The Biometrics system uses human biometrics as an identification condition. It has the characteristics of “everybody is different, has the same life span†and “carry-onâ€, so it has the characteristics of being unable to counterfeit and borrow, not afraid of losing, not carrying, not It will be forgotten and has the characteristics of uniqueness, uniqueness, and security of individual characteristics. It applies to the protection of high-confidentiality sites. The main categories of human biometric identification systems are physiological features (such as fingerprints, palm prints, face images, and irises) and behavioral features (such as voice, handwriting, gait, etc.). Here mainly talk about human biometrics based on physiological characteristics.
Fingerprint identification fingerprint identification equipment mainly includes live fingerprint image acquisition, fingerprint extraction features and fingerprint comparison. Its application is divided into Verification and Identification. Verification is a one-to-one comparison of on-site fingerprints with registered fingerprints, and identification is to find fingerprints in the fingerprint database that match the on-site fingerprints, which is one-to-many matching. Fingerprinting is used for a wide range of applications, including access control, Internet security, financial and commercial retailing.
Fingerprint recognition attendance machines usually preprocess images when collecting fingerprints. The main purpose of preprocessing is to transform the original grayscale image containing noise into a binary image suitable for feature extraction, which is the basis of the entire recognition algorithm and has an important influence on the performance of the system. Fingerprint image preprocessing methods mainly include steps of calculation, image enhancement, binarization and refinement. Among them, the pattern calculation is to obtain the direction information of the fingerprint image; the purpose of the image enhancement is to reduce the noise, enhance the contrast between the ridge line and the valley line in the fingerprint image, and improve the image quality; the meaning of the binarization is the ridge line of the fingerprint. The valley lines are denoted by 1 and 0 respectively; the purpose of the refinement is to remove the redundant width of the fingerprint ridge line as much as possible so that the width becomes a pixel width, and the thinning can remove a large amount of redundant information while retaining The topological connection of fingerprints facilitates subsequent feature extraction.
Palm-shaped palm recognition uses the shape of the palm of the hand in three-dimensional space, the length of four fingers, the width and thickness of the palm, and the width and height of the two joints of each finger as a condition for discrimination. The length and width of the hand are obtained by looking through the palm. The data is obtained from the side view of the thickness data of the hand, and the resulting hand outline data is finally converted into identification vectors of a plurality of character lengths, which are stored as user templates. The hand shape recognition is invasive and saves resources. It only requires 9 bytes (72 bits), which can be combined with the access control system and can be widely used.
Palmprint palmar vein recognition palmar vein recognition is a living body authentication technology. It uses active illumination of a near-infrared light source and uses human hemoglobin to absorb infrared light of a specific wavelength band to obtain an infrared image of the human subcutaneous vein network, which is not easily affected by palm peeling or injury. , moist and other external factors. However, since the thickness of the palm skin of different people is different, the sharpness of the obtained vein image is different, which has a great influence on the recognition accuracy. Under normal circumstances, the images captured by the palm recognition device include palmprint information, and the palmprint information assists in palm identification.
Iris Recognition In all biometric technologies, iris recognition is the most convenient and accurate one for current applications. The iris of the eye is composed of rather complex fibrous tissue. The detailed structure of the iris is determined in a random combination before birth. The iris recognition technology converts the visible characteristics of the iris into the iris code (Iris Code). This code template is Store it for later identification. In 2005, the State Key Laboratory of Pattern Recognition at the Institute of Automation of the Chinese Academy of Sciences obtained the second prize for national technology invention for its outstanding achievements in "Iris Image Acquisition and Recognition Technology".
The iris recognition iris can be positioned in less than 1 second. The time required to generate the iris code is only 1 second. The database retrieval time is also very fast. There are thousands of iris information databases. The time spent on searching is not much, and some people may question the speed so fast. In fact, the algorithm of iris recognition technology is also constrained by the existing technology. We know that processor speed is a bottleneck for large-scale retrieval, and the performance of the network and hardware devices also limits the speed of retrieval. Of course, because the iris recognition technology uses monochrome imaging technology, it is difficult for some images to separate it from the image of the pupil. However, the algorithm used by the iris recognition technology allows the image quality to vary to some extent. The Iris Code generated by the same iris also has a 25% change, which sounds like the weakness of this technique, but in the recognition process, this Iris Code changes only the entire iris. 10% of the code, the proportion of code it occupies is quite small.
Face recognition Faces are the most effective distinguishing parts when compared to human features. The recognized features are the shape and positional relationship of the outlines (head, chin, and cheeks) of the eyes, nose, mouth, eyebrows, and face, and the contour shadows of the face. available. It has the advantages of a "non-invasive system," and in particular, it can also be used for active search by specific individuals in public places. It is the most promising non-aggressive identification technology. In complex scenes, quickly and accurately determining which part of a person's face is a result of complex mathematical calculations. Using modern video inter-frame motion detection and analysis techniques, the part identified as a face is continuously "tracked" while it is continuously "segmented" and compared.
Face recognition technology is divided into two categories: two-dimensional face recognition and three-dimensional face recognition. The two-dimensional face recognition is based on the face plane image, but actually the face itself is three-dimensional. The face plane image is only a projection of a three-dimensional face in a plane. In this process, part of the information is bound to be lost. The further improvement of the two-dimensional face recognition performance has been adversely affected by factors such as ambient light, posture, and expression.
Three-dimensional face recognition is based on three-dimensional face images. Theoretically speaking, it possesses some technical advantages of three-dimensional image information. However, there are always collection devices that are expensive, the acquisition system is complex, the storage is high, the face reconstruction algorithm is very complex, and the recognition speed is relatively high. Slow and other shortcomings.
Face recognition is currently the leading domestic Hanwang binocular stereoscopic face recognition technology. Hanwang face recognition system uses dual cameras, the right side of the camera picture for face positioning and eye positioning, at the same time, the left side of the camera and the right side of the camera picture for stereo integration, restore the three-dimensional face model, based on this, Feature extraction and comparison to achieve template input and identification functions.
Fingerprint identification fingerprint identification equipment mainly includes live fingerprint image acquisition, fingerprint extraction features and fingerprint comparison. Its application is divided into Verification and Identification. Verification is a one-to-one comparison of on-site fingerprints with registered fingerprints, and identification is to find fingerprints in the fingerprint database that match the on-site fingerprints, which is one-to-many matching. Fingerprinting is used for a wide range of applications, including access control, Internet security, financial and commercial retailing.
Fingerprint recognition attendance machines usually preprocess images when collecting fingerprints. The main purpose of preprocessing is to transform the original grayscale image containing noise into a binary image suitable for feature extraction, which is the basis of the entire recognition algorithm and has an important influence on the performance of the system. Fingerprint image preprocessing methods mainly include steps of calculation, image enhancement, binarization and refinement. Among them, the pattern calculation is to obtain the direction information of the fingerprint image; the purpose of the image enhancement is to reduce the noise, enhance the contrast between the ridge line and the valley line in the fingerprint image, and improve the image quality; the meaning of the binarization is the ridge line of the fingerprint. The valley lines are denoted by 1 and 0 respectively; the purpose of the refinement is to remove the redundant width of the fingerprint ridge line as much as possible so that the width becomes a pixel width, and the thinning can remove a large amount of redundant information while retaining The topological connection of fingerprints facilitates subsequent feature extraction.
Palm-shaped palm recognition uses the shape of the palm of the hand in three-dimensional space, the length of four fingers, the width and thickness of the palm, and the width and height of the two joints of each finger as a condition for discrimination. The length and width of the hand are obtained by looking through the palm. The data is obtained from the side view of the thickness data of the hand, and the resulting hand outline data is finally converted into identification vectors of a plurality of character lengths, which are stored as user templates. The hand shape recognition is invasive and saves resources. It only requires 9 bytes (72 bits), which can be combined with the access control system and can be widely used.
Palmprint palmar vein recognition palmar vein recognition is a living body authentication technology. It uses active illumination of a near-infrared light source and uses human hemoglobin to absorb infrared light of a specific wavelength band to obtain an infrared image of the human subcutaneous vein network, which is not easily affected by palm peeling or injury. , moist and other external factors. However, since the thickness of the palm skin of different people is different, the sharpness of the obtained vein image is different, which has a great influence on the recognition accuracy. Under normal circumstances, the images captured by the palm recognition device include palmprint information, and the palmprint information assists in palm identification.
Iris Recognition In all biometric technologies, iris recognition is the most convenient and accurate one for current applications. The iris of the eye is composed of rather complex fibrous tissue. The detailed structure of the iris is determined in a random combination before birth. The iris recognition technology converts the visible characteristics of the iris into the iris code (Iris Code). This code template is Store it for later identification. In 2005, the State Key Laboratory of Pattern Recognition at the Institute of Automation of the Chinese Academy of Sciences obtained the second prize for national technology invention for its outstanding achievements in "Iris Image Acquisition and Recognition Technology".
The iris recognition iris can be positioned in less than 1 second. The time required to generate the iris code is only 1 second. The database retrieval time is also very fast. There are thousands of iris information databases. The time spent on searching is not much, and some people may question the speed so fast. In fact, the algorithm of iris recognition technology is also constrained by the existing technology. We know that processor speed is a bottleneck for large-scale retrieval, and the performance of the network and hardware devices also limits the speed of retrieval. Of course, because the iris recognition technology uses monochrome imaging technology, it is difficult for some images to separate it from the image of the pupil. However, the algorithm used by the iris recognition technology allows the image quality to vary to some extent. The Iris Code generated by the same iris also has a 25% change, which sounds like the weakness of this technique, but in the recognition process, this Iris Code changes only the entire iris. 10% of the code, the proportion of code it occupies is quite small.
Face recognition Faces are the most effective distinguishing parts when compared to human features. The recognized features are the shape and positional relationship of the outlines (head, chin, and cheeks) of the eyes, nose, mouth, eyebrows, and face, and the contour shadows of the face. available. It has the advantages of a "non-invasive system," and in particular, it can also be used for active search by specific individuals in public places. It is the most promising non-aggressive identification technology. In complex scenes, quickly and accurately determining which part of a person's face is a result of complex mathematical calculations. Using modern video inter-frame motion detection and analysis techniques, the part identified as a face is continuously "tracked" while it is continuously "segmented" and compared.
Face recognition technology is divided into two categories: two-dimensional face recognition and three-dimensional face recognition. The two-dimensional face recognition is based on the face plane image, but actually the face itself is three-dimensional. The face plane image is only a projection of a three-dimensional face in a plane. In this process, part of the information is bound to be lost. The further improvement of the two-dimensional face recognition performance has been adversely affected by factors such as ambient light, posture, and expression.
Three-dimensional face recognition is based on three-dimensional face images. Theoretically speaking, it possesses some technical advantages of three-dimensional image information. However, there are always collection devices that are expensive, the acquisition system is complex, the storage is high, the face reconstruction algorithm is very complex, and the recognition speed is relatively high. Slow and other shortcomings.
Face recognition is currently the leading domestic Hanwang binocular stereoscopic face recognition technology. Hanwang face recognition system uses dual cameras, the right side of the camera picture for face positioning and eye positioning, at the same time, the left side of the camera and the right side of the camera picture for stereo integration, restore the three-dimensional face model, based on this, Feature extraction and comparison to achieve template input and identification functions.
Tenyu Tools Co., Ltd. , http://www.dydrillbits.com