Shulou(Shulou.com)06/02 Report--

The iris is regarded as one of the "most reliable biological features" except DNA. It is non-contact, high-precision, unique and stable for life. It plays an important role in detecting criminal cases and cracking down on illegal exit. In today's society where the flow of people and logistics are highly developed, it is difficult to accurately determine the identity of personnel only on the basis of ID cards, passports and other documents, and the accurate authentication of citizenship by means of science and technology is an important direction of public security work in the future. Some foreign developed and developing countries have begun to collect iris biometric information of entry and exit passengers and carry out relevant testing and research. In this open class, the speaker will deeply share the development status of modern iris recognition technology and its application in security and other fields.

Open course playback link: http://www.mooc.ai/open/course/681

The sharing content of Lei Feng's AI study Society is as follows:

Hello, everyone. Thank you very much for providing such a platform to share with you the topic of modern iris recognition technology and its applications. I am also very grateful to the engineers in the field of AI who have taken the time to participate in this sharing. I hope to bring some valuable information to you in this introduction. I also hope that you will pay attention to the talent demand of Juhong Optoelectronics in the field of AI and be able to actively deliver resumes.

Let's move on to the topic of this sharing, modern iris recognition technology and its applications. I am Gong Yazhuo from Shanghai Juhong Optoelectronic Technology Co., Ltd.

First of all, let's take a look at the framework of this sharing, which is divided into the following four parts, as shown in the figure:

Let's move on to the first part, concepts and technical principles. The concept of iris recognition is actually a very avant-garde concept. It has been around for less than 30 years, and it was not until the 1990s that the technology was formally developed. We can see the following picture, which shows very clearly which region of the iris is. The middle black is the pupil, next to the white is the sclera, the circular area between the pupil and the sclera is the iris, is a circular ring, which is distributed with a large number of fine textures, each person's texture is different. The biometric recognition based on these textures is iris recognition.

Let's take a look at the basic principles of iris recognition: if we express its algorithm process in one sentence, it is to turn a circular iris principle image into a series of texture coding, as shown in the following figure (please watch video playback for details):

Let's take a look at the key steps in iris recognition. The first step is eye detection. Finding eyes quickly in a very clear and large resolution image is the first algorithmic challenge of iris recognition. The following picture shows three pictures, some with only one iris and some with two iris. How to find an eye quickly requires some very complex algorithms, we can use traditional digital image processing methods, such as circle detection, pupil detection, spot detection to help determine which part of the eye is the eye. New methods of machine learning can also be used, which have more advantages in speed and accuracy. (please watch the video playback for details.)

Next, we start to segment. When we get an original iris image of the eye, we need to separate it, that is, to identify the location of the ring structure. The steps are shown below. Please watch the video playback for details:

After the segmentation is completed and a circular texture area is obtained, it can be normalized. The right side of the following image shows a typical normalized image: a pair of circular iris is unfolded into a rectangular iris. After getting the rectangle, we divide it into sub-blocks, extract the features of each small block, and then connect all the iris features formed by the small blocks together to get a complete iris code. please watch the video playback for a detailed explanation:

The above is the general complete process of iris recognition. Next, let's take a look at the physiological structure of iris features, as shown in the figure. For details, please watch video playback:

Let's take a look at the advantages of iris recognition. Here are five advantages, such as the following figure. Please watch the video playback for details:

Now let's move on to the second part, the development and current situation of iris recognition. First of all, let's take a look at the development process, as shown in the figure. Please watch the video playback for a detailed explanation:

Let's take a look at the overseas development of iris recognition. Take Iridian as an example, the company was founded by the founder of iris recognition. For more information, please watch the video playback:

Just now we were looking at the situation overseas, and now let's take a look at the current situation of the whole industry. from the perspective of the whole industry, foreign manufacturers are gradually withdrawing from the market, while domestic manufacturers are emerging as a new force. Please watch the video playback for detailed explanation:

Next, let's take a look at the current situation of the whole industry. Domestic iris recognition companies with independent intellectual property rights were academic institutions at first. Shanghai Juhong's core R & D team comes from the Image Institute of Shanghai Jiaotong University and Hong Kong Polytechnic University. It combines the top iris recognition technology talents on both sides of the strait. Please watch the video replay for detailed explanation:

Let's take a look at the third part, the application trend and technical difficulties of iris recognition. The first application trend of iris recognition is long distance. When the iris was first born, the imaging distance was very close, only about ten to 20 centimeters, and the maximum distance was only 30 to 40 centimeters, and users were required to cooperate very well with the acquisition process of iris recognition. therefore, the user experience is not very good, and there is a strong sense of aggression in the whole recognition process, which also affects the collection of iris recognition and the promotion of iris recognition technology. For more information, please watch the video playback:

The second is mobile, mobile, miniaturization is actually a direction of biometric development, two product forms can be mentioned here. One is Samsung's previously mentioned and manufactured Galaxy series, which integrates iris recognition into the phone's front camera to unlock the user's iris or pay for iris, which is the application form of iris recognition made by Samsung on mobile phones. The other is dedicated mobile devices such as mobile phones and tablets. For more information, please watch video playback:

The third trend is multimodal. There is no challenge for iris as a single biometric feature that can provide the highest recognition accuracy, but if it is integrated with other features, it may bring a better user experience. For example, iris and face, please watch the video playback for more details:

The next trend is scale. We don't seem to see many iris applications in our daily life, but in fact, large-scale iris applications have become a trend of globalization. Here are three regional cases. For more information, please watch the video playback:

You can take a look at the cases of overseas iris database construction. The most representative case is India, and the other cases are Indonesia and Singapore. Please watch the video playback for details:

Next, let's take a look at the application of iris in foreign law enforcement departments. For more information, please watch the video playback:

Why do so many overseas countries adopt iris recognition? what are the characteristics of its audience? We have made a simple summary, as shown in the figure. For more information, please watch the video playback:

No matter at home or abroad, iris recognition technology has such a large market potential demand, so what are its difficulties? The first difficulty is the recognition of black iris. Please watch the video playback for more details:

The next technical difficulty is low-quality sample processing. If a very high-quality sample is collected, the eyelid opening is very large, all iris textures are exposed, and the texture is very rich and delicate, and there are no light spots and occlusion on the surface. Such high-quality samples can help us to do high-precision iris recognition. However, in the real use of the scene, there is no way to require users to get such a high-quality sample, there will be a variety of low-quality images. Please watch the video playback for details:

The third difficulty is long-distance identification. Please watch the video playback for details:

Finally, let's take a look at the introduction of Juhong Optoelectronics. As a professional iris recognition algorithm company, Juhong Optoelectronics has done something, what is our position in the whole industry, and what height we expect to achieve in the future.

First of all, let's take a look at the company's profile. Juhong Optoelectronics is the earliest technical team to carry out iris recognition research in China, as shown in the following figure. Please watch the video playback for detailed explanation:

Next, let's take a look at the overview of Juhong optoelectronic technology, with iris recognition as the core, combined with face recognition, and multi-dimensional feature fusion of eye movement, eye perimeter, and micro-expression line of sight tracking. to provide users with highly automated, non-collaborative, non-inductive identity authentication, living detection products and solutions. It can be understood that all interactive recognition technologies related to iris recognition and movement features around the eyes are within the scope of Juhong's research and development, and we have also made a series of achievements. Please watch the video playback for details:

Next, let's take a look at technological innovation. The technological innovation part is mainly aimed at the black iris of Easterners. The following picture shows some of our research results. Please watch the video playback for details:

The following are the authoritative qualifications obtained by Juhong Optoelectronics to meet the large-scale database construction requirements of the Ministry of Public Security, as shown in the following figure:

Next, let's talk about Juhong Optoelectronics's desire for biometric talents, especially AI talents. on the whole, biometrics belongs to the field of artificial intelligence, and biometric recognition has also received great attention in recent years. From the earliest fingerprint to face, from 2D face to 3D face, it can be expected that the iris will become a new technical hotspot in the next three to five years. Especially the multimodal combination of iris and human face. The job demand is shown in the following figure:

That's all for today's sharing. Thank you.

Https://www.leiphone.com/news/201910/mM0n4eGFkomcDxIg.html

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.