Alcorn Receives Face Recognition Patent


Dr. Yufeng Zheng, associate professor Department of Advanced Technologies

Dr. Yufeng Zheng, associate professor in the Department of Advanced Technologies, was recently awarded a patent by United States Patent and Trademark Office for his invention. Zheng is the principal inventor for the utility patent (8,917,914), “Face recognition system and method using face pattern words and face pattern bytes.”

The present invention provides a novel system and method for face recognition utilizing facial features. The system and method of the invention comprise the creation of face patterns called “face pattern words” (FPWs) and “face pattern bytes” (FPBs) for face identification.

This newly-invented technique can be applied to visible and thermal face recognition, and the recognition accuracy is substantially improved. In other words, the accuracy of this face recognition technique is comparable with that of fingerprint or iris recognition.

The face recognition system can be used in various occupations.

“The face recognition technique uses biometric measurement (i.e., face image) for human identification. However, unlike fingerprinting and iris scanning, imaging a face does not require much collaboration from a user and; thus, performs faster. This technique can be widely applied to customs and borders as well as the court system and police offices,” he said. “In the future, the face recognition technique will be used wherever and whenever showing a picture ID is required.”

According to Zheng, this face recognition technique is expected replace or complement fingerprint recognition in the near future. This technique is well suitable for the applications in crowd sites, such as, customs and borders. Meanwhile, the thermal face recognition capability in this technique enhances the facility security at nighttime, for example, in federal buildings and military bases.

“Imagine that you are entering the United States Customs after your foreign vacation; your face picture is automatically taken when you enter the line. You are identified from the national face database and further recognized as a ‘good citizen’ after matching other databases,” he said. “Then you are instructed to walk to the ‘Green Lane.’ The whole process takes only a few seconds when you walk through the line.”

Zheng added, another future application will be to register your face with Walmart, and associate it with your credit card; thus, every time you finish your shopping at Walmart, you can walk out by “swiping your face” through a camera (other than a card) – assume all items can be automatically scanned using RFID (radio frequency ID, pasted on each shopping item).

The FPWs and FPBs are sets of face patterns created from a face image. The FPWs and FPBs are then used for face identification, and the identification rate and recognition speed using FPWs and FPBs are significantly improved. The face images can be either visible images (red, green and blue color model, i.e. RGB, for example) or thermal images (long wave infrared, i.e., LWIR, for example). Another embodiment of the present invention provides a computer software means for accurate face recognition using visible and/or thermal images.

The system was designed, according to Zheng, because the current face recognition techniques (such as linear discriminate analysis [LDA], elastic bunch graph match [EBGM]) have low accuracy or are a slower process. The invented technique (FPWs, FPBs) has not only high face recognition accuracy but, also, high face matching speed. To identify a face image, the computer has to search for it from hundreds of millions of faces stored in the database. In addition, it is proven that the invented technique performs face recognition using thermal images as well as using visible images; thus, allowing the face recognition system to fully function both day and night without a requirement for illumination.

As for the next invention, Zheng stated, we are continuously improving this technique in terms of accuracy, speed, and reliability. We are preparing for the next phase – technology transfer and product commercialization.

“I feel excited about the patent. The invented face recognition technique has wide applications related to human identification,” said Dr. Zheng.

For more information on the invention, contact Dr. Zheng at (601) 877-6490 or [email protected] .

Links to the patent: