowchart illustrating a method for fast indexing a small-area fingerprint image according to an embodiment of the present invention;
FIG. 2 is a schematic view of a detailed flow chart of an embodiment of step S20 in FIG. 1;
FIG. 3 is a schematic view illustrating a detailed flow of an embodiment of a compiling rule of the small-area fingerprint image fast indexing method according to the present invention;
FIG. 4 is a schematic view of a detailed process of step S204 in FIG. 2;
FIG. 5 is a schematic view of a detailed process of step S205 in FIG. 2;
FIG. 6 is a schematic view of a detailed flow chart of an embodiment of step T30 in FIG. 1;
FIG. 7 is a functional model diagram of an embodiment of the apparatus for fast indexing a small-area fingerprint image according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
The present embodiment is only for explaining the present invention, and it is not limited to the present invention, and those skilled in the art can make modifications of the present embodiment without inventive contribution as needed after reading the present specification, but all of them are protected by patent law within the scope of the claims of the present invention.
Referring to fig. 1, fig. 1 is a flowchart illustrating a first embodiment of a method for fast indexing a small-area fingerprint image according to the present invention. In this embodiment, the method for quickly indexing a small-area fingerprint image includes:
when the indexing mechanism is established:
step S10: extracting a preset number of matching points from the reference fingerprint image, wherein the extracted matching points at least comprise coordinate information and reference direction information;
in the embodiment of the invention, one fingerprint image is determined as a reference fingerprint image and the other fingerprint image to be compared from 2 fingerprint images to be compared, and N matching points are extracted from the reference fingerprint image. The method for extracting the matching points from the reference fingerprint image can be a traditional minutiae extraction method, or can be a method for extracting deep-level gray-level semantic images of the fingerprint image, and then extreme points in the deep-level semantic images are selected as feature points or a network capable of identifying the feature points at a deep learning network training position is adopted for feature point extraction. The invention extracts the deep gray level semantic image of the fingerprint image, and then selects the extreme point in the deep level semantic image as the characteristic point, and the specific extraction process is as follows:
converting the reference fingerprint image into a gray level image, calculating a gradient field of the gray level image, and separating the reference fingerprint image into a foreground area and a background area according to the gradient field, wherein the foreground area is a fingerprint area in the reference fingerprint image, and the background area is a blank area. And acquiring a gray level histogram of the foreground region, calculating data information such as a gray level mean value, a variance and the like of the foreground region according to the gray level histogram, and classifying the reference fingerprint image according to the data information. The fingerprint image is divided into a dry image, a wet image and a normal image, corresponding filtering processing is selected according to the image category of the reference fingerprint image to reduce noise influence, wherein the normal image is processed by mean value filtering, Gaussian curvature filtering and mean value filtering are selected for the dry image, histogram normalization processing is firstly carried out for the wet image, and then Gaussian curvature filtering and mean value filtering are carried out. And finally, calculating the curvature value extreme value difference value of the pixel points of the gray level image of the filtered reference fingerprint image, defining the curvature value extreme value difference value as the deep semantic expression information of the reference fingerprint image, and selecting the maximum value point from the deep semantic expression information as the characteristic point of the reference fingerprint image.
Step S20: calculating to obtain the description information of the matching points;
in the embodiment of the present invention, the specific process of calculating the description information of the matching point is to determine the description area according to the size of the preset area by taking the matching point as a center, for example, by taking one of the matching points as a center, determining a description area according to the size of the preset area by 10, and determining a piece of description area according to the reference direction of the matching point in the fingerprint. After the description area is determined, multi-stage block processing is performed on the description area to obtain a plurality of grid small blocks, the multi-stage block can be 2 x 2, 3 x 3, 4 x 4 and the like, 4, 9 and 16 grid small blocks can be correspondingly obtained, then pairwise pairing is performed on the grid small blocks, the grid block pairs are extracted to perform characteristic data difference calculation, and characteristic data difference information of the grid block pairs is obtained, namely description information of the description area. The characteristic data of the grid small blocks comprise a gray level mean value, an X-direction gradient mean value, a Y-direction gradient mean value and a curvature mean value.
For example, if the multi-level tile is 3 x 3, then all 9 grid tiles, then there are grid tile pairsAnd randomly extracting the grid block pairs with preset logarithms to perform characteristic data difference calculation to obtain characteristic data difference information of the multiple pairs of grid block pairs, namely the description information of the description area.
Step S30: compiling the description information into binary data according to a preset compiling rule to obtain a first descriptor;
in the embodiment of the invention, the description information compiling rule is preset. And compiling the description information according to a preset compiling rule, and converting the description information into a first descriptor which represents the description information by binary data. The compiling rule can be set in a user-defined mode according to actual application.
For example, the encoding rules may be defined as follows:
wherein, xValue represents the difference information, the variation range of xValue around 0 is coded as 00, minV and maxV represent the minimum value and the maximum value of the difference information xValue, the range code is 01, and the range code is 10. The compiling rule can adjust the compiling length according to the requirements of the system, the longer the compiling length is, the more the description information is, and simultaneously, the occupied memory and the calculated amount are correspondingly increased. The specific application can be adjusted in the following ways: the first mode is to reduce the number of the sampling grid block pairs, the number of the tested sampling grid block pairs is preferably more than 16, otherwise, because the description information is too little, the characteristic data of the description area cannot be fully embodied; in the second method, the number of feature data of each pair of grid block pairs is reduced, for example, only the difference information of the X-direction gradient mean and the Y-direction gradient mean is used to describe the difference between the pair of grid blocks, and if the number of sampling grid block pairs is 16, and the difference information only uses the X-direction gradient mean and the Y-direction gradient mean, the length of the coded information of the feature point description area is 16 × 2=64 bits of binary data, that is, 8 bytes.
Step S40: performing hash calculation on the first descriptor through a hash function to obtain a hash value of the first descriptor, and taking the hash value as a first index value of the first descriptor;
in the embodiment of the invention, the descriptor is subjected to hash calculation to obtain the hash value of the descriptor, and the hash value is used as the second index value of the descriptor. For example, the descriptor is 64-bit binary data, and may be hashed into 8-bit binary data by a hash function calculation, and the 8-bit binary data is used as an index value of the descriptor.
When the indexing mechanism is applied:
step T10: extracting a characteristic point of the fingerprint image to be compared, and building the characteristic point according to an index mechanism to obtain a second descriptor and a second index value;
step T20: screening out feature points corresponding to the first index value with the Hamming distance within a preset threshold value by calculating the Hamming distance between the first index value and the second index value to obtain a similar point set of the feature points;
in the embodiment of the present invention, a feature point is extracted from the fingerprint image to be compared according to the steps S10 to S40, a second descriptor and a second index value of the feature point are calculated, then a hamming distance between the first index value and the second index value is calculated, two points within a preset range according to a preset hamming distance are similar points, and a similar point set similar to the feature point is screened from matching points of the reference fingerprint image.
For example, two points corresponding to the index value with the preset hamming distance between 0 and 2 are similar points, the index value of the feature point a1 of the fingerprint image to be compared is 01010000, the hamming distances of a1 and b1 and a1 and b2 are respectively calculated by referring to the index value 01010001 of the index value b1 and the index value 01010111 of the match point b2 in the fingerprint image, and then the b1 is judged to be the similar point of a 1.
Step T30: judging whether the matching point corresponding to the closest Hamming distance and the feature point are matching point pairs or not according to the Hamming distance of the descriptor between the feature point and each matching point in the similar point set;
in the embodiment of the invention, after the similar point set similar to the feature points in the matching points is determined through the steps, the hamming distance of the descriptor corresponding to each matching point in the feature point set and the similar point set is respectively calculated, the nearest hamming distance and the next-nearest hamming distance in the similar point set are determined, then the ratio of the nearest hamming distance to the next-nearest hamming distance is calculated, and if the ratio is smaller than a preset threshold value, the matching point corresponding to the nearest hamming distance and the feature point are judged to be a pair of matching points.
For example, if the ratio of the minimum hamming distance to the next minimum hamming distance is within 0.8, it is determined that the matching point corresponding to the minimum hamming distance and the feature point are a pair of matching points, the descriptor of the feature point a1 is 00001111, the descriptor of the matching point b1 in the similarity point set is 01010111, the descriptor of the matching point b2 is 00000011, and the descriptor of the matching point b3 is 01010100, the hamming distances of a1, b1, b2, and b3 are 3, 2, and 5, respectively, the ratio of the minimum hamming distance to the next minimum hamming distance is 2/3=0.67, and is less than the preset ratio 0.8, and it is determined that a1 and b2 are a pair of matching points.
Step T40: continuously extracting n characteristic points from the fingerprint image to be compared, traversing each characteristic point of the image to be compared to execute the application of an indexing mechanism, and counting the logarithm of the matching points;
step S50: judging whether the logarithm of the matching points is greater than a preset logarithm threshold value or not;
step T60: if yes, the fingerprint image to be compared is judged to be similar to the reference fingerprint image.
In the embodiment of the invention, n feature points are extracted again from the image to be compared, matching points matched with the feature points are searched in the reference fingerprint image according to the application steps from T10 to T40, then whether the number of the matched point pairs reaches a preset threshold value or not is counted, and if the number of the matched point pairs reaches the preset threshold value, the fingerprint image to be compared is judged to be similar to the reference fingerprint image.
For example, if the number of the preset matched point pairs reaches 70 pairs, the fingerprint image of the comparison graph is considered to be similar to the reference fingerprint image. 100 feature points are extracted from the image to be compared currently, according to the application steps, the matching points of each feature point are indexed in the reference fingerprint image, and if the corresponding matching points can be matched in the 100 feature points and the number of the matched point pairs reaches 70 or more, the image to be compared is considered to be similar to the reference fingerprint image.
Referring to fig. 2, fig. 2 is a schematic view of a detailed flow of the step S20 in fig. 1. In an embodiment of the present invention, the step S20 includes:
step S201: determining a description area according to the size of a preset area by taking the matching point as a center;
step S202: performing multi-level blocking on the description area to obtain grid small blocks of the description area;
step S203: respectively calculating the characteristic data of each grid small block;
step S204: pairing the grid small blocks pairwise according to a preset pairing rule to obtain a grid block pair;
step S205: and extracting the grid block pair with a preset logarithm to calculate the characteristic data difference value of 2 grid small blocks in the grid block pair to obtain the description information.
In the embodiment of the present invention, the specific process of calculating the description information of the matching point is to determine the description area according to the size of the preset area and the reference direction of the matching point by taking the matching point as a center, and for example, determine a description area according to the size 10 × 10 of the preset area by taking one matching point with the reference direction of 30 ° as a center. After the description area is determined, multi-stage blocking processing is carried out on the description area to obtain a plurality of grid small blocks, the multi-stage blocking can be to divide the description area into 2 × 2, 3 × 3, 4 × 4 and the like, 4, 9 and 16 grid small blocks can be obtained correspondingly, then the grid small blocks are sequenced, the grid small blocks with corresponding preset serial numbers are paired, pairwise pairing is carried out on the grid small blocks according to the preset pairing serial numbers to obtain different grid block pairs, a certain number of grid block pairs are extracted from the grid block pairs, difference calculation is carried out on the characteristic data sequences of the 2 grid small blocks in the grid block pairs, then the characteristic data difference sequences of the grid block pairs are arranged to obtain the description information of the matching points. The characteristic data may be a gray level mean, an X-direction gradient mean, a Y-direction gradient mean, and a curvature mean.
For example, if the multi-level block is 3 × 3, all the multi-level blocks are 9 grid small blocks, a grid block pair exists, and a grid block pair with a preset logarithm is randomly extracted to perform feature data difference calculation, so that feature data difference information of the multiple pairs of grid block pairs is obtained, that is, description information of the description area.
Further, the feature data obtained by the calculation in step S203 includes a gray level mean, an X-direction gradient mean, a Y-direction gradient mean, and a curvature mean.
In the embodiment of the invention, the feature data can be properly reduced according to the requirement when being calculated. In order to improve the calculation efficiency, only 2 or 3 feature data may be calculated, so as to improve the establishment efficiency of the index establishment mechanism. For example, only the difference information of the X-direction gradient mean value and the Y-direction gradient mean value is used to describe the difference of the pair of lattice blocks.
Referring to fig. 3, fig. 3 is a detailed flowchart illustrating an embodiment of a compiling rule of the small-area fingerprint image fast indexing method according to the present invention. In the embodiment of the present invention, the compiling rule includes:
step S01: presetting a variation range of the description information near 0;
step S02: extracting a maximum value and a minimum value from the description information, and obtaining different value ranges of the description information by combining the change range;
step S03: and compiling the description information into corresponding binary data according to the preset description information corresponding to different binary data in different value ranges.
In the embodiment of the invention, the variation range of the description information near 0 is determined based on the test data, the value range of the description information is determined based on the variation range by combining the maximum value and the minimum value determined from the difference sequence corresponding to the description information, and the binary data is compiled according to the determined value range and a preset compiling rule. For example, currently, a variation range (-5, 5) of the descriptor around 0 is determined based on the test data, the maximum value in the difference sequence of the descriptor is 10, the minimum value is-9, the value ranges of the descriptor are (-9, -5), (-5, 5) and (5, 10), the differences of the difference sequence corresponding to the descriptor are set to be compiled into 00,01,10 in the 3 value ranges, respectively, and the binary data compiled for the descriptor of { -6,1,7,0} is 00011001. It should be noted that, in order to improve the accuracy, in this embodiment, a value range, such as (-9, -7) or (-5, 0) or (0, 5), may be further divided from 3 value ranges, so that the description information in the value range is compiled into 11, so that the description information expressed by the descriptor is finer, and the comparison details are improved.
Referring to fig. 4, fig. 4 is a schematic view of a detailed flow of the step S204 in fig. 2. In the embodiment of the present invention, step S204 includes:
step S241: sequencing all grid small blocks to obtain the serial number of each grid small block;
step S242: and according to the preset matching sequence number, matching the 2 grid small blocks corresponding to the matching sequence number to obtain a grid block pair.
In the embodiment of the present invention, the grid tiles are sorted according to the multi-level partitions of the description area, and if the multi-level partitions of the description area are 3 × 3 or 4 × 4, the total number of the grid tiles is 9 or 16, and the sorting may be 1 to 9 or 1 to 16. Taking the total number of grid small blocks as 9 for example, two pairs of the 9 grid small blocks are paired to form a pair of grid blocks, and the pairing serial numbers of the preset grid block pairs are No. 1 and No. 2, No. 2 and No. 3, No. 3 and No. 4, and the like.
Referring to fig. 5, fig. 5 is a schematic view of a detailed flow of the step S205 in fig. 2. In the embodiment of the present invention, step S205 includes:
step S251: extracting corresponding logarithm grid block pairs according to the preset logarithm of extracted grid block pairs;
step S252: carrying out differential processing on the characteristic data of 2 grid small blocks in each pair of grid blocks to obtain a difference value sequence of the characteristic data of each pair of grid small blocks;
step S253: and arranging and combining the difference value sequences to obtain the description information.
In the embodiment of the invention, the number 18 of pairs of extracted grid blocks is preset, 18 pairs of grid blocks are extracted, the characteristic data of 2 grid small blocks in each pair of grid blocks are subjected to differential processing, if the serial numbers of the paired grid small blocks in a pair of grid block pairs are 1 and 2, the characteristic data sequence of the grid small block 1 is {5,6,8,10}, the characteristic data sequence of the grid small block 2 is {4,7,9,8}, and the characteristic data sequences of the grid small block 1 and the grid small block 2 in the pair of grid block pairs are subjected to differential processing to obtain the difference sequence {1, -1, -1,2} of the characteristic data of the pair of grid blocks. If the sequence of the difference values of the plurality of pairs of grid blocks obtained after calculation is {1, -1, -1,2}, {1,2, -1,2}, {1, -1,2, -1}, { -1,1, -1,2}, {1, -1, -1,1}, {1, -1,2, -1}, { -1,1, -1,2}, {3, -1, -1,2}, and are arranged to obtain description information {1, -1, -1,2}, {1,2, -1,2} {1, -1,2,2}, {1, -1,2, -1}, {1, -1, -1,1},1, 2, -1}{ -1,1, -1,2}{3, -1, -1,2}.
Referring to fig. 6, fig. 6 is a schematic view of a detailed flow of an embodiment of step T30 in fig. 1. In the embodiment of the present invention, step T30 includes:
step T301: calculating the Hamming distance between the first descriptor and the second descriptor corresponding to the matching points in the similar point set, and screening out the nearest Hamming distance and the next nearest Hamming distance;
step T302: judging whether the ratio of the nearest Hamming distance to the next nearest Hamming distance is within a preset threshold value;
step T303: if yes, judging that the matching point of the descriptor corresponding to the closest Hamming distance and the feature point are a pair of matching point pairs.
In the embodiment of the invention, after the similar point set similar to the feature points in the matching points is determined through the steps, the hamming distance of the descriptor corresponding to each matching point in the feature point set and the similar point set is respectively calculated, the nearest hamming distance and the next-nearest hamming distance in the similar point set are determined, then the ratio of the nearest hamming distance to the next-nearest hamming distance is calculated, and if the ratio is smaller than a preset threshold value, the matching point corresponding to the nearest hamming distance and the feature point are judged to be a pair of matching points.
For example, if the ratio of the minimum hamming distance to the next minimum hamming distance is within 0.8, it is determined that the matching point corresponding to the minimum hamming distance and the feature point are a pair of matching points, the descriptor of the feature point a1 is 00001111, the descriptor of the matching point b1 in the similarity point set is 01010111, the descriptor of the matching point b2 is 00000011, and the descriptor of the matching point b3 is 01010100, the hamming distances of a1, b1, b2, and b3 are 3, 2, and 5, respectively, the ratio of the minimum hamming distance to the next minimum hamming distance is 2/3=0.67, and is less than the preset ratio 0.8, and it is determined that a1 and b2 are a pair of matching points.
Referring to fig. 7, fig. 7 is a functional model diagram of an embodiment of the small-area fingerprint image fast indexing device of the present invention. In this embodiment, the small-area fingerprint image fast indexing device includes:
a first extraction module 10, configured to extract a preset number of matching points from a reference fingerprint image, where the extracted points at least include coordinate information and reference direction information;
a calculating module 20, configured to calculate and obtain description information of the matching point;
the compiling module 30 is configured to compile the description information into binary data according to a preset compiling rule, so as to obtain a first descriptor;
the processing module 40 is configured to perform hash calculation on the first descriptor through a hash function to obtain a hash value of the first descriptor, and use the hash value as a first index value of the first descriptor;
the second extraction module 50 is configured to extract a feature point of the fingerprint image to be compared, and obtain a second descriptor and a second index value through processing of the calculation module, the compiling module and the processing module; the screening module 60 is configured to screen out feature points corresponding to the first index value with the hamming distance within a preset distance threshold by calculating the hamming distance between the first index value and the second index value, so as to obtain a similar point set of the feature points;
a first judging module 70, configured to judge whether a matching point corresponding to a closest hamming distance and each matching point in the similar point set are matching point pairs according to the hamming distance of a descriptor between the feature point and the matching point;
the counting module 80 is used for continuously extracting n characteristic points from the fingerprint image to be compared, traversing each characteristic point of the image to be compared, and counting the logarithm of the matching point through the processing of the calculating module, the compiling module, the processing module, the screening module and the first judging module;
a second judging module 90, configured to judge whether the logarithm of the matching point is greater than a preset logarithm threshold;
the determining module 100 is configured to determine that the fingerprint image to be compared is similar to the reference fingerprint image if the number of matching point pairs is greater than a preset logarithm threshold.
The invention provides a small-area fingerprint image fast indexing device which is characterized by comprising a memory, a processor and a small-area fingerprint image fast indexing program which is stored on the memory and can run on the processor, wherein the small-area fingerprint image fast indexing program realizes the steps of the small-area fingerprint image fast indexing method in the embodiment when being executed by the processor.
The invention also provides a computer storage medium.
In this embodiment, the computer storage medium stores a small-area fingerprint image fast indexing program, and the small-area fingerprint image fast indexing program, when executed by a processor, implements the steps of the small-area fingerprint image fast indexing method as described in any one of the above embodiments.
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In addition, the term "and/or" herein is only one kind of association relationship describing an associated object, and means that there may be three kinds of relationships, for example, a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" herein generally indicates that the former and latter related objects are in an "or" relationship, unless otherwise specified.
Claims (9)
1. A small-area fingerprint image fast indexing method is characterized by comprising the following steps:
when the indexing mechanism is established:
extracting a preset number of matching points from the reference fingerprint image, wherein the extracted matching points at least comprise coordinate information and reference direction information;
calculating to obtain the description information of the matching points;
compiling the description information into binary data according to a preset compiling rule to obtain a first descriptor;
performing hash calculation on the first descriptor through a hash function to obtain a hash value of the first descriptor, and taking the hash value as a first index value of the first descriptor;
when the indexing mechanism is applied:
extracting a characteristic point of the fingerprint image to be compared, and building the characteristic point according to an index mechanism to obtain a second descriptor and a second index value;
screening out matching points corresponding to the first index value with the Hamming distance within a preset distance threshold value by calculating the Hamming distance between the first index value and the second index value to obtain a similar point set of the feature points;
judging whether the matching point corresponding to the closest Hamming distance and the feature point are matching point pairs or not according to the Hamming distance of the descriptor between the feature point and each matching point in the similar point set;
continuously extracting n characteristic points from the fingerprint image to be compared, traversing each characteristic point of the image to be compared to execute the application of an indexing mechanism, and counting the logarithm of the matching points;
judging whether the logarithm of the matching points is greater than a preset logarithm threshold value or not;
if yes, judging that the fingerprint image to be compared is similar to the reference fingerprint image;
wherein the calculating to obtain the description information of the matching point comprises:
determining a description area according to the size of a preset area by taking the matching point as a center;
performing multi-level blocking on the description area to obtain grid small blocks of the description area;
respectively calculating the characteristic data of each grid small block;
pairing the grid small blocks pairwise according to a preset pairing rule to obtain a grid block pair;
and extracting the grid block pair with a preset logarithm to calculate the characteristic data difference value of 2 grid small blocks in the grid block pair to obtain the description information.
2. The method for rapidly indexing a small-area fingerprint image according to claim 1, wherein the feature data comprises a gray-scale mean, an X-direction gradient mean, a Y-direction gradient mean, and a curvature mean of the grid patches.
3. The method for rapidly indexing a small-area fingerprint image according to claim 1, wherein the compiling rule comprises:
determining a variation range of the description information around 0 based on the test data;
extracting a maximum value and a minimum value from the description information, and obtaining different value ranges of the description information by combining the change range;
and compiling the description information into corresponding binary data according to the preset description information corresponding to different binary data in different value ranges.
4. The method for rapidly indexing a small-area fingerprint image according to claim 1, wherein pairwise pairing of grid patches according to a preset pairing rule to obtain a grid patch pair comprises:
sequencing all grid small blocks to obtain the serial number of each grid small block;
and according to the preset matching sequence number, matching the 2 grid small blocks corresponding to the matching sequence number to obtain a grid block pair.
5. The method for fast indexing a small-area fingerprint image according to claim 1, wherein the extracting a preset logarithm of a grid block pair to calculate a difference value of feature data of 2 grid small blocks in the grid block pair to obtain the description information comprises:
extracting corresponding logarithm grid block pairs according to the preset logarithm of extracted grid block pairs;
carrying out differential processing on the characteristic data of 2 grid small blocks in each pair of grid blocks to obtain a difference value sequence of the characteristic data of each pair of grid small blocks;
and arranging and combining the difference value sequences to obtain the description information.
6. The method for rapidly indexing a small-area fingerprint image according to claim 1, wherein the determining whether the matching point corresponding to the closest hamming distance and the feature point are matching point pairs according to the hamming distance of the descriptor between the feature point and each matching point in the set of similar points comprises:
calculating the Hamming distance between the first descriptor and the second descriptor corresponding to the matching points in the similar point set, and screening out the nearest Hamming distance and the next nearest Hamming distance;
judging whether the ratio of the nearest Hamming distance to the next nearest Hamming distance is within a preset threshold value;
if yes, judging that the matching point corresponding to the closest Hamming distance and the feature point are a pair of matching point pairs.
7. A small-area fingerprint image fast indexing device is characterized by comprising:
the first extraction module is used for extracting a preset number of matching points from the reference fingerprint image, and the extracted matching points at least comprise coordinate information and reference direction information;
the calculation module is used for calculating and obtaining the description information of the matching points;
the compiling module is used for compiling the description information into binary data according to a preset compiling rule to obtain a first descriptor;
the processing module is used for carrying out hash calculation on the first descriptor through a hash function to obtain a hash value of the first descriptor, and the hash value is used as a first index value of the first descriptor;
the second extraction module is used for extracting a characteristic point of the fingerprint image to be compared, and obtaining a second descriptor and a second index value through the processing of the calculation module, the compiling module and the processing module;
the screening module is used for screening out matching points corresponding to the first index value with the Hamming distance within a preset distance threshold value by calculating the Hamming distance between the first index value and the second index value to obtain a similar point set of the feature points;
the first judgment module is used for judging whether the matching point corresponding to the closest Hamming distance and the feature point are matching point pairs or not according to the Hamming distance of the descriptor between the feature point and each matching point in the similar point set;
the statistical module is used for continuously extracting n characteristic points from the fingerprint image to be compared, traversing each characteristic point of the image to be compared, and counting the logarithm of the matching points through the processing of the calculation module, the compiling module, the processing module, the screening module and the first judgment module;
the second judgment module is used for judging whether the logarithm of the matching points is greater than a preset logarithm threshold value;
and the judging module is used for judging that the fingerprint image to be compared is similar to the reference fingerprint image if the matching point number is greater than the preset logarithm threshold value.
8. A small-area fingerprint image fast indexing device, characterized in that the small-area fingerprint image fast indexing device comprises a memory and a processor and a small-area fingerprint image fast indexing program stored on the memory and executable on the processor, the small-area fingerprint image fast indexing program when executed by the processor implementing the steps of the small-area fingerprint image fast indexing method according to any one of claims 1 to 6.
9. A computer storage medium, characterized in that the computer storage medium has stored thereon a small-area fingerprint image fast indexing program which, when executed by a processor, implements the steps of the small-area fingerprint image fast indexing method according to any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010526646.9A CN111428064B (en) | 2020-06-11 | 2020-06-11 | Small-area fingerprint image fast indexing method, device, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010526646.9A CN111428064B (en) | 2020-06-11 | 2020-06-11 | Small-area fingerprint image fast indexing method, device, equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111428064A CN111428064A (en) | 2020-07-17 |
CN111428064B true CN111428064B (en) | 2020-09-29 |
Family
ID=71551451
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010526646.9A Active CN111428064B (en) | 2020-06-11 | 2020-06-11 | Small-area fingerprint image fast indexing method, device, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111428064B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113705519B (en) * | 2021-09-03 | 2024-05-24 | 杭州乐盯科技有限公司 | Fingerprint identification method based on neural network |
CN115455218A (en) * | 2022-06-09 | 2022-12-09 | 中国公路工程咨询集团有限公司 | Road image distributed storage method, search method and device |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104268602A (en) * | 2014-10-14 | 2015-01-07 | 大连理工大学 | Shielded workpiece identifying method and device based on binary system feature matching |
DE102016005636A1 (en) * | 2015-06-08 | 2016-12-22 | Cross Match Technologies, Inc. | Transformed representation of fingerprint data with high recognition accuracy |
CN108182205A (en) * | 2017-12-13 | 2018-06-19 | 南京信息工程大学 | A kind of image method for quickly retrieving of the HASH algorithms based on SIFT |
CN108664940B (en) * | 2018-05-16 | 2020-02-18 | 山东大学 | Partial fingerprint matching method and system |
CN109766850B (en) * | 2019-01-15 | 2021-06-01 | 西安电子科技大学 | Fingerprint image matching method based on feature fusion |
CN109993129B (en) * | 2019-04-04 | 2022-10-18 | 郑州师范学院 | Fingerprint identification method based on fingerprint thin node cylindrical code |
CN110704651A (en) * | 2019-10-18 | 2020-01-17 | 哈尔滨理工大学 | Rapid fingerprint database retrieval method based on core detail node support system |
-
2020
- 2020-06-11 CN CN202010526646.9A patent/CN111428064B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN111428064A (en) | 2020-07-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111429359B (en) | Small-area fingerprint image splicing method, device, equipment and storage medium | |
Wang et al. | Application of ReliefF algorithm to selecting feature sets for classification of high resolution remote sensing image | |
CN104463141B (en) | A kind of fingerprint template input method and device based on smart card | |
WO2011044058A2 (en) | Detecting near duplicate images | |
CN106203539B (en) | Method and device for identifying container number | |
CN111428064B (en) | Small-area fingerprint image fast indexing method, device, equipment and storage medium | |
CN101114335A (en) | Full angle rapid fingerprint identification method | |
CN101751550A (en) | Fast fingerprint searching method and fast fingerprint searching system thereof | |
CN109325507A (en) | A kind of image classification algorithms and system of combination super-pixel significant characteristics and HOG feature | |
CN111428701B (en) | Small-area fingerprint image feature extraction method, system, terminal and storage medium | |
CN114359632A (en) | Point cloud target classification method based on improved PointNet + + neural network | |
CN110866136B (en) | Face image stacking method and device, electronic equipment and readable storage medium | |
CN104881668B (en) | A kind of image fingerprint extracting method and system based on representative local mode | |
CN115600194A (en) | Intrusion detection method, storage medium and device based on XGboost and LGBM | |
EP1930852B1 (en) | Image search method and device | |
CN109544614B (en) | Method for identifying matched image pair based on image low-frequency information similarity | |
JP4802176B2 (en) | Pattern recognition apparatus, pattern recognition program, and pattern recognition method | |
CN108694411B (en) | Method for identifying similar images | |
CN116415210A (en) | Image infringement detection method, device and storage medium | |
CN106326927A (en) | Shoeprint new class detection method | |
CN110705569A (en) | Image local feature descriptor extraction method based on texture features | |
CN111931229B (en) | Data identification method, device and storage medium | |
CN115527241A (en) | Fingerprint template updating method and device, embedded equipment and storage medium | |
CN114972540A (en) | Target positioning method and device, electronic equipment and storage medium | |
CN110619273B (en) | Efficient iris recognition method and recognition device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |