A Brief Talk on Practical Application and Prospect of Image Processing in Daily Life

A Brief Talk on Practical Application and Prospect of Image Processing in Daily Life

*Yixuan FAN(范艺璇） Glasgow College, University of Electronic Science and Technology of China *

I am not an expert in image processing. I can not talk about the algorithm or the ideas in the field, but I would like to mention some contributions of image processing in our daily life.
Face scanning is becoming more and more popular in the 21st century. The technology, face recognize is an application of image processing. Face recognize system first widely appeared in bank. It is very common that many banks have used a face recognition verification system for supplementary authentication. As the technology is not completely reliable and the accuracy is still in need of improvement, face recognition system can not replace other traditional verification process. Next it is applied in the access control system, which used to be fingerprints or irises recognition. It is undeniable that the identification of fingerprints and irises is more accurate. However, many people might feel uncomfortable of recording fingerprints (collecting fingerprint information) or using their eyes to smash the camera every day (acquisition of iris information). Further more, some people argue that it is invasive in the process of collection and verification. In contrast, the recognition of the face is much more convenient, and most people do not mind taking a photo (collecting face information). For example, my university have the face recognize access control system in the library. But some student can not pass every time or they might be recognized as another person. Monitoring is also deserve to be mentioned in the application of face recognition. Face recognition system can make a huge difference in monitoring. As we know the monitoring of a camera can record hundreds of people from different angles (such as the monitoring of dense places such as stations). If the videos of the monitoring can only be identified by human beings, it will be a huge project. However, with the use of face recognition technology, the system can automatically identify face information. It will undoubtedly bring great convenience to police. Let’s have more discussion about security monitoring. Security monitoring might be the most promising application area in the field of image processing. Nowadays, every city installs surveillance cameras in an incredible number. The countless cameras in the country insure the recording, but the back-end processing of security monitoring has not kept up. The back-end processing can be image processing of surveillance video. Note that image processing here can include not only face recognition, but also pedestrian detection, anomaly detection, saliency detection, and cooperative tracking. As face recognition has been mentioned in previous, I would like to explain pedestrian anomaly detection here. Pedestrian anomaly detection can make a camera monitor the video to determine who is a good person and who is a bad person in a playing video (of course, it is too arbitrarily divided into good and bad people) seems impossible in general thinking. It seems that is too difficult. However, most analyzing processing of surveillance video is done by people. That means when the police is solving a case, they often watch the surveillance video of the last few days from the beginning to the end which is time-consuming. Because there is a practical need in this regard, the study of intelligent monitor is in need. Although we would like to have a smart monitor, our video analysis program does not directly give arbitrary and one-sided judgments such as “good guys or bad guys.” As far as the current technology is concerned, it is possible to count the number of people in a playing monitoring video(pedestrian detection), locate faces (face detection), identify identities (face recognition), discriminate their expressions (expression recognition), and detect their actions (abnormal detection). And these functions already reduce amount of work. In this way, people will no longer watch dozens or even hundreds of hours of surveillance video, but directly analyze the data given by the computer such as how many people in the video, who are they, and whose actions are suspicious.