TW201937392A - System and method for biometric authentication in connection with camera-equipped devices - Google Patents

Abstract

Disclosed are systems for providing selective access to resources available in connection with a device comprising software executed on suitable computer hardware, said system comprising: at least one camera associated with said device, said camera being capable of taking at least one photograph of a human palm print; a detector module using local classifiers to locate and segment the region of interest of the palm without physical contact; a conversion processor which converts raw pixel data associated with said region of interest of a human palm print into a unique signature associated with said palm print; and an authentication and identification engine, said authentication and identification engine determining whether access to one or more of said resources should be granted based upon said unique signature and at least one database containing a plurality of user models.

Description

System and method for biometric measurement and identification connected with a device equipped with a camera

The present invention relates generally to the use of biometric metrology techniques for identification and identification, and more particularly to the identification and identification of users via computers (such as mobile devices) to selectively allow or deny A contactless solution for accessing various resources. In the present invention, the identification or recognition is performed by using one of the following key steps: one (1) using the local classifier to detect the palm print area; (2) extracting from the region of interest. And (3) calculating a matching score for a user model stored in a database dynamically expandable through a learning program.

Mobile devices such as smart phones, tablets and notebooks have become widely used and used by many people on a daily basis. These devices have become more powerful and as the developers have created more and more applications and services running on the applications, these devices have become more integrated into our daily lives. These mobile devices not only provide a powerful computing platform by themselves, but they also provide connectivity to one of the available platforms on a remote platform that typically accesses a cell site via a wireless link and then reloads to the backbone of the Internet. Unlimited set of services, applications and materials. In addition to accessing such remote platforms, mobile devices also have the ability to connect to other mobile devices via short-range and long-range wireless connections.
Perhaps most importantly, the ever-increasing penetration of such devices, along with the ever-decreasing cost associated with component parts in such devices, has resulted in devices that have become more capable while still affordable for many users. . For example, as a result of the cost reduction of component parts and the development of more powerful software, a large number of smart phones now include powerful cameras that can take extremely detailed photos of about 8 million or more pixels.
An important issue that arises in the context of mobile devices and their interaction with such versatile connections and their need to interact with so many different resources is the need to control access to each of these resources so that only Such individuals or devices that are authorized to access the applicable resources can actually do so. In the normal case, resource access is controlled by input of alphanumeric strings (such as user IDs and passwords). For example, a smartphone user may be required to enter a four bit code before being allowed access to any functionality on the device. In addition, each local application or other resource on the device may require the user to enter one or more alphanumeric strings before obtaining access resources. In this case, the correct data (user ID, password, etc.) can be stored in the memory of the device. Alternatively, the user and/or device may be required to send a set of correct text/numeric strings to the remote resource for accessing resources (applications, data, communication capabilities, etc.) remote from the device, which in turn Confirm that the sent data matches the correct data before allowing access to the resource.
As can be imagined, for a typical smartphone user, for example, the aforementioned techniques for authentication and identification have several drawbacks. As one of them, remember that the need for user IDs and passwords for so many different applications, services and other resources (each with its own requirements for how to construct such IDs and passwords) can be quite The fear of the user and the user often forget the ID and password of the resource that is infrequently accessed. Another disadvantage is that there is a security issue with using literal/numeric strings to control access to resources. For example, there are powerful software programs available for hacking such strings to gain unauthorized access to resources.
Moreover, the typical contact-based approach of a user entering their password and user ID on their smartphone screen using their finger poses a security risk. Experienced hackers are often able to "extract" fingerprint patterns from the screen based on grease residue left by fingers to gain unauthorized access. This is especially true in the context of inputting a short digit string such as one of four digits to unlock the device. Once the device is unlocked, it is not even possible to preserve many of the resources on the device, resulting in various security risks.
One solution to set a goal to eliminate or reduce the disadvantages discussed above relates to using biometric measurement techniques to control access to resources available via mobile devices. While these solutions have eliminated some of the shortcomings discussed above in some cases, they still suffer from several disadvantages. For example, some contact-based solutions require the user to place their finger on a device sensor that has the ability to capture the user's fingerprint, and then the fingerprint data for the local or remote location matches the user's Fingerprint to determine if there is enough to allow the user or device to access one of the one or more resources to match. In this case, as referenced above, a fingerprint may be extracted by a hacker from the device sensor and used to gain unauthorized access to one or more resources at the subsequent time using the fingerprint. Such solutions are also typically subject to the disadvantage that a busy user is unable to accept processing in the context of attempting to obtain access to many different resources on the device during a typical schedule to determine if the fingerprint is a match.
There are additional health issues associated with contact-based methods involving the transmission of bacteria, viruses or other biological hazards, especially in the case of sharing devices between users. As is known in the art, a person's fingertips (and, more generally, the hands of a person) are often one of the primary vehicles for delivering bacteria, viruses, or other biological hazards between people. In the case of an individual device shared by multiple people, one of the users uses their fingertips to type in a recognition string or to identify or identify the user through contact-based biometric measurement methods (such as fingerprint or palmprint recognition, etc.). Their own contact-based identification and identification methods create the risk of transmitting such biohazards via a shared contact medium.

Accordingly, it is an object of the present invention to provide a non-contact based biometric feature that supports accurate, secure and fast authentication and identification of users and devices to provide selective access through a camera-accessible device. Measurement system and method.
In an embodiment of the invention, such a camera-equipped device is required to identify or identify itself as one of the conditions for accessing one or more resources (sometimes referred to hereinafter as convenient for convenience) A smart phone, although such devices should be understood to include all devices having a camera performance, both mobile and stationary devices, such as desktop computers, use a smart phone camera to shoot their palms or two A photo or series of photos of a palm. Next, the system of the present invention analyzes the palm print image using computer vision technology and confirms that the palm print signature matches the user's model in a database (user authentication) or finds a match in many models in a database. Model (user identification).
Further features and aspects of the present invention will become apparent from the Detailed Description of the <RTIgt;

Reference will now be made in detail to the exemplary embodiments embodiments It is to be understood that the following discussion of the exemplary embodiments is not intended to be a limitation of the invention. The following discussion is provided to provide the reader with a more detailed understanding of the specific aspects and features of the invention.
Before the embodiments of the present invention are described in detail, it is to be understood that Unless otherwise defined, all terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. Although any methods and materials similar or equivalent to those described herein can be used in the practice of the present invention, the preferred methods and materials are now described. All publications mentioned herein are hereby incorporated by reference to the extent of the disclosure of the disclosure of the disclosure of the disclosure. The present invention controls its extent of conflict with any incorporated disclosure.
The singular forms "a", "the" and "the" Thus, for example, reference to "a palm" includes a single palm or two palms of a person and references to "one image" include references to one or more images. In addition, use of terms that may be described using equivalent terms includes the use of such equivalent terms. Thus, for example, the use of the term "camera" shall be taken to include any device that is capable of obtaining an image of an object. As another example and as mentioned above, the term "smartphone" encompasses all devices that have a camera performance.
The description of the invention in the preferred embodiment of the invention will be as follows. Referring to Figure 1, a discussion of one of the key components of the system of the present invention and the interaction of each of these components with each other to derive the advantages of the present invention are as follows. Device 100 can be any device that includes a camera capable of taking a high quality photo. Preferably, the camera of device 100 also includes a flash element that can be selectively and quickly activated and deactivated for illumination of the captured area. Examples of such devices 100 include smart phones, tablets, notebooks, and any other device that can be carried by a user and that provides a computing platform that allows the functionality of the present invention to be operated, as well as a desktop Computer or various fixed embedded devices. Examples of such fixed embedded devices may include camera devices that are fixed to a facility portal or other strategic location that provides secure access to physical space or other resources, or camera devices that are fixed to strategic locations for purposes such as time and attendance agreements, and others application. Although not required, the device 100 may also include various other features such as an observation screen (which may also be a touch screen), a keypad, an accelerometer, GPS performance, storage capacity, and a central processing unit (CPU). .
The device 100 includes at least one camera 105 that is preferably capable of producing high quality photos of, for example, two million or more pixels. Camera data processor 110 receives image material from camera 105 and processes the image material as is known in the art to create pixels of photographs that can be used in a variety of ways, including for purposes of linking the present invention to the presently described invention. Information indicated. The data from camera data processor 110 is fed to region of interest detector 115, which is used to position the palm region within a wider image and to characterize the region using a high level of accuracy and consistency. The various individual images of the lighting conditions and the orientation of the palm to the camera provide a mask with a palm area of substantially the same shape and position on the palm of the hand.
In one embodiment of the region of interest detector 115, the local region classifier based on the sliding window is used to detect the region of interest to mark the palm and non-palm pixels by the classification score, followed by grouping the adjacent palm pixels into inputs. One of the steps of dividing the connected components in the image. Since a significant number of regions are learned from a large collection of sample images to capture the stable characteristics of the palm appearance to form a powerful classifier, accuracy and robustness to one of the high levels of image noise can be achieved. Thus, the Detector can accurately position and characterize areas of interest on input images in free form with various hand orientations and lighting conditions.
In one embodiment of the region of interest detector 115, the local classifier based on Haar Wavelets and AdaBoost (Reference 1) is used to detect the region of interest in the palm region of a user's hand. In another embodiment of the region of interest detector 115, the local classifier based on the support vector machine (reference 2) is used to detect the region of interest in the palm region of a user's hand. In another embodiment of the region of interest detector 115, a gyroscopic neural network is used to detect regions of interest in the palm region of a user's hand, such as in U.S. Patent Nos. 5,067,164 and 5,058,179. (Refer to Reference 3 and Reference 4).
Next, the region of interest detector 115 feeds image data (including the palm region mask) to the conversion processor 120 for extracting from the image block representing the characteristic features of the palm region of the individual for distinguishing between the individual and another user. A signature 125, wherein the image blocks are small sampling windows within the mask of the palm area.
In one embodiment, the signature 125 calculates one of the vectors as follows. First, a histogram of the edge orientations in several selected regions in the image is calculated. This can be performed using one of the well-known methods of computer vision for extracting the local image descriptor, such as a variable invariant feature transform (see reference 5), a directional gradient histogram (see reference 6), and in the art. Other references are known. Second, for example, using the well-known K-means cluster algorithm, each orientation histogram and several prototypes that have been calculated from the training data. Finally, a signature vector is formed such that the component k of the vector corresponds to the kth prototype mentioned above. Component k contains the number of histograms closer to the prototype k than to all other prototypes. This sequence of operations is known in the literature as a "Bag of Features" representation (see, for example, Reference 7). It should be understood from the present teachings that in another embodiment of the invention, a plurality of feature packages can be used to preserve geometric relationships between local regions.
The signature 125 is then fed to an authentication and recognition engine (AID engine) 130 for implementing many of the key procedures of the present invention, as described below. The AID engine 130 communicates with the user model repository 135 (if present) to store a local copy of a user model. Thus, in the case where the local end resides on the device 100 and does not require an application or service to communicate externally, for example, to a remote server or remote device, the palmprint image captured by the camera 105 can be compared. A user signature and a known user model stored in the database of the user model 135 for identification or identification. A user model is a statistical model calculated from a collection of one's palm images, from which the signatures are derived from the images defining the model. In one embodiment, the user model consists of a so-called Gaussian density model, one of the signatures calculated from the reference image of the user. Given the signature of the query image S, the user model is used to calculate a match score.

Where M _i and V _i are the average and variance of the i-th component of the signature vector of all reference images of the given user, and u is a small constant. If the match score R is greater than one of the pre-selected thresholds for this user model, the signature is considered to match the user model. The identification and recognition engine 130, the model building engine 155, and the user model database 135 form an AID unit 160.
The signature 125 is also fed to the model building engine 155 to initialize the user model during the first login of the user or to selectively incorporate the information of the signature when the model is already present to define the database 135 stored in the user model. User model. In one embodiment of the invention, the model building engine 155 updates the above mentioned averages and variances M _i and V _i using signatures extracted from the user's new image.
Device 100 also preferably includes a remote resource interface 145 that is in communication with AID engine 130. The remote resource interface 145 acts as an interface between the authentication and identification functionality implemented on the device 100 and the same functionality occurring on external/remote resources, such as remote servers and remote devices. Thus, for example, the remote resource interface 145 interacts with an application residing on the remote server to coordinate authentication or identification as needed by the remote application. This may include managing and responding to requests by external resource management or identification for a user operating device 100 or for authentication or identification of device 100 itself.
The remote resource interface 145 can communicate with the network interface 150 to transmit and receive data linked to the authentication and identification activities. Various wireless communication protocols can be used, including radio frequency and others (including and without limitation, Bluetooth and other near field communication technologies). In a preferred embodiment of the invention, by way of encryption and/or other methods (which reduce or eliminate user data and other information associated with the authentication and identification methods of the present invention, may be unauthorized The possibility of party interception) preserves the information communicated back and forth from device 100 over a public wireless link as is known in the art. Network interface 150 typically includes one of the radio frequency transceiver modules as is known in the art and allows device 100 to communicate with wireless network 400 via a wireless link. In turn, wireless network 400 typically retransmits data transmitted or received by device 100 to data network 500, again as is known in the art.
By way of example only, the present invention allows the user or device 100 itself to authenticate or identify the device 100 by remote servers and applications and other resources residing on the remote server. As illustrated in FIG. 1, remote server 200 can communicate with device 100 via the communication paths discussed above. In this manner and as controlled by the remote resource interface 145 residing on the device 100, the AID unit 205 residing on the remote server 200 can request and receive authentication and identification data from the device 100 for use with the resident A comparison is made on the remote server 200 or by a known and validated user model accessible by it, as described more fully below. This authentication and identification performance provides selective access to one or more applications 210, data 215, and other resources residing on the remote server 200. The same performance may also provide selective access to the local resources 140 (including applications, materials, and/or other resources residing on the device 100 and wherein the local resources seek to access data or other resources away from the device 100) Happening).
In another embodiment of the invention, communication as discussed above may occur between device 100 and one or more remote devices 300. The remote device 300 can be the same or different device type than the device 100 and the authentication/recognition functionality can occur bidirectionally in accordance with the teachings of the present invention. In other words, the device 100 can respond to the authentication/identification request from the remote device 300 to access, for example, one or more applications 310 resident on the remote device 300 via the AID unit 305 on the remote device 300. / or information 315. Moreover, remote device 300 can receive and respond to authentication and identification requests initiated by device 100 for remote device 300 (or one of its users) to access resources residing on device 100. In some cases, device 100 and remote device 300 will require the identification and/or identification of the other before sharing the resource. For example, this can occur in the context of a secure communication between device 100 and a user of remote device 300.
Turning now to Figure 2, a method of user/device identification and/or identification in accordance with a preferred embodiment of the present invention will now be described. By way of initial discussion, the differences between authentication and identification in the context of the teachings of the present invention are first described.
In the case of authentication, the user presents one of the forms in the form of a user ID or user name and the system of the present invention confirms that the user is indeed the person he claims to be. Next, the system compares the corresponding models in the database of the signature and user models derived from one of the images or images of the user's palm. If they match, the user is authenticated. If the matches do not match, the user is rejected.
A flow chart for user authentication in accordance with the teachings of the present invention in a preferred embodiment is shown in FIG. As a first step, the user at device 100 can enter his or her name or other identifying information into device 100, or the identity of the user can be preloaded into device 100. Separately, the user uses the camera 105 of the device 100 to take a photo or collection of photos of the palm of his or her hands. Next, the camera data processor 110 sends the raw pixel data to the region of interest detector 115, which determines the palm region within the image. The masked palm area from the region of interest detector 115 is fed to a conversion processor 120 that derives the user's unique signature. This conversion function can be handled alternatively on a remote resource or partially on a remote resource and partially on the device 100. There is no direct contact between the imaged palm area and the device 100, using a high resolution image of the hand, in the case where the end user presents a free form and is in any orientation and does not have any special hardware other than a normal digital camera, The system of the present invention identifies individuals using a multi-step software solution involving feature extraction, processing features as user signatures, and matching user signatures to stored user signatures or user models, where: (i) Detecting and segmenting a single or multiple regions of interest from the input image to remove extraneous pixel data and isolating the palm region (ii) extracting several high dimensional sparse feature vectors from the image (see, for example, reference 8); (iii) The proximity feature vector is merged into a single compact signature for one of the more compact and robust image representations; and (iv) the plurality of image signatures are combined into one identity model for each individual user.
Next, the authentication and recognition engine 130 queries the user's model (based on previously presented user identification data) in the user model database 135. At this time, if the derived user signature matches the stored user model, the user is authenticated and allowed to access the desired resource or resource group. Or, if the user signature does not match the model, the user is denied access to the desired resource or resource group. The aforementioned functionality with respect to queries and matches can be performed alternately away from device 100.
In the case of recognition, the user presents only one palm print image or image set, and the recognition and recognition engine 130 compares all models or models in the signature and user model database 135 derived from the palm print image or images. A subset. If a match is found, the user is identified. If a match is not found, the user is unknown.
A flow chart for user identification is shown in FIG. In this case, as in the case of authentication, the user takes a photo or collection of photos of the palm of his hand. This data is again processed as pixels formed by the camera data processor 110 and sent to the region of interest detector 115 to determine the palm region within the image. The masked palm area from the region of interest detector 115 is fed to a conversion processor 120 that derives the user's unique signature and then the AID engine 130 compares the derived signature with the database stored in the user model 135. A subset of all models or models. The conversion and comparison functions referenced above may alternatively be processed on a remote resource or partially on a remote resource and partially on the device 100. In any event, if a match is found, the user is identified and can be granted access to a resource or group of resources. If no match is found, the user cannot be identified and will not be granted access to a desired resource or resource group.
Which mode to use (identification or identification) depends on the application. In general, authentication provides a higher degree of accuracy but provides a lower level of user experience due to an additional factor that requires a user to take additional steps to enter their identity. The second identity factor can take any of the general forms, such as a user name, user ID, password, unique employee ID, social security number, email address, various other biometric measurement modalities, and the like. In an embodiment of the invention, the second identity factor is a signature derived from a palm print image of the second hand of the individual, wherein the individual signatures of each of the two palm print images or image sets of the individual are used together Identification or identification.
It is important to note that in each of the situations (identification or identification) described above, not one of the models in the database 135 of the user model that is locally located in the device 100 matches a user signature and is replaced by Generating a signature by capturing an image or collection of images of one of the user's palms at device 100 may be directed to one or both of remote server 200 or one or more remote devices 300. A database contains one or several models to match. In this case, the user of device 100 will typically seek to access one or more resources residing at such remote platforms rather than local to locate one of the resources within device 100. By way of example, in the case of unlocking, for example, a smart phone, the process can be done locally at the smart phone/device 100, but if authentication is initiated (eg, with a remote-based application) Some portions of the process may be performed at a remote server 200, wherein the matching user model may be stored on the remote server 200 rather than locally stored on the smart phone. Additionally, it should be understood from this teaching that user models, signatures, and/or other biometric measurement data can be synchronized between any of AID units 160, 205, 305 to allow for device 100, remote server 200. The remote device 300 performs local authentication or identification without requiring the device 100, the remote server 200, or the remote device 300 to locally generate the user model, signature, and/or other creatures. Feature measurement data.
Returning now to Figure 2, it can be seen that in a preferred embodiment of the present invention, in step (1), the apparatus 100 is adapted to capture a user to be identified (step (2)) by the camera 105 (step (3)). A photo or a series of photos of the palm of your hand. A flash assembly (step (4)) can be embedded in the device 100 to provide the necessary pre-processing of the image, particularly when the flash assembly provides minimal light for focus area detection, feature extraction and signature processing for individual palm images. . Then, the palm area of the image is covered by the region of interest detector 115 (step (5)) and fed to the conversion processor 120 (step (6)) to convert the original pixel into a uniquely identified user signature. Chapter 125. The user signature is a database that contains one of the relevant identification information associated with the palm print image of the user and can be quickly and accurately matched to a large database of user models, such as a database of user models 135 or A database of one of the remote platforms (step (7)). One benefit of the derived user signature is that it is essentially impossible to reconstruct a user's palm image from one of the user models' databases. In step (8), the AID engine 130 compares the user signatures in the database of user signatures and user models from the palm image or image collection to identify or identify the user as applicable. The conversion and comparison functions referenced above may alternatively be processed on a remote resource or partially on a remote resource and partially on the device 100.
Turning now to FIG. 3, it can be seen that in the case where the authentication or identification is completed with respect to a remote resource, the communication between the device 100 and the remote resource preferably occurs on a secure connection, such as in this item. Known in the technology. This may involve one or more techniques known in the art to include, for example, strong encryption, public or private key encryption, digital certificates and/or digital signatures, and the like.
Now that the system and primary methods of the present invention have been described, additional novel features, such as various methods for preventing spoofing from authentication/identification links, and novel methods for encoding and exchanging transaction information with remote resources, will be discussed.
Deception protection is an important aspect of the present invention. For example, it prevents an intruder from using one of the palms to print a photo instead of a real hand for authentication. One novel aspect of the present invention relating to fraud protection involves detecting and using the three-dimensional characteristics of a person's hand in order to provide security against fraud.
In one example of fraud detection, to distinguish between a photo and a real hand, the system of the present invention takes a series of photos in a quick sequence in which the camera flash is used intermittently and for varying lengths of time. Photographing a 3D object (a real hand) with a flash will have a specific highlight area and a shadow produced by the flash, however in the position of the hand, where one of the hands is 2D (for example, in another action) These highlighted areas and shadows will not be displayed if one of the palms or one of the palm images is displayed on the display of the device. This allows the system of the present invention to utilize a comparison of highlights and shadows on the hand created between the flash photo and the non-flash photo to distinguish between a printed photo and a real hand. In this way, one of the photos of one of the palms of an authorized user is accidentally obtained. The unauthorized party cannot use the photo to gain unauthorized access to the local or remote resources.
A further method for detecting a real hand involves 3D modeling of the hand. In this case, the system of the present invention can prompt the user to turn their hand when taking a series of multiple photos. A real 3D object will show different parts of the hand as each successive image, whereas a 2D object will always show the exact same part of the hand, with only varying distortion. This allows the system of the present invention to distinguish between a printed photo and a real hand. Similarly, instead of rotating the hand, the user can be prompted to punch or punch from a fist when taking a series of photos. Other methods of distinguishing between a real hand and a one-handed photo are also possible.
Another novel aspect of the present invention is one of the methods in which replay attacks can be detected and prevented. In this case, an intruder modifies a mobile device such that it sends a photo or a series of previously recorded photos from a legitimate user to the network for authentication or identification rather than by camera. The image taken. It is assumed here that an intruder can take a photo of an authorized user's hand without being recognized by an authorized user or authorizing the user to prevent the situation. If this is in fact a risk (eg, one of the authorized users is sleeping or unconscious), then preferably one or more additional identity factors are required (such as a user ID or other independent of the palm print image). Formal information) This method of identifying a user uses the system.
To detect and defend against a replay attack, the system of the present invention emits a series of photos and flashes at various intervals, i.e., it records a series of photos, some of which turn off the flash and others turn on the flash. The particular photo can be selected at random or according to a predetermined sequence and the flash sequence can be turned on/off and the like can be changed for each authentication or identification request. The system of the present invention can easily detect whether an intruder uses a previously recorded series of photos because the photo and flash on/off mode will not match the mode actually sent to the mobile device.
Another method of detecting a replay attack involves storing all previously used images and comparing the new images to the database. Since the pixel data based on the images of the two different palm prints can be essentially identical or substantially identical in a particular tolerance, the system can detect when a previously captured image is used again. Other methods of detecting a replay attack are also possible.
Yet another novel aspect of the present invention is the ability to embed transactional information or other data at the timing of a series of photos and/or flash modes. This timing mode can be further used to encode information about the transaction itself. Then, a password hash can be applied to this information. The hash code makes the resulting code compact (short) and also prevents anyone viewing the flash mode from deriving any information about the original content of the code. In one embodiment of the invention, the timing of the sequence of image and/or flash modes is used as part of an anti-spoofing mechanism to determine whether the image sequence provided for authentication or recognition matches information from the transaction itself. A particular implementation can include:
1. Low-resolution video in one of the palm areas with flash mode.
2. One of the palm areas or several high resolution still images.
3. Computer vision technology to ensure that high-resolution images are from objects that are identical to objects in video.
Based on the above description of the systems and methods of the present invention, it will be appreciated that a variety of applications are possible. Examples include, without limitation, accessing one or more devices, accessing one or more applications, various transaction applications residing on or remotely located on a server or other remote device User authentication required (such as election voting, allocation of state benefits, financial payments) and any other type of transaction.
While the invention has been shown and described with reference to the embodiments of the invention
reference
[1] Paul Viola and Michael Jones, Rapid Object Detection using a Boosted Cascade of Simple Features, Proceedings of IEEE Computer Vision and Pattern Recognition, 2001, pp. I: 511-518.
[2] Corinna Cortes and Vladimir N. Vapnik, Support-Vector Networks, Machine Learning, 20, 1995.
[3] Yann LeCun, Léon Bottou, Yoshua Bengio, Patrick Haffner: Gradient-Based Learning Applied to Document Recognition, Proceedings of the IEEE, 86(11): 2278-2324, November 1998.
[4] Pierre Sermanet, Koray Kavukcuoglu, Soumith Chintala, and Yann LeCun: Pedestrian Detection with Unsupervised Multi-Stage Feature Learning, Proc. International Conference on Computer Vision and Pattern Recognition (CVPR '13), IEEE, June 2013.
[5] David G. Lowe, Distinctive Image Features From Scale-Invariant Keypoints, International Journal of Computer Vision, 60, 2 (2004), pp. 91-110.
[6] N. Dalal and B. Triggs, Histograms of Oriented Gradients for Human Detection. In Proceedings of Computer Vision and Pattern Recognition, 2005.
[7] Y-Lan Boureau, Jean Ponce, and Yann LeCun: A Theoretical Analysis of Feature Pooling in Vision Algorithms, Proc. International Conference on Machine Learning (ICML '10), 2010.
[8] Yann LeCun, Koray Kavukvuoglu, and Clément Farabet: Convolutional Networks and Applications in Vision, Proc. International Symposium on Circuits and Systems (ISCAS '10), IEEE, 2010.

100‧‧‧ device

105‧‧‧ camera

110‧‧‧ camera data processor

115‧‧‧Following Area Detector

120‧‧‧Transition processor

125‧‧‧Signature

130‧‧‧Acknowledgment and Identification (AID) Engine

135‧‧‧Database

140‧‧‧Local resources

145‧‧‧ Remote resource interface

150‧‧‧Internet interface

155‧‧‧Model Building Engine

160‧‧‧AID unit

200‧‧‧Remote Server

205‧‧‧AID unit

210‧‧‧Application

215‧‧‧Information

300‧‧‧ Remote device

305‧‧‧AID unit

310‧‧‧Application

315‧‧‧Information

400‧‧‧Wireless network

500‧‧‧Information Network

1 is a diagram depicting the major components of the system of the present invention in a preferred embodiment of the present invention;

Figure 2 is a block diagram useful in illustrating the method of the present invention in a preferred embodiment of the present invention;

3 is a diagram illustrating a secure connection between a mobile device and one or more remote servers in accordance with a preferred embodiment of the present invention;

4 is a flow chart illustrating one of the key steps in identifying a user or device in accordance with the present invention in a preferred embodiment of the present invention;

Figure 5 is a flow chart illustrating one of the key steps in identifying a user or device in accordance with the present invention in a preferred embodiment of the present invention.

Claims (21)

An anti-spoofing system for providing selective access to resources available for connection to a smart phone based on a deep learning method, the system comprising: (a) the smart phone includes a digital processor, a memory module, an operating system, and a non-transitory storage medium including instructions executable by the digital processor; (b) at least one camera associated with the smart phone and configured to capture a plurality of images comprising a flash image and a non-flash image of an anatomical feature of the first user, wherein the anatomy The feature does not physically contact the smart phone to capture the plurality of images; and (c) at least one flash component associated with the smart phone and configured to emit a flash during capture of the flash image and not to emit a flash during capture of the non-flash image; Where the digital processor performs the following: (1) analyzing the anatomical features using a sliding-window-based local classifiers and a classifier formed by discriminative local features learned from a sample image set. The flash image and the non-flash image are labeled with anatomical features and non-anatomical features by classification scores; (2) locating and segmenting a region of interest of the anatomical feature in the flash image and the non-flash image using a trained detector; (3) using a small sampling window in the region of interest to identify image patches that enclose the characteristic biometrics of the anatomical feature; (4) extracting a signature from the pixel level data of the image blocks, wherein the signature is unique to the first user and is used to distinguish the first user from a second user, and wherein the signature is extracted The signature includes creating a histogram of edge orientations in a plurality of image blocks; (5) determining a fraud attempt based on a three-dimensional characteristic of the anatomical feature and a plurality of stored user models, wherein the three-dimensional characteristic is based on comparing the flash image with the non-flash image, wherein the flash image and the non-flash image are The comparing includes determining a highlighted area and a shaded area in at least one of the flash image and the non-flash image to determine that at least one of the flash image and the non-flash image represents a 3D object or the anatomical feature a 2D representation; (6) rejecting the first user from accessing one or more of the resources based on the determination of the fraud attempt; and (7) storing the at least one of the plurality of images of the anatomical feature and the three-dimensional characteristic in the sample image set. The system of claim 1, wherein the digital processor further performs the at least one image selectively including the anatomical feature, the pixel level data of the image block, and the signature to improve the use using new data a model to augment the at least one database and the determination of the fraud attempt. A system as claimed in claim 1, wherein the digital processor further performs the use of descriptors extracted from the image blocks on the region of interest. A system of claim 3, wherein the descriptors are encoded as high dimensional sparse vectors and wherein the sparse vectors are combined into at least one group. The system of claim 1, wherein the signature is calculated by a Bag of Features or a plurality of signature packages. The system of claim 1, wherein the flash image and the non-flash image analyzing the anatomical feature use Haar Wavelets and AdaBoost algorithms. The system of claim 1, wherein the flash image and the non-flash image analyzing the anatomical feature use a support vector machine, a gyrotron network, or both. The system of claim 1, wherein the user models comprise a statistical model calculated from a set of anatomical feature images of the first user. The system of claim 1, wherein the user models comprise a mixture of a Gaussian density model or a Gaussian density model. The system of claim 1, wherein at least one of the resources is remote from the smart phone or resides in the smart phone. The system of claim 10, wherein the at least one of the resources comprises an application or a database. The system of claim 1, wherein the individual signatures of the two anatomical feature images of the first user are used together for the determination of the fraud attempt. A system as claimed in claim 1, wherein one or more of the resources are further denied based on one or more other modalities. The system of claim 13, wherein the one or more other modalities comprise one or more of the following: passcode, security question, fingerprint identification, facial recognition, iris recognition, writing signature identification, and other biometric measurements and Non-biological feature measurement mode. The system of claim 1, wherein the denying access to one or more of the resources is further based on selectively rejecting one or more users for one or more transactions. A system of claim 1, wherein at least two of the plurality of images are captured during movement of one of the anatomical features, and wherein the three-dimensional characteristic is further based on the movement. The system of claim 1, wherein the plurality of images comprises a sequence of flash and non-flash images, and wherein the determining of the fraudulent attempt is further based on the sequence of the flash and non-flash images and the previously recorded flash and non-flash images One of the sequences is compared between one of the sequences. The system of claim 1, wherein the determining of the spoofing attempt is further based on comparing at least one of the plurality of images with at least one of the images of the first user anatomy in the set of sample images. [0057] The system of claim 15, wherein a transaction information or other material is embedded in one or more time intervals between consecutive images, and wherein the determination of the fraudulent attempt is further based on the transaction information and one of the transaction rooms Comparison. The system of claim 1, wherein the capture of at least one of the plurality of images is separated from the capture of a subsequent image by a time interval, and wherein the determining of the fraud attempt is based on the time interval and the previously recorded continuous image One of the previous time intervals is compared. An anti-spoofing method implemented by a smart phone for providing selective access to a smart phone connection based on a deep learning method, the method comprising: (a) using at least one camera associated with the smart phone and configured to capture a plurality of images of a flash image and a non-flash image of an anatomical feature of the first user, wherein the anatomy The feature does not physically contact the smart phone to capture the plurality of images; and (b) using at least one flash component associated with the smart phone and configured to emit a flash during capture of the flash image; (c) analyzing the flash image of the anatomical feature and the non-flash image by using a localization classifier based on a sliding window and a classifier formed by the localization feature learned from a sample image set to classify the score Marking anatomical features and non-anatomical features of pixels; (d) locating and segmenting a region of interest of the anatomical feature in the flash image and the non-flash image using a trained detector; (e) using a small sampling window in the region of interest to identify image blocks containing characteristic biometric features of the anatomical feature; (f) extracting a signature from the pixel level data of the image blocks, wherein the signature is unique to the first user and is used to distinguish the first user from a second user, and wherein the signature is extracted Signature includes creating a histogram of edge orientation in a plurality of image blocks; (g) determining a fraud attempt based on a three-dimensional characteristic of the anatomical feature and a plurality of stored user models, wherein the three-dimensional characteristic is based on a comparison between the flash image and the non-flash image, wherein the flash image and the non-flash image The comparing includes determining a highlighted area and a shaded area in at least one of the flash image and the non-flash image to determine that at least one of the flash image and the non-flash image represents a 3D object or an anatomical feature a 2D representation; (h) refusing the first user to access one or more of the resources based on the determination of the fraud attempt; and (i) storing the at least one image of the anatomical feature in the sample image set.