Portal field news

Portal field news

in ,

📱 | Overseas Mobile Topics 249th Xiaomi screens the side of the display as a sub-display…


Overseas Mobile Topics 249th Xiaomi displays screen table with side display as sub-display ...

If you write the contents roughly
The front camera will be embedded under the display, and face recognition etc. can be used even during full screen display.

In February of every year, each smartphone manufacturer announces the 2 flagship model.Samsung is January ... → Continue reading

 My Navi News

Mynavi News is a comprehensive information news site with more than 100 categories updated daily. It consists of the latest news and analysis reports and columns from a wide range of genres such as business, digital, lifestyle, and entertainment to niche categories such as management, technology, railways, and cars.

Wikipedia related words

If there is no explanation, there is no corresponding item on Wikipedia.

Face recognition system

Face recognition system(The Kaoninshiki system,English: Facial Recognition System) IsSurveillance cameraIs a computer application for automatically identifying a person from a digital image of. In the live imagefaceExtract the part that seems to be, face imageDatabaseIdentification is performed by matching with.


Some face recognition algorithms identify by extracting salient features from face images. For example, the relative positions and sizes of face parts, the shapes of eyes, nose, cheekbones, and chin are used as features. Search for images with matching features using such features[1].. Another algorithm creates standardized data from a large number of face images, compresses each face image only by the difference from the standard data, and leaves only the data necessary for face recognition. The target image is compared with such face data[2].. One of the first successful face recognition systems was[3], Based on a template matching technique that used a set of compressed facial data and a set of salient facial features[4].

The recognition algorithms can be roughly classified into two types. There are a method of directly comparing visual features geometrically and a method of statistically digitizing an image and comparing the numerical value with a template.

The main face recognition algorithms are:Principal component analysisUsedPeculiar face,Linear discriminant analysis,Hidden Markov model, By the motivation of neurons.

3D face recognition

As a new trend in recent years, try to achieve accuracy that considers the invisible part3D face recognitionThere is. In this technique, a three-dimensional sensor is used to acquire the stereoscopic information of the face. From there, extract and use distinctive features such as the orbit, nose, and chin contour[5].

The advantage of 3D face recognition is that it is less sensitive to image brightness than other techniques. Also, face images from various angles can be recognized.[1][5].

3D face recognition is also weak against changes in facial expressions.Israel Institute of TechnologyThen change the facial expressionIsometric mappingResearching techniques[6].

Skin texture analysis

Recently, there is a tendency to apply the details of the appearance of the skin to face recognition. It identifies wrinkles and spots from images and digitizes them.[1].. This is called Skin texture analysis, and there is a report that when it is used for face recognition, the recognition rate is improved by 20% from the conventional 25%.[1][5].


The following image management software has a face recognition system built-in, and it is possible to extract a picture of the same person.

ラ イ ブ ラ リ

  • - Open Source. It is based on Google's FaceNet algorithm.
  • - Open Source.

System usage example


People's Republic of ChinaIs the world's largest integrated face recognition system with over 1 million unitsSurveillance cameraAs a networkSky netIs famous[7].

UK OfNewham LondonHas a face recognition system built into the surveillance camera system for the entire district.

ア ッ プ ルDeveloped bySmartphone-iPhone XThen,Face IDFace recognition system called2017It has been adopted from. By registering the user's face in the device in advance and identifying the user's face with the built-in camera,iPhoneIt can be used to unlock. In addition, purchases from Apple's online storeApple PayFace ID can also be used for authentication. At present, many Android smartphones are also using face recognition.

Other new uses for face recognition are being developed. For example, it is being considered for use in ATM security. In other words, instead of using a cash card and a personal identification number, an ATM takes an image of the face and matches it with the image of the face in the database. The same idea can be applied to login to various sites on the Internet.[1].

German Federal Police TheFrankfurt airportIs conducting a fully automatic immigration examination for voluntary registrants. The registration target isEuropean UnionOrスイスLimited to citizens of[8][Broken link].

Australian Customs used facial recognitionSmartGateHas introduced an automatic immigration system called[9]. The system compares each person's face to the images recorded on the electronic passport microchip to verify that the passport holder is who he or she is.

Hong Kong OfShenzhen Bay PortThen,Mainland ChinaWithAccess controlso,Automobile OfLicense plateThe world's first system that identifies drivers and recognizes facesNECIntroduced from[10].

2017May 10,Ministry of Justice(I.e.It is,Tokyo International Airport(Haneda Airport)ImmigrationToPanasonic OfFace recognition systemIntroduced[11]. A self-gate was created for Japanese returnees to take a picture of their face and check it against their passport.[12].2019ToNarita International AirportHowever, the NEC face recognition system was introduced.[13],2020 Tokyo OlympicsIntroduced from the Olympic venues to airports around the world[14], Target is expected to expand to foreigners[15].

Criminal investigation

2001 Year of 1 MonthSuper bowlPolice at Tampa, Florida are using FaceIt to look for crowds of criminals and terrorists[1]Found 19 people without arrest warrants[16].

Madeline McKern disappearance caseAs part of an investigation intoPortugalIn Paria da Luz Hotel Ocean Club Resort and surrounding areas2007May 5We are asking people who visit within 2 weeks to provide photos taken in the area. This is to check if the kidnapper is shown by face recognition.[17][18].

Police carThere is also a system that mounts a 360-degree camera connected to a face recognition system on the roof of the and searches for criminal suspects just by patrol[19].

Mech robotThere is also a system that is equipped with a camera connected to a face recognition system and searches for and traces criminal suspects just by patrol[20].

2018In the People's Republic of China,Jackie ChunWe have conducted an investigation using a face recognition system in places where people gather, such as in a concert venue, and arrested more than 100 criminal suspects and wanted criminals.[21],Chinese New YearPolice installed face recognition system at the stationSunglassesType ofSmart glassesWas also reportedly used to arrest seven criminal suspects and forge identification documents, and 7 were banned from traveling[22][23].

Ticket resale prevention

concertTo prevent the ticket from being resold at a high price, we sometimes use a face recognition system to check whether the ticket purchaser is present. In 2014Momoiro Clover ZIntroduced for the first time in the field of entertainment entrance management[24].NECFace authentication system "" is used to register a face photo when purchasing a ticket, and face authentication is performed at the entrance of the venue to issue a ticket[25].

NeoFace, including Hong Kong,The United States of America,チリUsed by police agencies around the world, the accuracy is rated as the highest level in the world.[26][27]. It is spreading to the concerts of other musicians,B'z,Fukuyama Masaharu,Mr. Children,BABYMETALUsed for some concerts[28][29]. Meanwhile, from April 2016, 4stormThe national tour was held from February to March of the same year.Shibuya SubaruAs with the solo live tour of, it is face recognition by multiple staff visual inspection[30][31].

Shoplifting prevention

Count about 100 stores nationwideMaruzen Junkudo BookstoreFace recognition system used in all storesShoplifting prevention systemWas introduced.ShopliftingFace data of customers who may haveDatabaseRegister at and detect when you visit the store. Although he announced that the security camera was operating, he did not announce that there was a face recognition function. On the other hand, an attorney who is familiar with privacy issues said, "A face recognition system that has the function of tracking a specific individual has a higher degree of portrait rights and privacy infringement, and it is necessary to distinguish between the two. It is important to ensure transparency by clearly indicating what you are doing and not having to use the store if you feel uncomfortable.” Maruzen Junkudo announced that it would consider changing the display after that.[32].


In the 2000 presidential election, the Mexican government prepared facial recognition software to prevent fake voting. This is because some individuals were registered with multiple names and were trying to cast multiple votes. Significantly reduced double votes by comparing new face images with those already in the voter database[33][1]. Similar technology is used in the United States to prevent the acquisition of fake ID cards and licenses[34][35].

In addition, a vending machine that uses a built-in camera to determine whether the person who shot the image is an adult or a minor[36]Has also been put to practical use.

Medical errorSome hospitals have introduced a face authentication system to prevent this and identify the person with the face photograph embedded in the medical examination ticket. If you forget your voucher, you can use the face recognition system to receive one.

Besides the use as biometric authentication,Digital camera,SmartphoneThere are many things that can recognize the face of the subject and focus on it or adjust the exposure. There is also a function that recognizes facial expressions and automatically shoots when a smile is confirmed.

IbarakiTsukuba CityThen,Society5.0In the proof experiment of using the Internet that introduced the face recognition systemElectronic votingI went.

Face recognition technology installed in general smartphones until the 2010sIdentical twinsAre not identified, or the false recognition rate is significantly increased.[37].

Comparative study

Face recognition is not the most reliable and efficient method compared to other biometric technologies, but it has the important advantage of not requiring the cooperation of the target. Systems such as airports can distinguish criminals from crowds. Fingerprint, iris, and voice authentication cannot be used in this way. However, there are some doubts about its effectiveness in terms of security.

Another advantage is that it is possible to use inexpensive cameras that are commonly used (existing surveillance cameras, webcams, etc. can be used as they are), and face images remain as evidence, so unauthorized users, etc. It is easy for a person to identify a person by looking at the face image recorded when it is detected. Is difficult).


Recognition rate

The world's best face recognition system reaches a recognition rate of 99.5% for still images[38].

Face recognition is not perfect, and if the conditions are not met, the recognition rate will be low. Ralph Gross of the Carnegie Mellon University Robotics Laboratory said about the angle at which the face was photographed, "Face recognition is based on capturing the face from the front, and somehow it can recognize up to 20 degrees. Will occur"[5].

In addition, the recognition rate will be extremely low if the screen is dark, if you are wearing sunglasses, if your hair is growing, if part of your face is hidden by something, or if the resolution is low.[1].

In addition, it is often impossible to recognize when the facial expression changes. Canada says passport portraits must be expressionless[39].


Technical critics are critical of the system not knowing any criminals, even though there were criminals in Newam actually in the database[40][41].. This is inconsistent with the information that the number of crimes decreased by 34% due to the introduction of facial recognition system, but a similar system was introduced in Birmingham from the viewpoint of crime prevention.[42].

FloridaTampaAtPolicemenExperiments with also yielded similar disappointing results[43].

BostonGeneral Edward Lawrence Logan International AirportHowever, a large-scale experiment on terrorist identification using a face recognition system was conducted, but it ended in failure.[44].

Privacy issues

Even considering the potential benefits of this technology,PoliciesThere is still concern about the infringement of. The governmentbig BrotherSome people are worried that they will constantly monitor each and every person and grasp their actions. That power can cause such a runaway isHistoryHas proved[45].

Criticism in Japan

  • mainlyCapital AreaWhereaboutsShopping centerAnd large scaleApartmentIt has been revealed that 29 of the cameras equipped with the face recognition system installed in the are being photographed without being informed to users or passers-by. It is said that there is a purpose to analyze the gender and generation of viewers and use it for customer analysis, and the camera installer side claims that there is no problem because the individual is not specified, but among experts There are many criticisms such as, "There are many people who are not satisfied with commercial purposes, and some kind of rule development is necessary"[46].
  • National Institute of Information and Communications TechnologyIt is,JR West JapanWith the cooperation ofOsaka StationOn the premisesOsaka Station CityA large number of face recognition cameras were installed in the field, and a demonstration experiment was conducted to track the passersby on the premises.20144It was scheduled to be implemented from. However, the plan was the same yearMay 1Was reported to the public and many protests were received from citizens, 20143Currently, there is no prospect of implementation. Although the organization and JR West are said to be "for disaster prevention purposes", it is considered that there is concern that they will be used for commercial purposes etc.[47].


Pioneers of automated face recognition systems are Woody Bledsoe, Helen Chan Wolf and Charles Bisson.

1964から1965Bledsoe worked on computer-based face recognition with Helen Chan and Charles Bisson. He was proud of this work, but because it was funded by an anonymous intelligence agency, it was rarely publicized and only a small portion of the work was published. Given a large image database (like a facebook) and a single photo, the task was to extract a few records that matched the photo from the database. The success of the method is represented by the ratio of the number of extracted records to the number of records in the entire database. Bledsoe describes the difficulty as follows:

"This recognition problem is compounded by significant changes in head movements up and down, left and right, illuminance and angle, facial expression and aging. Other attempts at machine face recognition have taken these changes into account with little or no consideration. Some researchers have used correlations (or pattern matching) on ​​optically unprocessed image data, but they have always failed when the changes are dramatic. Correlation is extremely low for photographs of the same person in different orientations (Woody Bledsoe, 1966).

This project was classified as "Man Machine". This is because humans extracted features from photographs and used them for computer recognition. GRAFACON and RAND TABLETpen tabletUsing, the operator extracted feature points such as the center of the pupil, the inner and outer edges of the eye, and the hairline at the center of the forehead. From these coordinates, 20 types of distances, such as the width of the mouth and eyes and the distance between pupils, are calculated. The operator was able to process about 40 photos per hour. At the time of building the database, these data and the person name of the face photograph were stored in the computer in association with each other. In the recognition phase, these distance data groups are compared with the distance data group extracted from the target photograph. And the closest record comes out as an answer.

This explanation is so simplistic that most of the time it will fail. This is because the conditions such as the up/down/left/right orientation of the face and the distance from the camera do not completely match in the two separately taken photographs. Therefore, the distance data is normalized to the value when the face is viewed from the front. To perform this normalization, the program tries to determine the face orientation and tilt values. Then, using these values ​​(angles), the computer cancels the change in the direction and the inclination and calculates the data when viewed from the front. To calculate the angle, the computer needs to know the three-dimensional position of the head. Bledsoe measured the heads of seven people and determined a standard head position, since he could not use the actual head.

Bledsoe1966This work was done by Peter Hart et al.Stanford Research InstituteContinued in. In experiments carried out on a database of over 2000 photographs, the computer has always been superior in comparing face recognition ability between the computer and humans.

Around 1997Ruhr University BochumAnd the AmericanUniversity of Southern California TheMassachusetts Institute of Technology,University of Maryland College ParkWe have developed a face recognition system that is even better than the above system. In BochumsoftwareWas commercialized as ZN-Face,Deutsche Bank,空港Was adopted by. The face recognition system was able to recognize even incomplete facial images, and it was possible to recognize even a mustache, a beard, different hairstyles, or wearing glasses or sunglasses.[48].

Image retrieval generally requires some tagging of images. January 2007,Polar Rose Is a service that uses face recognition technology to create a three-dimensional image of a person's head from a photo in about 3 seconds, asks the user for the person's name, matches the name with the face, and uses it as an image search tag. Started[49].

2006 years,Face Recognition Grand Challenge (FRGC) The latest face recognition system was evaluated at. As a result, it was shown that the recognition was 2002 times more accurate than the 10 system and 1995 times more accurate than the 100 system. There is also an algorithm that surpasses human face recognition ability, and it was possible to identify identical twins as different people.[5].

As a technique to enhance the resolution of low-resolution face images face hallucination There is. Further, in recent years, the number of pixels of the camera itself has increased, and the problem of resolution is being solved.


  1. ^ a b c d e f g h Bonsor, K .. “How Facial Recognition Systems Work". 2008th of February 6Browse.
  2. ^ Smith, Kelly. “Face Recognition (PDF)". 2008th of February 6Browse.
  3. ^ R. Brunelli and T. Poggio, "Face Recognition: Features versus Templates", IEEE Trans. on PAMI, 1993, (15)10:1042-1052
  4. ^ R. Brunelli, Template Matching Techniques in Computer Vision: Theory and Practice, Wiley, ISBN 978-0-470-51706-2 2009 ([1] TM book)
  5. ^ a b c d e Williams, Mark. “Better Face-Recognition Software". 2008th of February 6Browse.
  6. ^ Kimmel, Ron. “Three-dimensional face recognition". 2005th of February 1Browse.
  7. ^ "China's surveillance network instantly identifies people Surveillance cameras with AI nationwide". with the BBC (September 2017, 12). 2019th of February 6Browse.
  8. ^ http://www.bundespolizei.de/cln_030/nn_484498/EN/AutomatedBorderControls/AutomatedBorderControls__anmod.html__nnn=true[Broken link]
  9. ^ SmartGate
  10. ^ "NEC delivers the world's first immigration gate management system for face recognition while riding in Hong Kong”. Business + IT (June 2007, 7). 2019th of February 9Browse.
  11. ^ "Pana and NEC clash, airport "face recognition" gate battle". 東洋 経 済 オ ン ラ イ ン (September 2018, 8). 2019th of February 6Browse.
  12. ^ "Will passport stamps disappear? Introducing "face recognition" for immigration procedures at Haneda Airport”. Vehicle News (January 2017, 10). 2018th of February 4Browse.
  13. ^ "Narita Airport, Face Pass Boarding First in Japan since spring 20 Adopts NEC's face recognition system”. AviationWire (June 2019, 6). 2018th of February 4Browse.
  14. ^ "Introduced facial recognition at the Tokyo Olympics. Pre-registered over 30 people involved in the competition.”. Nihon Keizai Shimbun (May 2018, 8). 2019th of February 6Browse.
  15. ^ "Countermeasures against terrorism by personnel with "face recognition" for foreigners leaving Japan”. Yomiuri Shimbun (March 2018, 4). 2018th of February 4Browse.
  16. ^ McNealy, Scott. “Privacy is (Virtually) Dead". 2006th of February 12Browse.
  17. ^ “Help find Madeleine McCann”. Child Exploitation and Online Protection Center. (May 2007, 5). オ リ ジ ナ ルArchived as of October 2007, 5.. https://web.archive.org/web/20070523075031/http://madeleine.ceopupload.com/ 2007th of February 5Browse. 
  18. ^ Brown, David (February 2007, 5). “We will travel anywhere to find Madeleine, say parents”. London: Times Online. http://www.timesonline.co.uk/tol/news/world/europe/article1826735.ece 2008th of February 6Browse. 
  19. ^ "A high-tech police car that finds the criminal just by scanning the surroundings with a face recognition system appeared”. GIGAZINE (May 2016, 4). 2018th of February 2Browse.
  20. ^ "Robots run the city of the future? Reality in Dubai and China". with the BBC (September 2018, 6). 2019th of February 9Browse.
  21. ^ ""Utagami" is too good! "Fugitive enemy" Jackie Chun finally arrested 100 people at concert-China". Record China (September 2019, 1). 2019th of February 9Browse.
  22. ^ "Chinese police are strengthening the surveillance system by introducing a sunglasses type device equipped with a face recognition function". GIGAZINE (September 2018, 2). 2019th of February 6Browse.
  23. ^ "Scanning travelers with face recognition glasses — — China, already 7 people arrested at station". Business insider (September 2018, 2). 2019th of February 6Browse.
  24. ^ “Face pass at a concert!! Enter face recognition quickly!”. NEC. http://jpn.nec.com/ad/onlinetv/concert.html 2016th of February 9Browse. 
  25. ^ "NEC, face authentication to prevent "Momokuro" ticket resale". NEC. http://www.nec.co.jp/press/ja/1001/2001.html 2019th of February 6Browse. 
  26. ^ "Innovators 100 Vol.04 Face recognition, the world's leading performer Hitoshi Imaoka". NEC. https://jpn.nec.com/info-square/innovators/h_imaoka/02.html 2019th of February 9Browse. 
  27. ^ “NEC Face Recognition Technology Receives First Place in All Participant Categories in Vendor Evaluation Project of National Institute of Standards and Technology (NIST)”. Nihon Keizai Shimbun. http://www.nikkei.com/article/DGXMZO80664930Y4A201C1H56A00/ 2015th of February 1Browse. 
  28. ^ "It is also used in Momokuro's live! What are the advantages of face recognition system?”. Nikkei BP (April 2016, 4). 2016th of February 5Browse.
  29. ^ "Face recognition system/concert/event ticket resale prevention". 2016th of February 5Browse.
  30. ^ "Arashi Live's "face recognition" was not NEC's "accuracy over 99%" system!”. Excite Japan (April 2016, 4). 2016th of February 5Browse.
  31. ^ "The latest system is sent off. "Visual" face recognition, which has been dissatisfied by Arashi fans”. Weekly Women's Prime (Shufu to Seikatsusha) (May 2016, 5). 2016th of February 5Browse.
  32. ^ "Using face data without notifying the customer... Understanding the customer base and preventing shoplifting". Yahoo! NewsYomiuri Shimbun (Yahoo Japan). (December 2015, 12). オ リ ジ ナ ルArchived as of October 2015, 12.. http://archive.is/NhC6d 
  33. ^ "Mexican Government Adopts FaceIt Face Recognition Technology to Eliminate Duplicate Voter Registrations in Upcoming Presidential Election”. Business Wire. As of July 2012, 7オ リ ジ ナ ルMore archives.2008th of February 6Browse.
  34. ^ House, David. “Facial recognition at DMV”. Oregon Department of Transportation. As of October 2007, 10オ リ ジ ナ ルMore archives.2007th of February 9Browse. “Oregon DMV is going to start using“ facial recognition ”software, a new tool in the prevention of fraud, required by a new state law. The law is designed to prevent someone from obtaining a driver license or ID card under a false name. ”
  35. ^ Schultz, Zac. “Facial Recognition Technology Helps DMV Prevent Identity Theft”WMTV News, Gray Television. オ リ ジ ナ ルArchived as of October 2012, 9.. http://archive.is/eTIO 2007th of February 9Browse"Madison: ... The Department of Motor Vehicles is using ... facial recognition technology [to prevent ID theft]" 
  36. ^ "Super vendor TOBACCO". Tobacco vending machine siteFujitaka. As of March 2006, 3オ リ ジ ナ ルMore archives.2016th of February 10Browse.
  37. ^ Tane, Kiyoshi. “Will Face IDs in the future be perfect for twins? Apple applies for vein pattern biometric patent-Engadget English”(Japanese). Engadget JP. 2019th of February 11Browse.
  38. ^ "NEC wins first place in face recognition accuracy evaluation by US national organization”(Japanese). NEC. 2019th of February 11Browse.
  39. ^ Passport photos
  40. ^ Meek, James (June 2002, 6). “Robo cop”. UK Guardian newspaper. http://www.guardian.co.uk/Archive/Article/0,4273,4432506,00.htm 
  41. ^ Krause, Mike (January 2002, 1). “Is face recognition just high-tech snake oil?”. Enter Stage Right. http://www.enterstageright.com/archive/articles/0102/0102facerecog.htm 
  42. ^ “Birmingham City Center CCTV Installs Visionics' FaceIt”.Business Wire. (June 2008, 6). オ リ ジ ナ ルArchived as of October 2008, 12.. https://web.archive.org/web/20081220212810/http://www.allbusiness.com/government/government-bodies-offices-regional/6111139-1.html 
  43. ^ Krause, Mike (January 2008, 6). “Is face recognition just high-tech snake oil?”. Enter Stage Right. http://www.enterstageright.com/archive/articles/0102/0102facerecog.htm 
  44. ^ Willing, Richard (September 2003, 9). “Airport anti-terror systems flub tests Face-recognition technology fails to flag'suspects'”. USA Today. http://pqasb.pqarchiver.com/USAToday/access/391952771.html?dids=391952771:391952771&FMT=ABS&FMTS=ABS:FT&date=Sep+2%2C+2003&author=Richard+Willing&pub=USA+TODAY&edition=&startpage=A.03&desc=Airport+anti-terror+systems+flub+tests+%3B+Face-recognition+technology+fails+to+flag+%27suspects%27 2007th of February 9Browse. 
  45. ^ "Civil Liberties & Facial Recognition Software”. About.com, The New York Times Company. Pp. pp. 2.As of August 2006, 3オ リ ジ ナ ルMore archives.2007th of February 9Browse. “A few examples which have already arisen from surveillance video are: using license plates to blackmail gay married people, stalking women, tracking estranged spouses ...”
  46. ^ “Customer shooting without “face recognition” without permission... Commercial facilities in the metropolitan area, etc.”. YOMIURI ONLINE (Yomiuri Shimbun). (November 2012, 11). オ リ ジ ナ ルArchived as of October 2012, 11.. http://archive.is/OUhy 
  47. ^ "Face authentication: not stand the prospect demonstration experiment citizen protest in Osaka Station Building". Mainichi Newspapers (Mainichi Newspaper). (May 2014, 3). オ リ ジ ナ ルArchived as of October 2014, 3.. https://web.archive.org/web/20140309055300/http://mainichi.jp/select/news/20140306k0000m040167000c.html 
  48. ^ “"Mugspot" Can Find A Face In The Crowd --Face-Recognition Software Prepares To Go To Work In The Streets”.ScienceDaily. (November 1997, 11). http://www.sciencedaily.com/releases/1997/11/971112070100.htm 2007th of February 11Browse. 
  49. ^ Polar Rose enters the face recognition field from Europe December 2006, 12, TechCrunch

Related item

外部 リンク


Back to Top