Face recognition systems

an introduction

The mind can identify the faces we see and we can know if we have met this person before or not, but in the field of information systems, this is different despite the storage capacity, scientists began work in the mid-sixties on the use of systems Information to know the human faces. Face recognition programs, known as face recognition programs, are developed to capture a picture of a face from a large gathering and compare the image to a large database of people. In order for the program to recognize the face it should initially know the difference between the main face and isolate it from the other faces as in the picture below and then the program measures many of the features that distinguish the face.

The use of an identity system was first introduced in 2001 in the United States, and was primarily intended to secure criminals, but failed to draw faces because the criminals wore masks on their faces. The face recognition system was carried out through the use of surveillance cameras High-tech images take three-dimensional images of faces in crowded places and compare the image to a large database of images of people, and because of the growing concern about security issues around the world increased interest in general about the accuracy of computer systems for recognition of faces, The range of security systems and applications in this field has developed significantly, and the algorithms used vary between simplicity and complexity.

Computer science researchers have worked on the development of many commercial products that have improved the performance of automated face recognition algorithms in a number of areas requiring face recognition functions. The need to develop an accurate facial recognition system, Face Recognition System is a predefined set of images for the face recognition information system.

These images consist of selected matrices from the original image matrices extracted from a known set of faces to different people. They represent important and fundamental values within the original images, thus reducing the size of the images to a ray representing the image abstract. Thus, the task of the recognition system is to find a ray The most similar and closest features within the training package to the ray of features from the image The image whose identity is required to be identified through the recognition system The identity of a person can be distinguished by passing the person's image to the face recognition information system

Face recognition technology is used to monitor people who violate traffic laws. Pedestrians who violate traffic laws are photographed directly on a large screen at the nearest intersection and avoid appearing on screen. They should pay a fine. To monitor legally required and every citizen aged 16 and over must possess an identity card bearing his image and address. Experts consider that China is ahead of the West in this area, especially because its laws on privacy are less stringent and because its citizens are used to being photographed and fingerprinted To give various personal data to the authorities of the types, and this technology has become used in many fields including as restaurants of  fast food network, "Kay FC", which uses a system called "smiled pay an" access to the latest uses

China has begun to apply facial recognition technology at the main train station to identify the passengers and a university in Beijing has installed a device that works at the entrances to the dormitory to make sure that access to the institution is limited to its students. "Banks began to process automatic payment machines to replace electronic payment cards, and travel and leisure professionals benefit from these services by using this technology instead of boarding passes.

In China, China also uses pay-as-you-go payment systems instead of cash payments in the transport and public services sectors to relieve pressure on these sectors in metro and high-speed rail stations, The traveler does not need to show his or her ticket or ticket, but just have to stand for a moment in front of the camera to clear his face and match him with his picture and personal data.

The technology works in a similar way to the fingerprint reader on the smartphone, where travelers' faces are linked to their bank accounts, meaning no delay because travelers wait for the rest of the fare. For example, the traveler only needs to stand in front of the camera to take his picture and wait for verification.

In recent years, China has expanded the use of facelift technology in electronic sales platforms to allow users to pay by facsimile instead of cash. This has reduced time and increased safety in day-to-day transactions.

China has announced its intention to build an artificially intelligent police station and not human beings. The center will be opened in Wuhan, the capital of central China's Hubei Province. The new center is designed to service issues related to driving licenses, cars and vehicles, More than a police station and provides the possibility to conduct simulated driving tests and registration services and the advanced scanning technology developed by Tencent.

How the facial recognition system works

Some information security systems rely on sophisticated protection systems that can identify undesirable people, such as thieves and security personnel, so that they can be controlled. These systems rely on their image databases and the program compares the images captured by the surveillance cameras to the database.

The program creates a set of network nodes in the image of the person to be identified and depends on the contract in defining facial features and then begin the process of creating a picture that corresponds to the owner of this contract network each face has many distinct features, which is the various aliases on the face. The program relies on these parameters as nodes. Each human face has approximately 80 nodes. The most common facial features measured by the program are:

(1) the distance between the eyes

(2) width of the nose

(3) the depth of the eye

(4) the shape of cheekbones

(5) the length of the jaw line

These parameters are measured by the program and translated into digital codes called face tags and used to represent the face in the database.

The phases of face recognition are generally summarized in the following steps:

1. Acquire the image (capture) Acquire

2. The stage of extracting the facial image from the overall image Detect

3. Phase Align and Image Standardization (ie adjust the angle of the face with the camera angle) Align

4. The stage of extracting the essential features of the image

5. Matching stage between the desired image and the picture store Match

6. The stage of issuing a report with the closest image to the image or the absence of a similar report

These stages differ in incremental increments according to the system being programmed and the target.

Modern face recognition systems

The new face recognition systems are based on 3D style. Special cameras capture 3D images of the suspicious person using the main features and features of each face, which have no significant change over time such as eye strain, nose shape, distance between the eyes and other features. These features are a source of information for face recognition systems where changes in lighting or surrounding environmental conditions do not affect these measurements. For example, these systems can operate in any lighting conditions even if the place is dim and even if the person is not facing the camera .

3D camera used in face recognition system

The use of depth and focus of the face that does not affect the change in lighting is known as three-dimensional on the face and software systems that rely on three-dimensional technology goes through a series of steps to be able to finally recognize the face.

(1) Detection Detection

The step is to capture a digital image with a two-dimensional digital camera or even a video camera.

(2) alignment Alignment

After capturing the image, the frame determines the position, size, and orientation of the head. A three-dimensional system can do this even if the image taken is a side image, which makes a 90-degree angle with the camera lens, while two-dimensional systems can only perform this step if the person looks directly at or towards the camera so that no The angle between the person's face and the camera lens increases from 35 degrees.

(3) Measurement Measurement

The system software calculates curves and aliases on the face accurately up to parts of the millimeter. This information turns into a face model.

(4) Representation Representation

In this step, the system translates the form into a code. The code for each model is unique and consists of a set of numbers.

(5) Comparison Matching

If the image is three-dimensional and matches three-dimensional images and is stored in the system database, the comparison of images is done directly. But the challenge for these systems is that most of the images stored in the databases are ordinary (two-dimensional) images. How can one compare a living image of someone who moves his head in front of the camera and takes a 3D picture with millions of two-dimensional images? This is why new technology has developed using three different points of recognition. These points represent the outside of the eye, the inside of the eye, and the nose. These systems make precise measurements of the dimensions between these points for three-dimensional images and begin to convert them into two-dimensional images by applying complex mathematical algorithms. After the conversion process, the system starts the comparison process.

 (6) Identification or Verification

In the recognition step, compare the image and match it with the database images that the system sorted in the previous step. However, if the goal is to verify the result of the previous step, the system compares the image with all images in the database and displays the results in percentages.

Dim lighting may cause system failure to recognize the face and be

The percentage of error is large and despite the success of these systems and development, but it does not reach the degree of perfection yet, because of the presence of some factors that may hinder the process of recognition of the face and these obstacles include:

(1) glare resulting from wearing sunglasses

(2) Long hair obscures the central part of the face

(3) Low light resulting in unclear images

(4) Double precision and clarity of images taken remotely.