3 Piece Can Production Line,Metal Tin Can Lid Making Machine,Capping Machine Twist Off Lids , Glass Jar Bottle Cap Making Line Zhoushan Putuo BODA Machinery Co., Ltd. , https://www.boda-machinery.com
The development of science and technology is accelerating our lives. In the past, when we were shopping, the cashier would ask “cash or credit card.†Now, this sentence has become “WeChat or Alipay?†Before we took cash in the street and later became a belt card, now we just need to With cell phone.
However, did you think about it? One day in the future, we don’t even need to bring our mobile phones to the streets. Just “bring your face†will do. Because, we are moving towards the era of "face brushing." When you arrive, put all your information and assets on your face, and go out and brush your face. Today, we will take a closer look at face recognition technology:
Face recognition overview
Face recognition is a kind of biometric recognition technology based on human facial feature information. A camera or camera captures an image or video stream containing faces, and automatically detects and tracks faces in the images, and then performs a series of related techniques on the detected faces, often called face recognition and face recognition.
Face recognition is a popular field of computer technology research. It belongs to biometric recognition technology. It distinguishes organisms from biological features of organisms (generally people).
The biometrics studied by the biometrics include face, fingerprint, palm print, iris, retina, voice (voice), figure, personal habits (such as the strength and frequency of typing on the keyboard, signature), and so on.
The corresponding recognition technologies include face recognition, fingerprint recognition, palmprint recognition, iris recognition, retina recognition, and speech recognition (using voice recognition for identity recognition and voice content recognition, only the former belongs to biometric recognition technology). Shape recognition, keyboard stroke recognition, signature recognition, etc.
Three key technologies
Feature-based face detection technology
Face detection is performed by using colors, outlines, textures, structures, or histogram features.
2, based on template matching face detection technology
The face template is extracted from the database, and then a certain template matching strategy is adopted to match the captured face image with the extracted picture from the template library, and the face size and position information are determined by the relevance level and the matched template size.
3, statistics-based face detection technology
By collecting a large number of face positive and negative sample banks for “face†and “non-face†images, the system is trained by statistical methods to achieve detection and classification of face and non-face modes.
Four characteristics
1, geometric features
From the distance and ratio between facial points as a feature, the recognition speed is fast, the memory requirements are relatively small, and the sensitivity to light is reduced.
2, based on model features
The facial image features are extracted based on the different probabilities of different feature states.
3, based on statistical characteristics
The face image is regarded as a random vector, and statistical methods are used to distinguish different face feature patterns. Typical face features, independent component analysis, and singular value decomposition are compared.
4, based on the characteristics of neural networks
A large number of neural units are used to associate and store facial image features, and accurate recognition of human face images is achieved based on the probability of different neural unit states.
Top ten difficulties
1, lighting problems
Illumination change is the most critical factor affecting the performance of face recognition. The degree of resolution of this issue is related to the success or failure of the process of face recognition. Due to the 3D structure of the human face, the shadow cast by the light will strengthen or weaken the original facial features. Especially at night, the shadow of the face caused by insufficient light causes a sharp drop in the recognition rate, making it difficult for the system to meet practical requirements.
At the same time, the theory and experiment also proved that the difference of the same individual due to different illumination is greater than the difference between different individuals under the same illumination. Illumination is an old problem in machine vision and it is particularly evident in face recognition. Solutions to solve the lighting problem include three-dimensional image face recognition and thermal imaging face recognition. However, these two technologies are still far from mature, and the recognition effect is not satisfactory.
2, posture problems
Face recognition is mainly based on human facial features. How to recognize face changes caused by posture has become one of the difficulties of this technology. The pose problem involves facial changes caused by the rotation of the head around the three axes in a three-dimensional vertical coordinate system, where deep rotations in two directions perpendicular to the image plane cause partial deletion of facial information. Making pose problems become a technical problem for face recognition.
There are relatively few researches on attitude. Most current face recognition algorithms mainly include frontal and frontal face images. When the pitch or left and right sides are relatively severe, the recognition rate of the face recognition algorithm will also be A sharp decline.
3, facial problems
Facial facial expression changes such as cries, laughs, and anger also show the accuracy of facial recognition. The existing technology has handled these aspects quite well, either on the open mouth or on some exaggerated facial expressions. The computer can correct it through three-dimensional modeling and pose correction methods.
4, occlusion problem
The occlusion problem is a very serious issue for face image acquisition in non-match situations. Especially in the monitoring environment, often the monitored objects are wearing glasses, hats and other ornaments, making the collected face images may be incomplete, thus affecting the subsequent feature extraction and recognition, and even result in face detection algorithms. Failure.
5, age change
As one's age changes, a person changes from a teenager to a young person and becomes an old person. His appearance may undergo a relatively large change, leading to a decrease in the recognition rate. For different age groups, the recognition rate of the face recognition algorithm is also different.
The most direct example of this problem is the identification of ID photos. The validity period of ID cards in China is generally 20 years. During these 20 years, each person's appearance will inevitably undergo considerable changes. All of them have a great deal of recognition. problem.
6. Face similarity
There is not much difference between different individuals. The structure of all faces is similar. Even the structure of the human face is very similar. Such a feature is advantageous for the positioning of human faces, but it is disadvantageous for distinguishing human individuals using human faces.
7, dynamic identification
In the case of non-combined face recognition, facial image blurring caused by motion or incorrect focus of the camera can seriously affect the success rate of facial recognition. This difficulty has become apparent in the use of security and surveillance identification in subways, highway bayonet, station bayonet, supermarket counters, border inspections, etc.
8, face security
The main method of deception to forge face images is to establish a three-dimensional model, or to embed some expressions. With the introduction of intelligent anti-counterfeiting technology, 3D facial recognition technology, camera and other intelligent computing vision technologies, the success rate of forging facial images for recognition will be greatly reduced.
9, image quality problems
Face images may come from a variety of sources. Due to the different acquisition devices, the quality of face images obtained is also not the same, especially for those face images with low resolution, high noise, and poor quality (such as those shot by mobile phone cameras. Face images, remote monitoring and shooting pictures, etc.) How to perform effective face recognition is a matter of concern.
Similarly, the influence of high-resolution images on face recognition algorithms needs further research. Nowadays, when we use face recognition, we generally use the same size and sharpness of face images. Therefore, image quality problems can be basically solved. However, in the face of more complex problems in the real world, we need to continue to optimize the process.
10, lack of sample
The face recognition algorithm based on statistical learning is currently the mainstream algorithm in the face recognition field, but statistical learning methods require a lot of training. Because the distribution of face image in the high-dimensional space is an irregular manifold distribution, the available sample is only a small part of the face image space. How to solve the statistical learning problem under small sample needs to be further studied. Research.
In addition, the face image libraries currently involved in training are basically images of foreigners. There are very few face images libraries for Chinese and Asian people, which makes training facial recognition models more difficult.