WO2018213946A1 - 图像识别方法、图像识别装置、电子装置及计算机存储介质 - Google Patents

图像识别方法、图像识别装置、电子装置及计算机存储介质 Download PDF

Info

Publication number
WO2018213946A1
WO2018213946A1 PCT/CN2017/085233 CN2017085233W WO2018213946A1 WO 2018213946 A1 WO2018213946 A1 WO 2018213946A1 CN 2017085233 W CN2017085233 W CN 2017085233W WO 2018213946 A1 WO2018213946 A1 WO 2018213946A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
feature
feature information
information
fingerprint
Prior art date
Application number
PCT/CN2017/085233
Other languages
English (en)
French (fr)
Inventor
李其昌
Original Assignee
深圳信炜科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳信炜科技有限公司 filed Critical 深圳信炜科技有限公司
Priority to PCT/CN2017/085233 priority Critical patent/WO2018213946A1/zh
Priority to CN201780000357.1A priority patent/CN107278308A/zh
Publication of WO2018213946A1 publication Critical patent/WO2018213946A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • 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
    • G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • 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
    • G06V40/1347—Preprocessing; Feature extraction
    • 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

Definitions

  • the present invention relates to the field of biometrics, and in particular, to an image recognition method, an image recognition device, an electronic device, and a computer storage medium.
  • Biometric technology especially fingerprint recognition, is widely used in smart terminals such as mobile phones. Since the setting position of the smart terminal is limited, when the biological image sensing is performed, the larger the sensing area is, the higher the cost is. Therefore, the image sensing area has a certain limit, and fingerprint identification is taken as an example, and the commonly used sensing area is currently used. It is 25% to 30% of the entire fingerprint area of the finger.
  • the sensing area is small, if the recognition accuracy is to be ensured, it is necessary to collect the registration multiple times.
  • the image information to be identified needs to be compared with all the registered templates to obtain the recognition result. In this way, image recognition takes a long time and is not conducive to user experience.
  • embodiments of the present invention aim to at least solve one of the technical problems existing in the prior art. To this end, embodiments of the present invention need to provide an image recognition method, an image recognition device, an electronic device, and a computer storage medium.
  • performing the feature extraction on the acquired image information in the step S2, the obtaining the first feature information specifically includes:
  • the initial coding features are classified according to the preset coding feature class, and the number of each class is counted.
  • the final coding feature information is included in the initial coding features.
  • the image encoding employs one or more of the following encoding methods: local binary mode, local phase quantization, image feature value statistics.
  • the feature classification uses one or more of the following clustering methods: k-means, hierarchical clustering, self-organizing mapping, fuzzy C-means clustering.
  • the preset coding feature core is preset or updated in real time by means of an online manner.
  • the step S4 is performed in parallel with the step S3 after obtaining the aligned image template.
  • the recognition process is ended, and the recognition result is output.
  • step S3 when the first feature information and the first feature information in all the registered image templates are inconsistent, the recognition process is ended, and the recognition result is output.
  • the method before the step S2, the method further includes:
  • step S04 determining whether the number of registration acquisitions reaches a preset threshold; if yes, executing step S05, otherwise returning to step S01;
  • step S05 the method includes:
  • the image information of the target object is collected by an image acquisition unit, and the image information to be recognized is identified by the steps in the image recognition method of any of the foregoing embodiments.
  • An image recognition apparatus includes a processor and a storage unit; and the processor is configured to execute the image recognition method of any of the foregoing embodiments.
  • An electronic device includes an image capturing device and a processor; the image capturing device is configured to acquire image information of the target object when the target object is placed on the electronic device; and the processor is configured to execute The image recognition method of any of the above embodiments.
  • a computer storage medium stores computer executable instructions for performing the image recognition method of any of the foregoing embodiments.
  • the first feature information and the second feature information can be extracted in parallel, which further improves the fingerprint recognition speed.
  • the comparison of the first feature information may be performed first, and after the second feature information is extracted, the feature of the second feature information is performed in parallel. The comparison further improves the speed of fingerprint recognition.
  • FIG. 1 is a schematic diagram of fingerprint registration using fingerprint recognition as an example in image recognition according to an embodiment of the present invention
  • FIG. 2 is a schematic diagram of fingerprint recognition taking fingerprint recognition as an example in image recognition according to an embodiment of the present invention
  • FIG. 3 is a schematic flowchart of a registration process of an image recognition method according to an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of a fingerprint template stored in an image recognition method according to an embodiment of the present invention.
  • FIG. 5 is a schematic flowchart diagram of an embodiment of an identification process of an image recognition method according to an embodiment of the present invention
  • FIG. 6 is a schematic flow chart of another embodiment of an identification process of an image recognition method according to an embodiment of the present invention.
  • FIG. 7 is a flow chart showing still another embodiment of the identification process of the image recognition method according to the embodiment of the present invention.
  • FIG. 8 is a functional block diagram of an image recognition device according to an embodiment of the present invention.
  • FIG. 9 is a schematic diagram of comparison of feature information and fingerprint template in feature matching in an image recognition apparatus according to an embodiment of the present invention.
  • FIG. 10 is a functional block diagram of a first feature extraction unit in the image recognition device according to the embodiment of the present invention.
  • FIG. 11 is a schematic diagram showing a plain fingerprint image and a coding feature distribution obtained by encoding the same according to BSIF encoding in the image recognition apparatus according to the embodiment of the present invention
  • FIG. 12 is a fingerprint image of a crepe pattern in an image recognition apparatus according to an embodiment of the present invention and encoded in accordance with BSIF Schematic diagram of the distribution of coding features obtained after encoding;
  • FIG. 13 is a plan view showing an embodiment of an electronic device to which an image recognition device according to an embodiment of the present invention is applied;
  • Fig. 14 is a functional block diagram showing another embodiment of an electronic device to which the image recognition device according to the embodiment of the present invention is applied.
  • first and second are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated.
  • features defining “first” or “second” may include one or more of the described features either explicitly or implicitly.
  • the meaning of "a plurality" is two or more unless specifically and specifically defined otherwise.
  • connection In the description of the present invention, it should be noted that the terms “installation”, “connected”, and “connected” are to be understood broadly, and may be fixed or detachable, for example, unless otherwise explicitly defined and defined. Connected, or integrally connected; may be mechanically connected, or may be electrically connected or may communicate with each other; may be directly connected or indirectly connected through an intermediate medium, may be internal communication of two elements or interaction of two elements relationship. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood on a case-by-case basis.
  • the image recognition in the embodiment of the present invention refers to image acquisition of a target object placed at a specified position by using an image sensor, and the collected image is transmitted to a processing system through a flexible printed circuit board or other printed circuit for image recognition, that is, Identify whether the image is registered or registered, or identify the user's identity, and so on.
  • the fingerprint image of the user must be registered before the fingerprint image is recognized.
  • the user places the finger 200 that needs to be registered in the designated position.
  • the fingerprint collecting unit 103 collects the fingerprint of the finger to obtain a fingerprint image.
  • the feature extraction unit 101 performs fingerprint feature extraction on the fingerprint image obtained by the fingerprint collection unit 103, obtains fingerprint feature information, and forms a corresponding fingerprint template, and stores it to a predetermined location of the storage unit 102. This is repeated a number of times to form N corresponding fingerprint templates. It should be noted that, since the sensing area of the fingerprint collecting unit 103 is small, the same fingerprint is collected multiple times to form complete fingerprint information. Therefore, assuming that one finger is collected five times and five fingers are collected, a total of 25 fingerprint templates are generated.
  • the finger 200 When the user needs to perform fingerprint recognition, the finger 200 is placed at the designated position, and the fingerprint collection unit 103 collects the fingerprint of the finger to obtain a fingerprint image.
  • the feature extraction unit 101 performs fingerprint feature extraction on the fingerprint extraction unit 101 to obtain fingerprint feature information to be compared.
  • the feature comparison unit 104 compares the fingerprint feature information to be compared with all the fingerprint templates in the storage unit 102, and outputs a comparison result, such as a successful recognition or a recognition failure.
  • the fingerprint collection unit 103 herein may include any one or more of an optical fingerprint collection unit, a capacitive fingerprint collection unit, an ultrasonic fingerprint collection unit, a radio frequency fingerprint collection unit, and the like.
  • the fingerprint feature information to be compared needs to be compared with all the fingerprint templates in the storage unit 102, the recognition result can be obtained. Therefore, the more the number of fingerprint templates, the longer the fingerprint recognition time is, so that the fingerprint recognition is performed. The speed is slow, which is not conducive to the user experience.
  • the present invention proposes a new image recognition method.
  • the image recognition method performs feature extraction on the image feature extraction according to different feature extraction rules to obtain different feature information, such as first feature information, second feature information, etc.; when the feature information is compared, First comparing the first feature information with the first feature information in the image registration information, if the comparison is consistent, comparing the second feature information with the second feature information in the registration information, and finally outputting the comparison result . Since the first feature information includes only a small number of features of the image, and the second feature information includes all features of the image, the comparison speed of the first feature information is faster than the second feature information, that is, the image recognition method is improved. Image recognition speed.
  • the image information may include fingerprint information, palm print information, ear pattern information, and skin texture information at other suitable locations on the living body. Since the skin texture and cortical structure of the organism have different characteristics, different organisms can be identified according to the characteristics.
  • the living body is, for example, a human body, but may not be limited to a human body. In the following embodiments, fingerprint recognition will be described as an example.
  • An image recognition method of an embodiment of the present invention may include a registration process and an identification process.
  • the registration process the user registers his or her own image information, such as fingerprint information, which will be stored in the form of a fingerprint template for use in the identification process.
  • fingerprint information such as fingerprint information
  • the pattern template performs feature comparison and outputs recognition results, such as recognition success or recognition failure.
  • the registration process specifically includes:
  • Step S101 acquiring a fingerprint image to be registered
  • the fingerprint image to be registered is acquired by the image sensor.
  • the fingerprint sensor Taking the fingerprint sensor as an example, when the user's finger is located on the fingerprint sensor, the fingerprint sensor will collect the fingerprint image of the finger and transmit the collected fingerprint image to the fingerprint recognition system.
  • the input interface 111 of the fingerprint recognition system 100 can be electrically connected to the fingerprint sensor through a flexible printed circuit board or other circuit board.
  • Step S102 performing feature extraction on the acquired fingerprint image, and acquiring first feature information and second feature information;
  • the first feature information and the second feature information reflect the fingerprint feature from different angles, and the feature information amount of the first feature information is less than the feature information amount of the second feature information. Therefore, feature extraction is performed on the acquired fingerprint image according to different feature extraction rules, and the first feature information and the second feature information are obtained.
  • the extraction of the first feature information and the extraction of the second feature information may be performed sequentially or in parallel. Moreover, the extraction of feature information in parallel can speed up feature extraction.
  • the first feature information is used to describe the fingerprint feature of the user as a whole, and only contains a small amount of feature information describing the characteristics of the user's fingerprint.
  • a human finger fingerprint may include three basic types: a concentric circle or a spiral line, which is called a bucket line; an opening on one side of the line, called a striated line; and a striate like a bow, called a bow line.
  • different types of fingerprints also differ in the degree of curvature of the texture, the width of the interval of the texture, and the like. Therefore, when the first feature information is extracted, the type of the fingerprint, the degree of curvature of the texture, the interval width of the texture, and the like can be extracted.
  • the second feature information is used to accurately describe the uniqueness of the fingerprint, and then includes all the feature information of the fingerprint information, such as the texture of the entire fingerprint. Therefore, when the second feature information extraction is performed, the texture information of the entire fingerprint is extracted.
  • first feature information and the second feature information are not limited herein.
  • the third feature information, the fourth feature information, and the like may also be extracted as needed.
  • Step S103 Form a corresponding fingerprint template according to the first feature information and the second feature information; that is, each fingerprint template includes corresponding first feature information and second feature information.
  • Step S104 determining whether the number of registration acquisitions reaches a preset threshold; if yes, executing step S105, otherwise returning to step S101;
  • the acquisition is set at least 3 times, that is, the preset threshold is 3. It should be noted that each acquisition is effective, that is, the fingerprint image acquired each time should meet the requirements of definition, finger scanning position and the like. If the currently collected fingerprint image does not meet the requirements, for example, the finger movement causes the captured image to be blurred, the current captured image is discarded, and the user is prompted to re-perform the fingerprint image. collection.
  • step S105 all the fingerprint templates are stored.
  • all fingerprint templates formed by the registration are stored.
  • the difference between the fingerprint templates generated when the same finger is registered is small, when storing, the same portion in each fingerprint template and a portion having a difference between the fingerprint templates may be stored. This will save storage space.
  • all the fingerprint templates contain only feature information, and the corresponding fingerprint image cannot be obtained through the feature information, thereby ensuring the information security of the fingerprint template.
  • each registered finger has a corresponding fingerprint template library, that is, each fingerprint template has a corresponding finger identifier, and each corresponding fingerprint template library has multiple fingerprint templates.
  • the fingerprint template corresponding to the finger A includes the template 1A, the template 2A, and the template 3A.
  • the registration process of the above image is not limited to the process described in the foregoing embodiment, and may also include other manners in order to obtain a fingerprint template including the first feature information and the second feature information.
  • the image template in the registration process it can be repeatedly used for the image recognition process, thereby saving each image registration process.
  • the identifying process may specifically include:
  • Step S201 acquiring a fingerprint image to be identified
  • the fingerprint image to be identified is acquired by the image sensor.
  • the fingerprint sensor when the user's finger is located on the fingerprint sensor, the fingerprint sensor will collect the fingerprint image of the finger and output the collected fingerprint image to the fingerprint recognition system.
  • the input interface 111 of the fingerprint recognition system 100 can be electrically connected to the fingerprint sensor through a flexible printed circuit board or other circuit board.
  • Step S202 performing feature extraction on the acquired fingerprint image to obtain first feature information and second feature information
  • the first feature information and the second feature information reflect the fingerprint feature from different angles, and the feature information amount of the first feature information is less than the feature information amount of the second feature information. Therefore, feature extraction is performed on the acquired fingerprint image according to different feature extraction rules, and the first feature information and the second feature information are obtained.
  • the extraction of the first feature information and the extraction of the second feature information may be performed sequentially or in parallel. Moreover, the extraction of feature information in parallel can speed up feature extraction.
  • the first feature information is used to describe the fingerprint feature of the user as a whole, and only includes describing the fingerprint feature of the user.
  • a small amount of feature information For example, a human finger fingerprint may include three basic types: a concentric circle or a spiral line, which is called a bucket line; an opening on one side of the line, called a striated line; and a striate like a bow, called a bow line.
  • different types of fingerprints also differ in the degree of curvature of the texture, the width of the interval of the texture, and the like. Therefore, when the first feature information is extracted, the type of the fingerprint, the degree of curvature of the texture, the interval width of the texture, and the like can be extracted.
  • the second feature information is used to accurately describe the uniqueness of the fingerprint, and then includes all the feature information of the fingerprint information, such as the texture of the entire fingerprint. Therefore, when the second feature information extraction is performed, the texture information of the entire fingerprint is extracted.
  • Step S203 comparing the first feature information with the first feature information of all the registered fingerprint templates to obtain a fingerprint template that is consistent with the comparison;
  • Step S204 comparing the second feature information with the second feature information in the fingerprint template that is aligned, to obtain a comparison result.
  • step S204 is performed.
  • step S204 is performed in parallel, and after the second feature information is successfully matched, the identification may be stopped, so that the comparison result may be obtained in advance, and the reference may be accelerated. Identify the speed.
  • the first feature information is coded feature information obtained by performing coded clustering processing on the fingerprint image.
  • the first feature information in the fingerprint template is coded feature information.
  • the identification process may specifically include:
  • Step S301 acquiring a fingerprint image to be identified
  • Step S302 performing feature extraction on the acquired fingerprint image to obtain coding feature information and second feature information
  • Step S303 comparing the coded feature information with the coded feature information in all the registered fingerprint templates to obtain a fingerprint template that is consistent in comparison
  • Step S304 comparing the second feature information with the second feature information in the fingerprint template that is aligned, to obtain a comparison result.
  • the process of obtaining the encoded feature information may include: performing image coding on the fingerprint image to obtain an initial coding feature; classifying the initial coding feature, and counting the number of each class to form a final coding feature information.
  • the encoding method may include Local Binary Pattern (LBP), Local Phase Quantization (LPQ), and Binaryized Statistical Image Features (BSIF).
  • LBP Local Binary Pattern
  • LPQ Local Phase Quantization
  • BSIF Binaryized Statistical Image Features
  • Feature categorization methods may include k-means (k-means), hierarchical clustering, Self-Organizing Maps (SOM), Fuzzy C-means (FCM) clustering, etc., and the number of clusters may be For fixed values, of course, you can also flexibly set according to the actual situation.
  • the initial coding feature is compared with the preset coding feature class core by using a distance metric or the like, and the coding feature is merged to the class with the smallest distance. Then count the number of each class, and finally get a statistical histogram, the final coding feature.
  • the preset coding feature class core includes a cluster center value or a cluster mean value of the feature, a feature range value of the cluster, and the like.
  • the preset coding feature core is pre-computed and saved to the image recognition device for reuse in feature classification.
  • the preset coding feature class core can also be updated in real time through an online manner to improve the previously set coding feature class.
  • the preset coding feature class core can also be obtained by processing the registered fingerprint template. For example, the fingerprint image in the registered fingerprint template is image-encoded to obtain a coding feature; then the feature features are clustered to obtain a coding feature core.
  • the encoded feature core is obtained by using the registered fingerprint template processing
  • the coding feature obtained by performing feature classification statistics according to the coding feature class is closer to the coding feature of the registered fingerprint template, so that when the feature comparison is performed, not only the acceleration is accelerated. Compare the speed and ensure the accuracy of the comparison.
  • the method may further include:
  • Step S205 preprocessing the acquired fingerprint image.
  • the fingerprint image may be subjected to certain preprocessing, for example, noise processing of the fingerprint image, enhancement processing of the fingerprint image, and the like. If there is noise in the fingerprint image, there will be burrs when extracting features, and there will be many false minutiae points, that is, pseudo-detail points.
  • the noise processing of the image may include processing of the spatial domain and processing of the frequency domain, such as averaging or intermediate values of image pixels; frequency domain processing such as low pass filtering.
  • the enhancement processing is mainly for highlighting the contrast between the ridge portion and the valley portion of the fingerprint image, for example, by adjusting the gray value of the image pixel, sharpening the fingerprint image, and increasing the outline of the fingerprint image and the sharpness of the line.
  • adjusting the gray value of the image pixel sharpening the fingerprint image
  • increasing the outline of the fingerprint image and the sharpness of the line there are other ways to deal with it, not to mention here.
  • an image recognition apparatus 100 is used to implement the method steps of the above embodiment.
  • the image recognition device 100 includes a processing circuit 110, a storage unit 120, a feature extraction unit 130, and a feature comparison unit 140.
  • Processing circuit 110 has, for example but not limited to, a microprocessor having an input interface 111 and an output interface 112.
  • the input interface 111 is configured to receive image information collected by an image acquisition unit (not shown), such as fingerprint image information collected by the fingerprint collection unit.
  • the fingerprint image information received by the input interface 111 may be an analog signal or a digital signal. If the fingerprint image information received by the input interface 111 is an analog signal, when it is sent to the processing circuit 110, it needs to be analog-to-digital converted and converted into a corresponding digital signal.
  • the output interface 112 is configured to output a processing result, such as fingerprint identification success or fingerprint recognition failure, etc., and the processing result may be displayed by driving the display screen through the display driving circuit, or by displaying the indicator light, or performing voice prompting through the speaker.
  • the processing circuit 110 outputs the corresponding fingerprint image information to the feature extraction unit 130, so that the feature extraction unit 130 performs feature extraction on the fingerprint image to obtain corresponding feature information.
  • the feature information may include first feature information and second feature information, the first feature information and the second feature information are feature information that reflects fingerprint characteristics from different angles, and the feature information amount of the first feature information is smaller than the second feature. The amount of characteristic information of the information is small.
  • the first feature information is used to describe the fingerprint feature of the user as a whole
  • the second feature information is used to accurately describe the texture of the fingerprint, including all feature point information describing the texture of the ridge portion, due to the portion of the valley in the fingerprint and the position of the portion of the ridge.
  • the position of the valley portion of the fingerprint can be determined by determining the position of the ridge portion of the fingerprint.
  • the first feature information may also include all feature point information describing the texture of the valley portion, or all feature point information of the grain of the valley portion and the ridge portion.
  • the first feature information and the second feature information may also be other suitable fingerprint information.
  • the fingerprint template stored in the storage unit 120 corresponds to the first feature information and the second feature information including the registered fingerprint.
  • the difference between the fingerprint templates generated when the same finger is registered is small, when storing, the same portion in each fingerprint template and a portion having a difference between the fingerprint templates may be stored. This will save storage space.
  • all the fingerprint templates contain only feature information, and the corresponding fingerprint image cannot be obtained through the feature information, thereby ensuring the information security of the fingerprint template.
  • the feature comparison unit 140 compares the first feature information extracted by the first feature extraction unit 131 with the first feature information of the registered N fingerprint templates to obtain a comparison.
  • the fingerprint template for example, M fingerprint templates, M ⁇ N; the second feature information extracted by the second feature extraction unit 132 is compared with the second feature information in the fingerprint template that is aligned.
  • the output comparison results such as recognition success or recognition failure. If the first feature information is compared with the first feature information in all the registered fingerprint templates, no comparison is obtained.
  • the fingerprint template directly outputs the comparison result, that is, the second feature information is no longer compared.
  • the unmatched fingerprint template can be quickly filtered, and the second feature information is avoided from being compared with all the fingerprint templates in the storage unit 120, thereby saving the comparison time of the second feature information. , improve the speed of fingerprint recognition.
  • the comparison of the first feature information may be performed first. After the second feature information is extracted, the second feature information is compared, thereby further improving the fingerprint recognition speed.
  • the feature extraction unit 130 may include a first feature extraction unit 131 and a second feature extraction unit 132.
  • the first feature extraction unit 131 performs feature extraction on the fingerprint information by using a preset first feature extraction rule to obtain first feature information.
  • the first feature information is used to describe the fingerprint feature of the user as a whole, and only contains a small amount of feature information describing the characteristics of the user's fingerprint.
  • a human finger fingerprint may include three basic types: a concentric circle or a spiral line, which is called a bucket line; an opening on one side of the line, called a striated line; and a striate like a bow, called a bow line.
  • the first feature extraction unit 131 may extract first feature information such as the type of the fingerprint, the degree of curvature of the texture, and the interval width of the texture.
  • the second feature extraction unit 132 performs feature extraction on the fingerprint information by using a preset second feature extraction rule to obtain second feature information.
  • the second feature information is used to accurately describe the uniqueness of the fingerprint. Therefore, the second feature information includes, but is not limited to, feature point information of each stripe path in the collected fingerprint image, for example, location information of each feature point and the feature. The eigenvalue of the point.
  • the feature extraction unit 130 may further include a third feature extraction unit (not shown), and the third feature extraction unit performs feature on the fingerprint information by using a preset third feature extraction rule. Extracting, obtaining third feature information.
  • the third feature information is used to describe several detailed feature points of the fingerprint, such as a bifurcation point, a termination point, a center point, and a triangle point, and the feature information amount of the third feature information is less than the feature information amount of the second feature information. However, there is more feature information than the first feature information. Therefore, after the comparison of the first feature information by the feature comparison unit 140, the first fingerprint template that is aligned is obtained, and the third feature information in the first fingerprint template that matches the third feature information and the comparison is performed.
  • Performing an alignment obtaining a second fingerprint template consistent with the comparison, and comparing the second feature information with the second fingerprint template that is aligned, to obtain a comparison result, such as a successful recognition or a recognition failure.
  • a comparison result such as a successful recognition or a recognition failure.
  • the feature comparison unit 140 may include a first feature comparison unit and a second feature comparison unit.
  • the first feature comparison unit is used for feature comparison of the first feature information
  • the second feature comparison unit is used for feature comparison of the second feature information. Since the first feature information is extracted faster than the second The feature information is extracted at a high speed. Therefore, after the first feature information is extracted, the first feature comparison unit can perform feature comparison on the first feature information to obtain a matching fingerprint template.
  • the second feature comparison unit may perform feature matching on the fingerprint template in which the second feature information is consistent with the comparison, without waiting for the first feature information to end the feature comparison with all the fingerprint templates.
  • the feature comparison of the feature comparison of the first feature information with the second feature information can also be processed in parallel, thereby speeding up the feature comparison, that is, improving the fingerprint recognition speed.
  • the first feature information is coded feature information obtained after performing coded clustering processing on the fingerprint image.
  • the first feature information in the fingerprint template is coded feature information.
  • the first feature extraction unit 131 may include an image encoding module 1311 and a feature categorization statistics module 1312 .
  • the image encoding module 1311 is configured to perform image coding on the fingerprint image to obtain a coding feature.
  • the feature classification statistics module 1312 is configured to perform feature classification on the coding features, and count the number of each class to form final coding feature information.
  • the encoding method may include Local Binary Pattern (LBP), Local Phase Quantization (LPQ), and Binaryized Statistical Image Features (BSIF).
  • LBP Local Binary Pattern
  • LPQ Local Phase Quantization
  • BSIF Binaryized Statistical Image Features
  • the encoding may be performed by using one of the encoding methods, or may be encoded by using multiple encoding methods, as long as the encoding rule of the first feature information in the fingerprint template is consistent with the encoding rule of the first feature information in the fingerprint to be detected. can.
  • Feature categorization methods may include k-means (k-means), hierarchical clustering, Self-Organizing Maps (SOM), Fuzzy C-means (FCM) clustering, etc., and the number of clusters may be For fixed values, of course, you can also flexibly set according to the actual situation.
  • the coding feature obtained by the image coding module 1311 is compared with the preset coding feature core by using a distance metric or the like, and the coding feature is merged into the class with the smallest distance.
  • the feature categorization statistics module 1312 then counts the number of each class, and finally obtains a statistical histogram, the final coding feature.
  • the preset coding feature class core includes a cluster center value or a cluster mean value of the feature, a feature range value of the cluster, and the like.
  • the preset coding feature core is pre-computed and saved to the image recognition device for reuse in feature classification.
  • the preset coding feature class core can also be updated in real time through an online manner to improve the previously set coding feature class.
  • the preset coding feature class core can also be obtained by processing the registered fingerprint template. For example, the fingerprint image in the registered fingerprint template is image-encoded to obtain a coding feature; then the feature features are clustered to obtain a coding feature core.
  • the encoded feature core is obtained by using the registered fingerprint template processing
  • the coding feature obtained by performing feature classification statistics according to the coding feature class is closer to the coding feature of the registered fingerprint template, so that when the feature comparison is performed, not only the acceleration is accelerated. Compare the speed and ensure the accuracy of the comparison.
  • FIG. 11 is a fingerprint image of a parallel texture and a feature distribution after BSIF encoding the fingerprint image
  • FIG. 12 is a fingerprint image of a curved texture, and BSIF encoding the fingerprint image.
  • Characteristic distribution The fingerprint images shown in FIG. 11 and FIG. 12 only reflect the characteristics of the fingerprint as a whole, that is, the first feature information including only the fingerprint information. It can be seen from the feature distributions in FIG. 11 and FIG. 12 that the difference between the parallel grain and the curved grain is obvious, and thus the feature comparison is performed by the coding feature, thereby further improving the comparison speed, that is, the fingerprint recognition speed.
  • the fingerprint image may be subjected to certain preprocessing, for example, noise processing of the fingerprint image, enhancement processing of the fingerprint image, and the like. If there is noise in the fingerprint image, there will be burrs when extracting features, and there will be many false minutiae points, that is, pseudo-detail points.
  • the noise processing of the image may include processing of the spatial domain and processing of the frequency domain, such as averaging or intermediate values of image pixels; frequency domain processing such as low pass filtering.
  • the enhancement processing is mainly for highlighting the contrast between the ridge portion and the valley portion of the fingerprint image, for example, by adjusting the gray value of the image pixel, sharpening the fingerprint image, and increasing the outline of the fingerprint image and the sharpness of the line.
  • adjusting the gray value of the image pixel sharpening the fingerprint image
  • increasing the outline of the fingerprint image and the sharpness of the line there are other ways to deal with it, not to mention here.
  • the image recognition device described above may be integrated as a processing chip, disposed separately from the image acquisition unit. In this way, the design of the image recognition device can be performed independently, thereby reducing the design cost.
  • the image recognition device described above may also be integrated with the image acquisition unit as an image recognition chip, such as a fingerprint recognition sensor. In this way, the fingerprint recognition sensor collects fingerprints and performs fingerprint recognition internally. Even if the terminal main system applied by the fingerprint recognition sensor has a security threat, the biological information is still safe, thereby ensuring the security of the biological information.
  • the image recognition device described above can be applied to implement a corresponding function on the smart terminal, for example, the image recognition device recognizes the user identity, and after the recognition succeeds, the user is unlocked, the application is started, the online payment is performed, and the like.
  • the smart terminal can be a consumer electronic product or a home-based electronic product or a vehicle-mounted electronic product.
  • consumer electronic products such as mobile phones, tablets, notebook computers, desktop monitors, computer integrated machines and other electronic products using biometric identification technology.
  • Home-based electronic products such as smart door locks, televisions, refrigerators, wearable devices and other electronic products that use biometric technology.
  • Vehicle-mounted electronic products such as car navigation systems, car DVDs, etc.
  • the image recognition device can be integrated with the image acquisition unit as a chip, such as a fingerprint sensor chip, a collection fingerprint collection function and a fingerprint matching function; the image recognition device can also integrate a separate processing chip, that is, the image recognition device is integrated in the chip. All or part of a component.
  • an electronic device 500 is provided with an image capture device 501 that collects image information and an image recognition device 100 according to any of the above embodiments.
  • the electronic device may also be configured with an image capture function.
  • An image recognition chip combined with an image matching function.
  • the image capturing device 501 can be a photoelectric sensing module, a capacitive sensing module, and of course, an imaging device.
  • the image capture device 501 collects image information of the target object, such as a fingerprint image, and transmits the acquired image information to the image recognition device 100.
  • the image recognition device 100 recognizes the image information and outputs a recognition result such as recognition success or recognition failure.
  • the image recognition apparatus 100 adopts a new recognition processing technique, at the time of feature extraction, feature extraction is performed on the image information according to different angles, and the first feature information and the second feature information are obtained, and the feature information amount of the first feature information is obtained.
  • the first feature information is first compared with all the registered fingerprint templates to obtain a matching fingerprint template, and the fingerprints that are inconsistent are removed.
  • the template is then compared with the fingerprint template that matches the alignment, thereby speeding up the feature comparison, that is, the image recognition speed.
  • the electronic device 500 is a mobile phone, and the front surface of the mobile phone is provided with a display device 400 , and the image capturing device 501 or the image recognition chip is disposed under the front cover of the electronic device 500 .
  • the image capture device 501 or the image recognition chip may also be disposed on the display device 400.
  • the image capture device 501 can also be integrated as an image recognition chip, or the image capture device 501 can also be integrated with the image recognition device 100 as an image recognition chip, correspondingly disposed on the front, back, and sides of the electronic device 500. The position may be exposed to the outer surface of the electronic device 500 or may be disposed inside the electronic device 500 and adjacent to the outer casing.
  • the electronic device 600 when the image recognition device is applied to the smart terminal, the existing structure of the smart terminal may be utilized, and some structures of the image recognition may be added.
  • the electronic device 600 includes a processor 601, a memory 602, a display unit 603, a user input unit 604, a power source 605, and a communication unit 606.
  • the memory 602 is used to store all data on the smart terminal 600 that needs to be stored, such as external data and internal processing data.
  • the memory 602 can include an internal memory unit and an external memory unit.
  • Display unit 603 includes, but is not limited to, an LCD, a TFT-LCD, an OLED, a flexible display, and the like.
  • the user input unit 604 can generate input data according to a command input by the user to control various operations of the smart terminal 600.
  • the user input unit 604 allows the user to input various types of information, and may include a button, a pot, a touch pad (eg, a touch sensitive component that detects changes in resistance, pressure, capacitance, etc. due to contact), a scroll wheel, a shaker. Rod and so on.
  • a touch panel is superimposed on the display unit 603 in a layer form, a touch screen may be formed.
  • the button in the user input unit 604 It may include a virtual button disposed on the touch panel, or a physical button disposed in a non-display area, such as a Home button, a power button, and the like.
  • the power source 605 is used to provide the working power required for the smart terminal to operate, and to provide the standby power required to maintain the standby state of the smart terminal.
  • the communication unit 606 may include a communication interface such as USB, Type-C, etc.; of course, may also include a wireless communication interface such as GPRS, WCDMA, wifi, radio frequency, Bluetooth, infrared, and the like. It should be noted that FIG. 14 only exemplifies some components of the electronic device 600, and may also include other functional components, such as a camera, a microphone, a speaker, etc., to implement functions required by the user.
  • the electronic device 600 is further provided with an image acquisition device 601 and an image recognition device.
  • the structure of the image recognition device is similar to that of the image recognition device 100 of the above embodiment, but only in the image recognition device 100.
  • the processing circuit 110 and the storage unit 120 are shared with the processor 601 and the memory 602 in the electronic device 600, and other structures of the image recognition device 100 may be stored in the memory 602 or in a separate storage medium in the form of a software program, such as an instruction code.
  • the existing structure of the smart terminal can be utilized, and only the software structure of the image recognition can be added, thereby greatly reducing the manufacturing cost of the image recognition.
  • the image capturing device 601 can be a photoelectric sensing module, a capacitive sensing module, and of course an imaging device.
  • the configuration of the image acquisition device 601 in the electronic device 600 can be referred to the previous embodiment as an example, and details are not described herein again.
  • the image capture device 601 collects image information of the target object, such as a fingerprint image, and transmits the acquired image information to the image recognition device 602.
  • the image recognition device 602 processes the image information, it recognizes the image information and outputs a recognition result such as recognition success or recognition failure.
  • the image recognition device adopts a new recognition processing technology, at the time of feature extraction, feature extraction is performed on the image information according to different angles, and the first feature information and the second feature information are obtained, and the feature information ratio of the first feature information is obtained.
  • the second feature information has a small amount of feature information. Therefore, when the feature comparison is performed, the first feature information is first compared with all the registered fingerprint templates to obtain a matching fingerprint template, and the fingerprint template that is inconsistent is removed. Then, the second feature information is compared with the fingerprint template that is aligned, thereby speeding up the feature comparison, that is, the image recognition speed.
  • each unit in the above embodiment may be integrated into one processing unit, or two or more units may be integrated.
  • the above integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.
  • the units integrated in the above embodiments may also be stored in a computer readable storage medium if they are implemented in the form of software functional modules and sold or used as separate products. Based on such understanding, the technical solution of the embodiments of the present invention may be embodied in the form of a software product in essence or in part contributing to the prior art.
  • the computer software product is stored in a storage medium and includes instructions for causing a smart terminal (consumer electronics, home electronics, or vehicle-mounted electronic product) to perform all or part of the methods of the various embodiments of the present invention .
  • first and second are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated.
  • features defining “first” or “second” may include at least one of the features, either explicitly or implicitly.
  • the meaning of "a plurality” is at least two, such as two, three, etc., unless specifically defined otherwise.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Human Computer Interaction (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Collating Specific Patterns (AREA)

Abstract

一种图像识别方法、图像识别装置、电子装置（500）。该图像识别方法包括步骤：获取待识别的图像信息；对所获取的图像信息进行特征提取，获得第一特征信息和第二特征信息；将第一特征信息与注册的图像模板中的第一特征信息进行比对，获得比对一致的图像模板；将第二特征信息与比对一致的图像模板中的第二特征信息进行比对，获得识别结果。该图像识别方法加快了图像识别速度。

Description

图像识别方法、图像识别装置、电子装置及计算机存储介质 技术领域
本发明涉及生物识别领域，尤其涉及一种图像识别方法、图像识别装置、电子装置及计算机存储介质。
背景技术
生物识别技术，尤其是指纹识别，被广泛应用于手机等智能终端上。由于智能终端的设置位置有限，再者进行生物图像感测时，感测面积越大则成本也越高，因此，图像感测面积有一定的限定，以指纹识别为例，目前常用的感应面积是整个手指指纹区域的25％～30％。
然，由于感测面积小，若要保证识别准确率，则必须采集多次进行注册，识别时则需要将待识别的图像信息与注册的所有模板一一比对后，才能获得识别结果。如此，使得图像识别耗时长，不利于用户体验。
发明内容
本发明实施方式旨在至少解决现有技术中存在的技术问题之一。为此，本发明实施方式需要提供一种图像识别方法、图像识别装置、电子装置及计算机存储介质。
本发明实施方式的一种图像识别方法，包括以下步骤：
S1，获取待识别的图像信息；
S2，对所获取的图像信息进行特征提取，获得第一特征信息和第二特征信息；
S3，将第一特征信息与注册的图像模板中的第一特征信息进行比对，获得比对一致的图像模板；
S4，将第二特征信息与比对一致的图像模板中的第二特征信息进行比对，获得识别结果。
在某些实施方式中，所述步骤S2中对所获取的图像信息进行特征提取，获得第一特征信息具体包括：
对所获取的图像信息进行图像编码，获得初始编码特征；
按照预设的编码特征类心对初始编码特征进行特征归类，并统计每个类的数目，形 成最终的编码特征信息。
在某些实施方式中，所述图像编码采用如下一种或多种编码方法：局部二值模式、局部相位量化、图像特征值统计。
在某些实施方式中，所述特征归类采用如下一种或多种聚类方法：k均值、层次聚类、自组织映射、模糊C均值聚类。
在某些实施方式中，所述预设的编码特征类心为预先设置或者通过在线的方式实时更新。
在某些实施方式中，所述步骤S4在获得比对一致的图像模板后，与步骤S3并行执行。
在某些实施方式中，所述步骤S4中当第二特征信息比对成功，则结束识别流程，并输出识别结果。
在某些实施方式中，所述步骤S3中，当所述第一特征信息与注册的所有图像模板中的第一特征信息均比对不一致，则结束识别流程，并输出识别结果。
在某些实施方式中，所述步骤S2之前还包括：
S5，对获取的图像信息进行预处理。
本发明实施方式的一种图像识别方法，包括以下步骤：
S01，获取待注册的图像信息；
S02，对获取的图像信息进行特征提取，获取第一特征信息和第二特征信息；
S03，根据第一特征信息和第二特征信息，形成相应的图像模板；
S04，判断注册采集次数是否达到预设阈值；是则执行步骤S05，否则返回执行步骤S01；
S05，将所有的图像模板进行存储，以用于对待识别的图像信息进行识别。
在某些实施方式中，所述步骤S05之后包括：
通过一图像采集单元对目标物体的图像信息进行采集，采用前述任一实施方式的图像识别方法中的步骤对待识别的图像信息进行识别。
本发明实施方式的一种图像识别装置，包括处理器和存储单元；所述处理器用于执行前述任一实施方式的图像识别方法。
本发明实施方式的一种电子装置，包括图像采集装置和处理器；所述图像采集装置用于在目标物体放置于电子装置上时，采集所述目标物体的图像信息；所述处理器用于执行前述任一实施方式的图像识别方法。
本发明实施方式的一种计算机存储介质，存储有计算机可执行指令，所述计算机可执行指令用于执行前述任一实施方式的图像识别方法。
通过新的图像识别技术，在进行图像识别时，具有如下有点：
(1)通过第一特征信息和第二特征信息的设置，在进行特征比对时，先进行第一特征信息的特征比对，快速过滤不匹配的指纹模板，再进行第二特征信息的特征比对时，就节省了第二特征信息的比对时间，提高了指纹识别速度。
(2)第一特征信息和第二特征信息可以并行进行特征提取，进一步提高了指纹识别速度。
(3)由于第一特征信息的提取速度比第二特征信息的提取速度快，因此可以先进行第一特征信息的比对，待第二特征信息提取后，再并行进行第二特征信息的特征比对，进一步提高了指纹识别速度。
本发明实施方式的附加方面和优点将在下面的描述中部分给出，部分将从下面的描述中变得明显，或通过本发明实施方式的实践了解到。
附图说明
本发明实施方式的上述和/或附加的方面和优点从结合下面附图对实施方式的描述中将变得明显和容易理解，其中：
图1是本发明实施方式的图像识别中以指纹识别为例的指纹注册示意图；
图2是本发明实施方式的图像识别中以指纹识别为例的指纹识别示意图；
图3是本发明实施方式的图像识别方法的注册过程的流程示意图；
图4是本发明实施方式的图像识别方法中存储的指纹模板示意图；
图5是本发明实施方式的图像识别方法的识别过程一实施例的流程示意图；
图6是本发明实施方式的图像识别方法的识别过程另一实施例的流程示意图；
图7是本发明实施方式的图像识别方法的识别过程又一实施例的流程示意图。
图8是本发明实施方式的图像识别装置的功能框图；
图9是本发明实施方式的图像识别装置中进行特征比对时特征信息与指纹模板的比对示意图；
图10是本发明实施方式的图像识别装置中第一特征提取单元的功能框图；
图11是本发明实施方式的图像识别装置中一种平纹的指纹图像及按照BSIF编码对其进行编码后获得的编码特征分布示意图；
图12是本发明实施方式的图像识别装置中一种箩纹的指纹图像及按照BSIF编码对 其进行编码后获得的编码特征分布示意图；
图13是本发明实施方式的图像识别装置所应用的电子装置一实施例的平面示意图；
图14是本发明实施方式的图像识别装置所应用的电子装置另一实施例的功能框图。
具体实施方式
下面详细描述本发明的实施方式，所述实施方式的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施方式是示例性的，仅用于解释本发明，而不能理解为对本发明的限制。
在本发明的描述中，需要理解的是，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括一个或者更多个所述特征。在本发明的描述中，“多个”的含义是两个或两个以上，除非另有明确具体的限定。
在本发明的描述中，需要说明的是，除非另有明确的规定和限定，术语“安装”、“相连”、“连接”应做广义理解，例如，可以是固定连接，也可以是可拆卸连接，或一体地连接；可以是机械连接，也可以是电连接或可以相互通信；可以是直接相连，也可以通过中间媒介间接相连，可以是两个元件内部的连通或两个元件的相互作用关系。对于本领域的普通技术人员而言，可以根据具体情况理解上述术语在本发明中的具体含义。
下文的公开提供了许多不同的实施方式或例子用来实现本发明的不同结构。为了简化本发明的公开，下文中对特定例子的部件和设定进行描述。当然，它们仅仅为示例，并且目的不在于限制本发明。此外，本发明可以在不同例子中重复参考数字和/或参考字母，这种重复是为了简化和清楚的目的，其本身不指示所讨论各种实施方式和/或设定之间的关系。此外，本发明提供了的各种特定的工艺和材料的例子，但是本领域普通技术人员可以意识到其他工艺的应用和/或其他材料的使用。
进一步地，所描述的特征、结构可以以任何合适的方式结合在一个或更多实施方式中。在下面的描述中，提供许多具体细节从而给出对本发明的实施方式的充分理解。然而，本领域技术人员应意识到，没有所述特定细节中的一个或更多，或者采用其它的结构、组元等，也可以实践本发明的技术方案。在其它情况下，不详细示出或描述公知结构或者操作以避免模糊本发明。
本发明实施方式中的图像识别，是指利用图像传感器对放置在指定位置的目标物体进行图像采集，该采集到的图像将通过柔性印刷电路板或其他印刷电路传输至处理系统进行图像识别，即识别该图像是否已注册或登记，或者识别用户的身份等等。
请参阅图1及图2，以指纹识别为例，在对指纹图像进行识别之前，必须对用户的指纹图像进行注册。首先用户将需要注册的手指200放置在指定的位置，此时指纹采集单元103对手指的指纹进行采集，获得指纹图像。特征提取单元101则对指纹采集单元103获得的指纹图像进行指纹特征提取，获得指纹特征信息，并形成相应的指纹模板，存储至存储单元102的预定位置。如此重复多次，形成N个相应的指纹模板。需要说明的是，由于指纹采集单元103的感测面积较小，因此同一个指纹要采集多次，才能形成完整的指纹信息。故，假设1个手指采集5次，采集了5个手指，则共产生25个指纹模板。
当用户需要进行指纹识别时，将手指200放置在指定的位置，指纹采集单元103对手指的指纹进行采集，获得指纹图像。特征提取单元101对其进行指纹特征提取，获得待比对的指纹特征信息。特征比对单元104将待比对的指纹特征信息与存储单元102中的所有指纹模板进行一一比对，并输出比对结果，例如识别成功或识别失败。这里的指纹采集单元103可包括光学式指纹采集单元、电容式指纹采集单元、超声波指纹采集单元、射频指纹采集单元等等中的任意一种或几种。
由于待比对的指纹特征信息需要与存储单元102中的所有指纹模板一一进行比对后，才能获得识别结果，因此指纹模板的数量越多，则指纹识别的时间越长，如此使得指纹识别的速度较慢，不利于用户体验。
为了提高图像识别速度，本发明提出了一种新的图像识别方法。该图像识别方法在进行图像特征提取时，将图像按不同的特征提取规则进行特征提取，以获得不同的特征信息，例如第一特征信息、第二特征信息等等；在特征信息比对时，先将第一特征信息与图像注册信息中的第一特征信息进行比对，若比对一致，再将第二特征信息与该注册信息中的第二特征信息进行比对，最后输出比对结果。由于第一特征信息仅包括图像的少量特征，第二特征信息包括图像的所有特征，因此第一特征信息的比对速度比第二特征信息的比对速度较快，即该图像识别方法提高了图像识别速度。该图像信息可包括指纹信息、掌纹信息、耳纹信息以及生物体上其它合适位置的皮肤纹理信息。由于生物体的皮肤纹理、皮层结构具有不同的特性，根据该特性可以识别出不同的生物体。所述生物体例如为人体，但可不限于人体。以下实施方式中，将以指纹识别为例进行描述。
本发明实施方式的一种图像识别方法可包括注册过程和识别过程。在注册过程中，使用者注册自己的图像信息，例如指纹信息，该注册的指纹信息将通过指纹模板的形式进行存储，以供识别过程使用。识别过程中，将采集到的待检测的指纹图像与存储的指 纹模板进行特征比对，并输出识别结果，例如识别成功或识别失败等。
请参阅图3，在一些实施方式中，注册过程具体包括：
步骤S101，获取待注册的指纹图像；
该待注册的指纹图像通过图像传感器采集获得，以指纹传感器为例，当用户的手指位于指纹传感器上，指纹传感器将采集手指的指纹图像，并将采集的指纹图像传输至指纹识别系统。指纹识别系统100的输入接口111可通过柔性印刷电路板或其他的电路板与指纹传感器电性连接。
步骤S102，对获取的指纹图像进行特征提取，获取第一特征信息和第二特征信息；
该第一特征信息和第二特征信息从不同的角度来体现指纹特点，而且第一特征信息的特征信息量比第二特征信息的特征信息量少。因此，按照不同的特征提取规则对获取的指纹图像进行特征提取，可获得第一特征信息和第二特征信息。第一特征信息的提取和第二特征信息的提取可先后进行，也可以并行处理。而且，并行处理特征信息的提取，可以加快特征提取速度。
具体地，该第一特征信息用于整体描述用户的指纹特点，仅包含描述用户指纹特点的少量特征信息。例如，人体的手指指纹可包括3种基本类型：有同心圆或螺旋纹线，称为斗形纹；纹线一边开口的，称为箕形纹；纹形像弓一样，称为弓线纹。而且不同类型的指纹，在纹路的弯曲程度、纹路的间隔宽度等也存在差异。故而在进行第一特征信息提取时，可提取指纹的类型、纹路的弯曲程度、纹路的间隔宽度等。第二特征信息用于精确描述指纹的独特性，则包含指纹信息的全部特征信息，例如整个指纹的纹路。故而在进行第二特征信息提取时，将提取整个指纹的纹路信息。
当然，这里并不限定包括第一特征信息和第二特征信息，在其他例子中，也可以根据需要提取第三特征信息，第四特征信息等等。
步骤S103，根据第一特征信息和第二特征信息，形成相应的指纹模板；即每个指纹模板中均包括对应的第一特征信息和第二特征信息。
步骤S104，判断注册采集次数是否达到预设阈值；是则执行步骤S105，否则返回执行步骤S101；
为了保证指纹图像的完整、准确地采集，需要采集多次，一些实施方式中，设定采集至少3次，即预设阈值为3。需要说明的是，每次采集都是有效采集，即每次采集的指纹图像要满足清晰度、手指扫描位置等要求。若当前采集的指纹图像不符合要求，例如手指移动造成采集图像模糊，则放弃当前采集的图像，提示用户重新进行指纹图像的 采集。
步骤S105，将所有的指纹模板进行存储。
待注册采集次数达到预设阈值，则将注册采集形成的所有指纹模板进行存储。在一些例子中，由于同一手指注册时生成的指纹模板之间差异较小，故而在存储时，可以存储每个指纹模板中相同的部分以及各指纹模板之间存在差异的部分。如此将节省存储空间。另外，所有的指纹模板中仅包含特征信息，通过该特征信息无法获得对应的指纹图像，因此保证了指纹模板的信息安全。
当存在多个手指注册时，为区分该多个手指，可以对所有的指纹模板进行标识。请参阅图4，目前应用于手机的指纹识别，可注册采集5个手指，例如手指A、手指B、手指C、手指D、手指E。每个注册的手指均具有对应的指纹模板库，即每个指纹模板具有对应的手指标识，而且每个对应的指纹模板库具有多个指纹模板。以手指A为例，该手指A对应的指纹模板包括模板1A、模板2A、模板3A。
可以理解的是，上述图像的注册过程并不局限于上述实施方式描述的过程，也可以包括其他方式，目的为了获得包括第一特征信息和第二特征信息的指纹模板。当然，该注册过程中的图像模板形成后，可以重复用于图像识别过程，从而节省了每次的图像注册过程。
请参阅图5，在一些实施方式中，识别过程具体可包括：
步骤S201，获取待识别的指纹图像；
该待识别的指纹图像通过图像传感器采集获得，以指纹传感器为例，当用户的手指位于指纹传感器上，指纹传感器将采集手指的指纹图像，并将采集的指纹图像输出至指纹识别系统。指纹识别系统100的输入接口111可通过柔性印刷电路板或其他的电路板与指纹传感器电性连接。
步骤S202，对所获取的指纹图像进行特征提取，获得第一特征信息和第二特征信息；
该第一特征信息和第二特征信息从不同的角度来体现指纹特点，而且第一特征信息的特征信息量比第二特征信息的特征信息量少。因此，按照不同的特征提取规则对获取的指纹图像进行特征提取，可获得第一特征信息和第二特征信息。第一特征信息的提取和第二特征信息的提取可先后进行，也可以并行处理。而且，并行处理特征信息的提取，可以加快特征提取速度。
具体地，该第一特征信息用于整体描述用户的指纹特点，仅包含描述用户指纹特点 的少量特征信息。例如，人体的手指指纹可包括3种基本类型：有同心圆或螺旋纹线，称为斗形纹；纹线一边开口的，称为箕形纹；纹形像弓一样，称为弓线纹。而且不同类型的指纹，在纹路的弯曲程度、纹路的间隔宽度等也存在差异。故而在进行第一特征信息提取时，可提取指纹的类型、纹路的弯曲程度、纹路的间隔宽度等。第二特征信息用于精确描述指纹的独特性，则包含指纹信息的全部特征信息，例如整个指纹的纹路。故而在进行第二特征信息提取时，将提取整个指纹的纹路信息。
步骤S203，将第一特征信息与注册的所有指纹模板中的第一特征信息进行比对，获得比对一致的指纹模板；
步骤S204，将第二特征信息与比对一致的指纹模板中的第二特征信息进行比对，获得比对结果。
在一些例子中，可以待步骤S203中第一特征信息跟所有的指纹模板比对结束后，再执行步骤S204。再另一些例子中，可以在步骤S203中获得比对一致的指纹模板时，就并行执行步骤S204，当第二特征信息比对成功后，即可停止识别，如此可以提前获得比对结果，加快识别速度。
进一步地，上述第一特征信息为对指纹图像进行编码聚类处理后获得的编码特征信息。对应地，指纹模板中第一特征信息为编码特征信息。请参阅图6，识别过程具体可包括：
步骤S301，获取待识别的指纹图像；
步骤S302，对所获取的指纹图像进行特征提取，获得编码特征信息和第二特征信息；
步骤S303，将编码特征信息与注册的所有指纹模板中的编码特征信息进行比对，获得比对一致的指纹模板；
步骤S304，将第二特征信息与比对一致的指纹模板中的第二特征信息进行比对，获得比对结果。
上述实施方式中，获得编码特征信息的过程具体可包括：对指纹图像进行图像编码，获得初始编码特征；对初始编码特征进行特征归类，并统计每个类的数目，形成最终的编码特征信息。
在一些实施方式中，编码方法可包括局部二值模式(Local Binary Pattern，LBP)、局部相位量化(Local Phase Quantization，LPQ)、图像特征值统计(Binarized Statistical Image Features，BSIF)等方法。当然，可以采用其中一种编码方法进行编码，也可以采 用多种编码方法结合的方式进行编码，只要指纹模板中第一特征信息的编码规则与待检测指纹中第一特征信息的编码规则一致即可。
特征归类方法可包括k-means(k均值)、层次聚类、自组织映射(Self-Organizing Maps，SOM)、模糊C均值(Fuzzy C-means，FCM)聚类等等，聚类数目可为固定值，当然也可以根据实际情况而灵活设置。具体地，特征归类处理时，利用距离度量等方法计算初始编码特征与预设的编码特征类心进行比较，并将编码特征归并到距离最小的那个类。然后统计每个类的数目，最终获得一个统计直方图，即最终的编码特征。
上述预设的编码特征类心包括特征的聚类中心值或者聚类均值，以及聚类的特征范围值等等。在一些例子中，该预设的编码特征类心为预先计算好，并保存至图像识别装置中，以供特征归类时重复使用。当然，该预设的编码特征类心还可以通过在线的方式实时更新，以对之前设置的编码特征类心进行完善。另外，该预设的编码特征类心还可以通过对注册的指纹模板进行处理获得。例如，对注册的指纹模板中的指纹图像进行图像编码，获得编码特征；然后对编码特征进行特征聚类，获得编码特征类心。由于利用注册的指纹模板处理获得编码特征类心，因此根据该编码特征类心进行特征归类统计所获得的编码特征更接近注册的指纹模板的编码特征，因此进行特征比对时，不但加快了比对速度，而且保证了比对的准确率。
进一步地，请参阅图7，在一些实施方式中，步骤S202之前还可包括：
步骤S205，对获取的指纹图像进行预处理。
为了保证指纹图像的处理效果，在对指纹图像进行提取之前，还可以对指纹图像进行一定的预处理，例如，指纹图像的噪声处理，对指纹图像进行增强处理等等。若指纹图像中存在噪声，则提取特征时会存在毛刺现象，而且会存在很多假的细节点，也就是伪细节点。其中图像的噪声处理可包括空间域的处理和频率域的处理，空间域处理例如求图像像素的平均值或中间值；频率域处理例如低通滤波。增强处理主要是为了突出指纹图像的脊部分和谷部分之间的对比度，例如通过调整图像像素的灰度值，对指纹图像进行锐化处理，增加指纹图像的轮廓及线条的清晰度。当然，还存在其他的处理方式，这里不一一例举。
本领域技术人员可以理解：实现上述方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成，前述的程序可以存储于一计算机可读取存储介质中，该程序在执行时，执行包括上述方法实施例的步骤；而前述的存储单元包括移动存储设备、ROM、RAM、磁碟或光盘等各种可以存储程序代码的介质。
请参阅图8，本发明实施方式的一种图像识别装置100用于实现上述实施方式的方法步骤。该图像识别装置100包括处理电路110、存储单元120、特征提取单元130、特征比对单元140。
处理电路110例如但不局限为微型处理器，具有输入接口111和输出接口112。该输入接口111用于接收一图像采集单元(图未示)采集的图像信息，例如指纹采集单元采集的指纹图像信息。该输入接口111接收到的指纹图像信息可以为模拟信号，也可以为数字信号。若输入接口111接收到的指纹图像信息为模拟信号，则在送入处理电路110时，需要对其进行模数转换，转换成相应的数字信号。输出接口112用于输出处理结果，例如指纹识别成功或指纹识别失败等等，该处理结果可以通过显示驱动电路驱动显示屏进行显示，或者通过指示灯显示，或者通过扬声器进行语音提示。
处理电路110输出相应的指纹图像信息给特征提取单元130，以使特征提取单元130对指纹图像进行特征提取，获得相应的特征信息。该特征信息可包括第一特征信息和第二特征信息，该第一特征信息和第二特征信息为从不同的角度体现指纹特点的特征信息，而且第一特征信息的特征信息量比第二特征信息的特征信息量少。例如，第一特征信息用于整体描述用户的指纹特点，第二特征信息用于精确描述指纹的纹路，包括描述脊部分的纹路的所有特征点信息，由于指纹中谷的部分和脊的部分的位置是交替设置的，因此在确定指纹的脊部分的位置，就可以确定指纹的谷部分的位置。同理，该第一特征信息也可以包括描述谷部分的纹路的所有特征点信息，或者谷部分和脊部分的纹路的所有特征点信息。然，可变更地，所述第一特征信息和第二特征信息也可为其它合适的指纹信息。
相应地，存储单元120中存储的指纹模板中对应包括注册指纹的第一特征信息和第二特征信息。在一些例子中，由于同一手指注册时生成的指纹模板之间差异较小，故而在存储时，可以存储每个指纹模板中相同的部分以及各指纹模板之间存在差异的部分。如此将节省存储空间。另外，所有的指纹模板中仅包含特征信息，通过该特征信息无法获得对应的指纹图像，因此保证了指纹模板的信息安全。
请一并参阅图8与图9，特征比对单元140将第一特征提取单元131提取的第一特征信息与注册的N个指纹模板中的第一特征信息进行比对，获得比对一致的指纹模板，例如M个指纹模板，M<N；再将第二特征提取单元132提取的第二特征信息与比对一致的指纹模板中的第二特征信息进行比对。最后输出比对结果，例如识别成功或识别失败。若第一特征信息与注册的所有指纹模板中的第一特征信息比对后，均未获得比对一 致的指纹模板，则直接输出比对结果，即不再对第二特征信息进行比对。
因此，通过第一特征信息的比对，可以快速过滤不匹配的指纹模板，避免了第二特征信息与存储单元120中所有的指纹模板进行比对，从而节省了第二特征信息的比对时间，提高了指纹识别速度。另外，由于第一特征信息的提取和第二特征信息的提取可以并行处理，而且第一特征信息提取速度比第二特征信息的提取速度快，因此可以先进行第一特征信息的比对，待第二特征信息提取后，再进行第二特征信息的比对，从而进一步提高了指纹识别速度。
在一些实施方式中，上述特征提取单元130可包括第一特征提取单元131、第二特征提取单元132。第一特征提取单元131采用预设的第一特征提取规则对指纹信息进行特征提取，获得第一特征信息。该第一特征信息用于整体描述用户的指纹特点，仅包含描述用户指纹特点的少量特征信息。例如，人体的手指指纹可包括3种基本类型：有同心圆或螺旋纹线，称为斗形纹；纹线一边开口的，称为箕形纹；纹形像弓一样，称为弓线纹。而且不同类型的指纹，在纹路的弯曲程度、纹路的间隔宽度等也存在差异。故而第一特征提取单元131在进行提取时，可提取指纹的类型、纹路的弯曲程度、纹路的间隔宽度等第一特征信息。第二特征提取单元132采用预设的第二特征提取规则对指纹信息进行特征提取，获得第二特征信息。第二特征信息用于精确描述指纹的独特性，因此该第二特征信息包含但不局限于采集到的指纹图像中每条纹路的特征点信息，例如每个特征点所在的位置信息以及该特征点的特征值。
当然，在另一些实施方式中，上述特征提取单元130还可以包括第三特征提取单元(图中未示出)，该第三特征提取单元采用预设的第三特征提取规则对指纹信息进行特征提取，获得第三特征信息。该第三特征信息用于描述指纹的几个细节特征点，例如分叉点、终止点、中心点、三角点，该第三特征信息的特征信息量比第二特征信息的特征信息量少，但比第一特征信息的特征信息量多。因此，上述特征比对单元140在进行第一特征信息的比对后，获得比对一致的第一指纹模板，再将第三特征信息与比对一致的第一指纹模板中的第三特征信息进行比对，获得比对一致的第二指纹模板，再将第二特征信息与比对一致的第二指纹模板进行比对，获得比对结果，例如识别成功或识别失败。如此，不但可以保证指纹识别的准确度，而且还能加快指纹识别的速度。
进一步地，为了进一步加快特征比对的速度，上述特征比对单元140可以包括第一特征比对单元和第二特征比对单元。第一特征比对单元用于第一特征信息的特征比对，第二特征比对单元用于第二特征信息的特征比对。由于第一特征信息的提取速度比第二 特征信息的提取速度快，因此当第一特征信息提取后，第一特征比对单元即可对第一特征信息进行特征比对，以获得比对一致的指纹模板。此时第二特征比对单元可以将第二特征信息与该比对一致的指纹模板进行特征比对，而不用等第一特征信息与所有的指纹模板进行特征比对结束。换句话说，第一特征信息的特征比对与第二特征信息的特征比对也可以并行处理，从而加快了特征比对的速度，即提高了指纹识别速度。
在一些实施方式中，上述第一特征信息为对指纹图像进行编码聚类处理后获得的编码特征信息。对应地，指纹模板中第一特征信息为编码特征信息。
具体地，请参阅图10，第一特征提取单元131可包括图像编码模块1311、特征归类统计模块1312。图像编码模块1311用于对指纹图像进行图像编码，获得编码特征；特征归类统计模块1312用于对编码特征进行特征归类，并统计每个类的数目，形成最终的编码特征信息。
在一些实施方式中，编码方法可包括局部二值模式(Local Binary Pattern，LBP)、局部相位量化(Local Phase Quantization，LPQ)、图像特征值统计(Binarized Statistical Image Features，BSIF)等方法。当然，可以采用其中一种编码方法进行编码，也可以采用多种编码方法结合的方式进行编码，只要指纹模板中第一特征信息的编码规则与待检测指纹中第一特征信息的编码规则一致即可。
特征归类方法可包括k-means(k均值)、层次聚类、自组织映射(Self-Organizing Maps，SOM)、模糊C均值(Fuzzy C-means，FCM)聚类等等，聚类数目可为固定值，当然也可以根据实际情况而灵活设置。具体地，特征归类处理时，利用距离度量等方法计算图像编码模块1311获得的编码特征与预设的编码特征类心进行比较，并将编码特征归并到距离最小的那个类。然后特征归类统计模块1312统计每个类的数目，最终获得一个统计直方图，即最终的编码特征。
上述预设的编码特征类心包括特征的聚类中心值或者聚类均值，以及聚类的特征范围值等等。在一些例子中，该预设的编码特征类心为预先计算好，并保存至图像识别装置中，以供特征归类时重复使用。当然，该预设的编码特征类心还可以通过在线的方式实时更新，以对之前设置的编码特征类心进行完善。另外，该预设的编码特征类心还可以通过对注册的指纹模板进行处理获得。例如，对注册的指纹模板中的指纹图像进行图像编码，获得编码特征；然后对编码特征进行特征聚类，获得编码特征类心。由于利用注册的指纹模板处理获得编码特征类心，因此根据该编码特征类心进行特征归类统计所获得的编码特征更接近注册的指纹模板的编码特征，因此进行特征比对时，不但加快了 比对速度，而且保证了比对的准确率。
请参阅图11和图12，图11是一个平行纹路的指纹图像，以及对该指纹图像进行BSIF编码后的特征分布；图12是一个弯曲纹路的指纹图像，以及对该指纹图像进行BSIF编码后的特征分布。其中，图11和图12中示出的指纹图像仅整体反映了指纹的特点，即仅包含指纹信息的第一特征信息。从图11和图12中的特征分布可以看出，平行纹路和弯曲纹路的差异明显，因此通过编码特征进行特征比对，进一步提高了比对速度，即指纹识别速度。
为了保证指纹图像的处理效果，在对指纹图像进行提取之前，还可以对指纹图像进行一定的预处理，例如，指纹图像的噪声处理，对指纹图像进行增强处理等等。若指纹图像中存在噪声，则提取特征时会存在毛刺现象，而且会存在很多假的细节点，也就是伪细节点。其中图像的噪声处理可包括空间域的处理和频率域的处理，空间域处理例如求图像像素的平均值或中间值；频率域处理例如低通滤波。增强处理主要是为了突出指纹图像的脊部分和谷部分之间的对比度，例如通过调整图像像素的灰度值，对指纹图像进行锐化处理，增加指纹图像的轮廓及线条的清晰度。当然，还存在其他的处理方式，这里不一一例举。
在一些实施方式中，上述图像识别装置可集成为处理芯片，与图像采集单元分开设置。如此，可以使得图像识别装置的设计可以独立进行，从而降低了设计成本。然，可替换的，上述图像识别装置也可以与图像采集单元一起集成为图像识别芯片，例如指纹识别传感器。如此，该指纹识别传感器通过指纹采集，并在内部进行指纹识别，即使指纹识别传感器应用的终端主系统发生安全威胁，生物信息依然安全，从而保证了生物信息的安全。
上述图像识别装置可应用于智能终端上实现对应的功能，例如通过图像识别装置识别用户身份，识别成功后进行解锁、启动应用程序、网上支付等等。该智能终端可为消费性电子产品或家居式电子产品或车载式电子产品。其中，消费性电子产品如为手机、平板电脑、笔记本电脑、桌面显示器、电脑一体机等各类应用生物识别技术的电子产品。家居式电子产品如为智能门锁、电视、冰箱、穿戴式设备等各类应用生物识别技术的电子产品。车载式电子产品如为车载导航仪、车载DVD等。
该图像识别装置可以与图像采集单元集成为一颗芯片，例如指纹传感器芯片，集合指纹采集功能和指纹匹配功能；该图像识别装置也可以集成单独的处理芯片，即该芯片内集成有图像识别装置的所有组件或部分组件。
请参阅图13，本发明实施方式的一种电子装置500，其上设置有采集图像信息的图像采集装置501以及上述任一实施方式的图像识别装置100，当然该电子装置也可以设置图像采集功能和图像匹配功能结合的图像识别芯片。该图像采集装置501可为一光电传感模组，也可为一电容式传感模组，当然还可以为摄像装置。当目标物体位于图像采集装置501上时，图像采集装置501采集目标物体的图像信息，例如指纹图像，并将采集到的图像信息传输至图像识别装置100。图像识别装置100对图像信息进行处理后，对图像信息进行识别，并输出识别结果，例如识别成功或识别失败。
由于该图像识别装置100采用了新的识别处理技术，在特征提取时，对图像信息按照不同的角度进行特征提取，获得第一特征信息和第二特征信息，而且第一特征信息的特征信息量比第二特征信息的特征信息量少，因此在进行特征比对时，先将第一特征信息与注册的所有指纹模板进行比对，以获得比对一致的指纹模板，去掉比对不一致的指纹模板；然后再将第二特征信息与比对一致的指纹模板进行比对，从而加快了特征比对的速度，即图像识别速度。
在图13的示例中，电子装置500为手机，手机的正面设置有显示装置400，图像采集装置501或图像识别芯片设置在电子装置500的前盖板下方。然，可变更地，在其它实施方式中，图像采集装置501或图像识别芯片也可设置在显示装置400上。另外，所述图像采集装置501也可集成为图像识别芯片，或者该图像采集装置501还可以和图像识别装置100集成为图像识别芯片，对应设置在电子装置500的正面、背面、以及侧面等合适位置，且，既可曝露出电子装置500的外表面，也可设置在电子装置500内部并邻近外壳。
请参阅图14，在一些实施方式中，图像识别装置应用于智能终端时，也可以利用智能终端已有的结构，再增加图像识别的一些结构即可实现。具体地，该电子装置600包括处理器601、存储器602、显示单元603、用户输入单元604、电源605、通讯单元606。其中存储器602用于存储智能终端600上所有需要存储的数据，例如外部数据以及内部处理数据等等。该存储器602可包括内部存储单元和外部存储单元。显示单元603包括但不局限于LCD、TFT-LCD、OLED、柔性显示器等等。用户输入单元604可以根据用户输入的命令生成输入数据以控制智能终端600的各种操作。用户输入单元604允许用户输入各种类型的信息，并且可以包括按键、锅仔片、触摸板(例如，检测由于被接触而导致的电阻、压力、电容等变化的触敏组件)、滚轮、摇杆等等。特别地，当触摸板以层地形式叠加至显示单元603上时，可以形成触摸屏。该用户输入单元604中的按键 可包括设置于触摸板上的虚拟按键，也可以设置在非显示区域的实体按键，例如Home按键、电源按键等等。电源605用于提供智能终端工作所需的工作电源，以及提供维持智能终端待机状态所需的待机电源。通讯单元606可以包括通讯接口，例如USB、Type-C等；当然还可包括无线通讯接口，例如GPRS、WCDMA、wifi、射频、蓝牙、红外等。需要说明的是，图14仅例举了该电子装置600的部分组件，其还可以包括其他的功能组件，为实现使用者需要的功能，例如摄像头、麦克风、扬声器等。
当然，为了实现图像识别功能，该电子装置600上还设置图像采集装置601以及图像识别装置，该图像识别装置的结构与上述实施方式的图像识别装置100的结构类似，只是图像识别装置100中的处理电路110和存储单元120与电子装置600中的处理器601和存储器602共用，而图像识别装置100的其他结构可以通过软件程序，例如指令代码的形式，存储在存储器602中或单独的存储介质，供处理器601调用，以实现图像识别功能。如此，该图像识别装置应用于智能终端时，可以利用智能终端的已有结构，只需再增加图像识别的软件结构即可，从而大大降低了图像识别的制作成本。
上述图像采集装置601可为一光电传感模组，也可为一电容式传感模组，当然还可以为摄像装置。该图像采集装置601在电子装置600的设置结构可参照前面一实施方式为例，在此不再赘述。当目标物体位于图像采集装置601上时，图像采集装置601采集目标物体的图像信息，例如指纹图像，并将采集到的图像信息传输至图像识别装置602。图像识别装置602对图像信息进行处理后，对图像信息进行识别，并输出识别结果，例如识别成功或识别失败。
由于该图像识别装置采用了新的识别处理技术，在特征提取时，对图像信息按照不同的角度进行特征提取，获得第一特征信息和第二特征信息，而且第一特征信息的特征信息量比第二特征信息的特征信息量少，因此在进行特征比对时，先将第一特征信息与注册的所有指纹模板进行比对，以获得比对一致的指纹模板，去掉比对不一致的指纹模板；然后再将第二特征信息与比对一致的指纹模板进行比对，从而加快了特征比对的速度，即图像识别速度。
上述实施方式中描述的实施例仅仅是示意性的，例如各单元的划分仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，如多个单元或组件可以结合，或者一些特征可以忽略，或不执行。另外，上述实施方式中的各功能单元可以全部集成在一个处理单元，也可以两个或两个以上的单元集成在一起。上述集成的单元可以采用硬件的形式实现，也可以采用硬件加软件功能单元的形式实现。
上述实施方式中集成的单元如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。基于这样的理解，本发明实施方式的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来。该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一智能终端(消费性电子产品、家居式电子产品、或车载式电子产品)执行本发明各个实施例所述方法的全部或部分。
需要指出的是，上述例子是为了方便理解本发明实施方式而作出的一些例子，而不应理解为对本发明保护范围的限制。
在本说明书的描述中，参考术语“一个实施方式”、“某些实施方式”、“示意性实施方式”、“示例”、“具体示例”、或“一些示例”等的描述意指结合所述实施方式或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个实施方式或示例中。在本说明书中，对上述术语的示意性表述不一定指的是相同的实施方式或示例。而且，描述的具体特征、结构、材料或者特点可以在任何的一个或多个实施方式或示例中以合适的方式结合。
此外，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。在本发明的描述中，“多个”的含义是至少两个，例如两个，三个等，除非另有明确具体的限定。
尽管上面已经示出和描述了本发明的实施方式，可以理解的是，上述实施方式是示例性的，不能理解为对本发明的限制，本领域的普通技术人员在本发明的范围内可以对上述实施方式进行变化、修改、替换和变型。

Claims (14)

  1. 一种图像识别方法，其特征在于，所述方法包括以下步骤：
    S1，获取待识别的图像信息；
    S2，对所获取的图像信息进行特征提取，获得第一特征信息和第二特征信息；
    S3，将第一特征信息与注册的图像模板中的第一特征信息进行比对，获得比对一致的图像模板；
    S4，将第二特征信息与比对一致的图像模板中的第二特征信息进行比对，获得识别结果。
  2. 如权利要求1所述的图像识别方法，其特征在于，所述步骤S2中对所获取的图像信息进行特征提取，获得第一特征信息具体包括：
    对所获取的图像信息进行图像编码，获得初始编码特征；
    按照预设的编码特征类心对初始编码特征进行特征归类，并统计每个类的数目，形成最终的编码特征信息。
  3. 如权利要求2所述的图像识别方法，其特征在于，所述图像编码采用如下一种或多种编码方法：局部二值模式、局部相位量化、图像特征值统计。
  4. 如权利要求2所述的图像识别方法，其特征在于，所述特征归类采用如下一种或多种聚类方法：k均值、层次聚类、自组织映射、模糊C均值聚类。
  5. 如权利要求2所述的图像识别方法，其特征在于，所述预设的编码特征类心为预先设置或者通过在线的方式实时更新。
  6. 如权利要求1所述的图像识别方法，其特征在于，所述步骤S4在获得比对一致的图像模板后，与步骤S3并行执行。
  7. 如权利要求6所述的图像识别方法，其特征在于，所述步骤S4中当第二特征信息比对成功，则结束识别流程，并输出识别结果。
  8. 如权利要求6所述的图像识别方法，其特征在于，所述步骤S3中，当所述第一特征信息与注册的所有图像模板中的第一特征信息均比对不一致，则结束识别流程，并输出识别结果。
  9. 如权利要求1所述的图像识别方法，其特征在于，所述步骤S2之前还包括：
    S5，对获取的图像信息进行预处理。
  10. 一种图像识别方法，其特征在于，所述方法包括以下步骤：
    S01，获取待注册的图像信息；
    S02，对获取的图像信息进行特征提取，获取第一特征信息和第二特征信息；
    S03，根据第一特征信息和第二特征信息，形成相应的图像模板；
    S04，判断注册采集次数是否达到预设阈值；是则执行步骤S05，否则返回执行步骤S01；
    S05，将所有的图像模板进行存储，以用于对待识别的图像信息进行识别。
  11. 如权利要求10所述的图像识别方法，其特征在于，所述步骤S05之后包括：
    通过一图像采集单元对目标物体的图像信息进行采集，采用如权利要求1-9任一项所述的图像识别方法中的步骤对待识别的图像信息进行识别。
  12. 一种图像识别装置，其特征在于，所述图像识别装置包括处理器和存储单元；所述处理器用于执行如权利要求1-9任一项所述的图像识别方法；或者用于执行如权利要求10或11所述的图像识别方法。
  13. 一种电子装置，其特征在于，所述电子装置包括图像采集装置和处理器；所述图像采集装置用于在目标物体放置于电子装置上时，采集所述目标物体的图像信息；所述处理器用于执行如权利要求1-9任一项所述的图像识别方法；或者用于执行如权利要求10或11所述的图像识别方法。
  14. 一种计算机存储介质，其特征在于，所述计算机存储介质存储有计算机可执行指令，所述计算机可执行指令用于执行权利要求1-9任一项所述的图像识别方法；或者用于执行权利要求10或11所述的图像识别方法。
PCT/CN2017/085233 2017-05-20 2017-05-20 图像识别方法、图像识别装置、电子装置及计算机存储介质 WO2018213946A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2017/085233 WO2018213946A1 (zh) 2017-05-20 2017-05-20 图像识别方法、图像识别装置、电子装置及计算机存储介质
CN201780000357.1A CN107278308A (zh) 2017-05-20 2017-05-20 图像识别方法、图像识别装置、电子装置及计算机存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2017/085233 WO2018213946A1 (zh) 2017-05-20 2017-05-20 图像识别方法、图像识别装置、电子装置及计算机存储介质

Publications (1)

Publication Number Publication Date
WO2018213946A1 true WO2018213946A1 (zh) 2018-11-29

Family

ID=60076513

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/085233 WO2018213946A1 (zh) 2017-05-20 2017-05-20 图像识别方法、图像识别装置、电子装置及计算机存储介质

Country Status (2)

Country Link
CN (1) CN107278308A (zh)
WO (1) WO2018213946A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110222789A (zh) * 2019-06-14 2019-09-10 腾讯科技（深圳）有限公司 图像识别方法及存储介质
CN112330728A (zh) * 2020-11-30 2021-02-05 维沃移动通信有限公司 图像处理方法、装置、电子设备以及可读存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108416262A (zh) * 2018-01-25 2018-08-17 杭州电子科技大学 一种基于多特征值的指纹图像特征匹配算法
CN110199295A (zh) * 2019-04-04 2019-09-03 深圳市汇顶科技股份有限公司 指纹识别的方法、装置和电子设备
TWI792017B (zh) * 2020-07-01 2023-02-11 義隆電子股份有限公司 生物特徵的辨識系統及辨識方法
CN111860279A (zh) * 2020-07-14 2020-10-30 武汉企秀网络科技有限公司 图像识别方法及装置、计算机存储介质
WO2024030105A1 (en) * 2022-08-02 2024-02-08 Havelsan Hava Elektronik San. Ve Tic. A.S. Multi-stage fusion matcher for dirty fingerprint and dirty palm

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101079102A (zh) * 2007-06-28 2007-11-28 中南大学 基于统计方法的指纹识别方法
CN101145198A (zh) * 2007-09-21 2008-03-19 清华大学 指纹识别中确定性编码方法与系统
CN101482916A (zh) * 2008-01-08 2009-07-15 祥群科技股份有限公司 手指热像的身份辨识方法
CN101604385A (zh) * 2009-07-09 2009-12-16 深圳大学 一种掌纹识别方法和掌纹识别装置
US20120257802A1 (en) * 2011-04-07 2012-10-11 Kwon Dongjin Apparatus and method for generating representative fingerprint template
CN105814585A (zh) * 2016-03-17 2016-07-27 深圳信炜科技有限公司 指纹处理方法、指纹处理装置、指纹识别系统及电子设备

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101540000B (zh) * 2008-03-20 2011-07-27 中国科学院自动化研究所 基于纹理基元统计特性分析的虹膜分类方法
CN109800741B (zh) * 2015-11-13 2023-07-14 Oppo广东移动通信有限公司 指纹注册方法、装置和终端设备
CN106055961B (zh) * 2016-05-31 2019-02-05 Oppo广东移动通信有限公司 一种指纹解锁方法及移动终端
CN106096585A (zh) * 2016-06-29 2016-11-09 深圳市金立通信设备有限公司 一种身份验证方法以及终端
CN208689589U (zh) * 2017-05-20 2019-04-02 深圳信炜科技有限公司 图像识别系统及电子装置
CN107358144A (zh) * 2017-05-20 2017-11-17 深圳信炜科技有限公司 图像识别系统及电子装置
CN107358145A (zh) * 2017-05-20 2017-11-17 深圳信炜科技有限公司 图像传感器及电子装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101079102A (zh) * 2007-06-28 2007-11-28 中南大学 基于统计方法的指纹识别方法
CN101145198A (zh) * 2007-09-21 2008-03-19 清华大学 指纹识别中确定性编码方法与系统
CN101482916A (zh) * 2008-01-08 2009-07-15 祥群科技股份有限公司 手指热像的身份辨识方法
CN101604385A (zh) * 2009-07-09 2009-12-16 深圳大学 一种掌纹识别方法和掌纹识别装置
US20120257802A1 (en) * 2011-04-07 2012-10-11 Kwon Dongjin Apparatus and method for generating representative fingerprint template
CN105814585A (zh) * 2016-03-17 2016-07-27 深圳信炜科技有限公司 指纹处理方法、指纹处理装置、指纹识别系统及电子设备

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110222789A (zh) * 2019-06-14 2019-09-10 腾讯科技（深圳）有限公司 图像识别方法及存储介质
CN110222789B (zh) * 2019-06-14 2023-05-26 腾讯科技（深圳）有限公司 图像识别方法及存储介质
CN112330728A (zh) * 2020-11-30 2021-02-05 维沃移动通信有限公司 图像处理方法、装置、电子设备以及可读存储介质

Also Published As

Publication number Publication date
CN107278308A (zh) 2017-10-20

Similar Documents

Publication Publication Date Title
WO2018213946A1 (zh) 图像识别方法、图像识别装置、电子装置及计算机存储介质
CN107209855B (zh) 通过指纹识别认证用户
US9141844B2 (en) System and method for three-dimensional biometric data feature detection and recognition
US9202104B2 (en) Biometric information correction apparatus, biometric information correction method and computer-readable recording medium for biometric information correction
WO2010119500A1 (ja) 生体情報登録装置、生体情報登録方法及び生体情報登録用コンピュータプログラムならびに生体認証装置、生体認証方法及び生体認証用コンピュータプログラム
KR101632912B1 (ko) 지문 인식을 이용한 사용자 인증 방법
CN105659253B (zh) 指纹提取方法和装置
US10552596B2 (en) Biometric authentication
CN111201537A (zh) 在指纹分析中通过机器学习来区分活体手指与欺骗手指
CN105303174B (zh) 指纹录入方法及装置
CN107077279B (zh) 一种压力检测的方法及装置
El_Rahman Multimodal biometric systems based on different fusion levels of ECG and fingerprint using different classifiers
CN107358144A (zh) 图像识别系统及电子装置
CN107408208B (zh) 用于对用户的生物测量进行分析的方法和指纹感测系统
CN107358145A (zh) 图像传感器及电子装置
WO2018213947A1 (zh) 图像识别系统及电子装置
Xie et al. Fingerprint quality analysis and estimation for fingerprint matching
JP6241230B2 (ja) 生体情報判定装置及びプログラム
WO2018213945A1 (zh) 图像传感器及电子装置
Yu et al. Fingerprint image preprocessing based on whole-hand image captured by digital camera
Khalil et al. Authentication of fingerprint biometrics acquired using a cellphone camera: a review
Patel et al. Fingerprint matching using two methods
Sehgal Palm recognition using LBP and SVM
Zhang et al. 3D fingerprint reconstruction and recognition
US11460928B2 (en) Electronic device for recognizing gesture of user from sensor signal of user and method for recognizing gesture using the same

Legal Events

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

Ref document number: 17911034

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17911034

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 27/05/2020)

122 Ep: pct application non-entry in european phase

Ref document number: 17911034

Country of ref document: EP

Kind code of ref document: A1