From key unlocking to fingerprint unlocking to face unlocking, the development of biometrics began to focus on liberating human hands. At the same time, biometrics technology has begun to serve every aspect of people's lives. The era of brushing the face is the straightforward example.
Baidu solves the problem, biometric recognition is based on human physiological characteristics (face, fingerprint, iris, etc.) and behavioral characteristics (attitude, movement, emotion, etc.) to achieve identity authentication technology. In the identification of human identity, it is closely combined with high-tech means such as optics, acoustics, biosensors and biostatistics, and uses the inherent physiological and behavioral characteristics of the human body to identify individuals. At present, commonly used biometric technologies mainly include: face recognition, fingerprint recognition, iris recognition, behavior recognition and gait recognition.
Face recognition, also known as face recognition, is a biometric recognition technology based on human facial feature information for identification. Because face recognition involves many organs and large areas, the recognition technology is not only complicated but also susceptible to interference from many factors, such as human expressions, gestures, etc., and the micro-expression recognition and emotion recognition mentioned now. Technology is a branch of face recognition technology.
HFSecurity 2019 NEW Face Recognition Products
Related reports indicate that the size of the face recognition market is expected to be 2019 as the government introduces increased user and data security initiatives, the use of more and more mobile devices, and the growing global demand for robust fraud detection and prevention systems. At $3.2 billion, the market will reach $7.9 billion in five years and then in 2024, with a compound annual growth rate of 16.6%. In other words, face recognition technology has become the mainstream biometric technology.
From a technical point of view, the face recognition system mainly consists of four parts, namely face image acquisition and detection, face image preprocessing, face image feature extraction, and matching and recognition. At present, with the improvement of various computing powers and algorithms, and the unstructured data gradually transform to the level of architecture, the accuracy of face recognition algorithms has reached an average of 99.69%, which means that face recognition technology has reached a high level. And in China. The face of the “Four Little Dragons” of Chinese face recognition represented by Yun from Science and Technology, Shangtang Technology, Yitu Technology and Defiance Technology has emerged.
From an application perspective, in China alone, with the promotion of Skynet, Xueliang and national policies, China has become one of the countries that have benefited from face recognition technology. Software and hardware products based on face recognition technology also continue. Updates, such as face down, face recognition mobile phone unlock, face recognition access control and attendance, authentication and unity verification, face recognition capture fugitives, pedestrian red light grab system, Alipay face recognition payment, hospital online card, etc. Etc., various "smart +" industries, such as smart security, smart education, smart hospitals, smart finance, smart transportation, etc., have adopted face recognition technology as the mainstream technology.
Although the world has accelerated into the “brushing era”, face recognition development still faces many challenges:
● 2D face recognition technology
Due to the limitation of deep data loss in 2D information, it is impossible to express the real face completely, so its robustness to complex faces is not good. There are also many difficulties in practical applications, such as cross-view (cross Attitude, age, makeup, etc.) face recognition, low-resolution face, deep learning (sample deception), face anti-spoofing, complex scenes (strong and weak illumination, blur, occlusion, etc.), external changes, computational power Restricted, fine-grained classification (non-human detection and agent detection) and more.
●3D face recognition technology
At present, 3D face acquisition technology includes binocular technology, structured light technology and TOF technology, and 3D live detection can effectively distinguish forged images, videos, masks, etc., but it also has many challenges in terms of algorithms and hardware. In terms of algorithms, it is necessary to solve how to obtain high quality 3D face images (depth image, point cloud image, grid image and how to solve the influence of face pose, expression, occlusion and other factors; at the hardware level, there are Challenges include lighting, distance, precision, complexity, and more.
In addition to the above problems, face recognition technology also faces the problems of multi-modal technology fusion, artificial intelligence chip monopoly, and data privacy issues.
Although face recognition is a popular biometric technology in the market, fingerprint recognition is still a widely used recognition method.
Fingerprint recognition is mainly through the comparison of the detailed feature points of different fingerprints. The process can be summarized as three parts: fingerprint acquisition, fingerprint feature extraction and fingerprint matching. At present, the development of the fingerprint identification industry can be summarized as the following major trends:
● The fingerprint identification industry includes chip design, chip manufacturing and packaging, module manufacturing, and complete machine manufacturers. The industry chain is improving.
● In recent years, more and more smartphones have begun to adopt fingerprint recognition technology. In 2013, the smartphone iPhone 5S with fingerprint recognition function and Huawei were launched;
● With the promotion and popularization of the full-screen mobile phone solution, the fingerprint recognition sensor starts to turn to the back of the back or built in the bottom of the display.
● Among all types of biometrics, fingerprint recognition still occupies the highest share. In 2017, the global fingerprint identification technology scale was about 7.456 billion US dollars, an increase of 7.71%.
It is foreseeable that with the development and improvement of related technologies, fingerprint recognition will be widely used in smart terminals, ID cards, motor vehicles, homes and other fields. But fingerprint recognition still has a lot of pain points to be solved:
●Quality assessment: fingerprint recognition relies heavily on the quality of fingerprint images obtained. It is very important to solve the problem of fingerprint quality assessment.
● Data compression: The fingerprint database has a large amount of stored data. When fingerprint recognition is performed, all fingerprints need to be matched and scored. How to compress data to achieve fast matching is worth considering.
● Authenticity identification: fingerprint information is easy to be counterfeited, and the fingerprint is false and difficult to distinguish.
● Privacy security: At present, most fingerprint identification systems do not perform non-reversible encryption. Once the fingerprint information is leaked, it will cause loss to the user and even threaten public safety.
● Fingerprint acquisition hardware and chip: The current screen fingerprint recognition technology is mainly used in OLED panels, which is difficult to do in liquid crystal panels (LCD).
In addition, fingerprint recognition does not actually have lifetime "uniqueness" and "each person's fingerprint is not unique": people with blood relationship will have extremely similar fingerprints, and human fingerprints will change with age and skin condition. The shape of the fingerprint will also be affected a lot.
Iris recognition technology is based on the iris in the eye for identification, applied to security equipment (such as access control, etc.), and places with high security requirements.
Iris recognition has high stability, accuracy and safety. Compared with fingerprint recognition and face recognition, it has obvious advantages. However, the technical difficulty of iris recognition is higher, and the production cost is higher than other biometric technologies. The requirement to identify distance is high, and these factors hinder their entry into the general consumer market to some extent.
In addition, the current development of iris recognition technology in China is also limited by many factors, one is the acquisition of technology and equipment, and the other is the national policy factor. For the time being, the iris collection behavior in China is still in its infancy. To achieve the same wide application as face recognition requires strong support from national policies. In addition, because iris recognition requires a more special acquisition lens than face recognition, Moreover, it is vulnerable to external environment and its popularity has yet to be improved.
However, fortunately, with the support of relevant departments, China's iris image recognition and identification work has begun, the iris recognition market will certainly be expected in the future!
With the continuous maturity and development of face recognition, fingerprint recognition and iris recognition technology, biometrics technology has gradually become a behavior recognition technology based on attitude estimation and motion recognition.
In a nutshell, behavior recognition is an algorithm that structures the main activity skeleton of a person, defines various behaviors according to the human motion trajectory, and forms an action system through deep learning algorithms, which can be efficiently recognized by the system. . The recognition technology is photographed by a camera, simulates a skeleton of a person, and analyzes a motion trajectory for various movements of a person, thereby judging which motion the motion trajectory belongs to. Then through the background server for calculation, once the management action set by the system is matched, the system will immediately alert to achieve the purpose of early warning.
At present, it benefits from several advantages of behavior recognition technology: many actions can achieve zero false positives, can accurately identify abnormal behaviors of people in the scene, high efficiency of server identification analysis, and the same camera can simultaneously analyze N Abnormal behavior, this technology can analyze the anomalous behaviors such as emergency help, fight fights, high-altitude parabolic objects, and crowd watching, and early warning, can be widely used in autonomous driving, medical, education, robotics, public safety, film and television entertainment and other fields.
However, technology always has defects. Although behavior recognition technology gradually brings security closer to active defense, the development of current behavior recognition technology faces algorithmic challenges and hardware challenges.
● Algorithmic challenges: lack of end-to-end models, diversity of human postures and movements, complex scenes, lack of well-marked large data sets, individual differences (different people's performance of unified movements);
● Hardware Challenge: There are certain requirements for high-precision, miniaturized sensors and high-computing, low-power chips.
In general with behavior recognition, gait recognition is also a rising star in the field of biometrics.
Gait recognition is designed to identify people by walking posture. From an anatomical point of view, the physical basis of gait uniqueness is the difference in the physiological structure of each person, the length of the leg bones, the different muscle strength, the different heights of the center of gravity, and the different sensitivity of the motor nerves. Determine the uniqueness of the gait. In addition, because the technology supports long-distance identification, no need for hard coordination, and strong environmental adaptability, it has begun to enter the security, transportation, industrial and other industries to develop related applications.
From a technical point of view, gait is a dynamic feature compared to static biometrics such as fingerprints, faces, palm prints, veins, etc., and is therefore more complicated in the recognition process. The whole process of gait recognition is divided into four parts: acquisition, analysis, extraction and comparison. However, in fact, each link is facing challenges. For example, how to collect data samples, how to obtain data, and how to build a database for gait recognition? How to segment the foreground and background after getting the data to make the identification more accurate? In the stage of feature expression, how to solve the problem of cross-view recognition and so on.
As an emerging biometric technology, gait recognition needs to go from the laboratory to the commercial scene. Gait recognition should not only be based on recognition accuracy, recognition speed, technology application cost and convenience, but also It is highly integrated with the industry and has gained industry recognition.
Relevant reports show that by 2020, the global biometric identification market will exceed 25 billion US dollars, and China's biometric identification market will reach 30 billion yuan.
More than the five recognition technologies mentioned above, biometrics include voiceprint recognition, eye pattern recognition, retina recognition, vein recognition, body recognition, and the like. The challenges faced by all these biometric technologies are derived from three aspects: algorithms, hardware, and laws and regulations: at the algorithm level, large data sets that need to be well-labeled are required to satisfy the interpretability of the deep learning model and the complexity of the actual application scenario; At the hardware level, we need to focus on the design and manufacture of sensors, the design and manufacture of chips, and the real-time computing of mobile devices. Finally, the state and relevant governments must start with laws and regulations, and formulate relevant policies to protect user privacy. Industry Standard.
It is undeniable that from fingerprint authentication to face recognition and iris recognition, biometric technology is moving into the “visual era”. It is a fact that technology is constantly improving, but behind this “vision of vision”, multiple biometrics is the king of future biometric technology development.
Compared with single biometric technology, multiple biometric technologies combining face recognition, behavior recognition, gait recognition with passwords, fingerprints, and iris will bring higher reliability and accuracy, while protecting users. Private issues.
Looking to the future, multiple biometric technologies will move toward more and more applications and markets.
Facebook: Huifan Technology
Youtube: Huifan Technology