Wednesday, March 20, 2013

Fujitsu Measures Pulse from Face Video Stream

Tech-on: Fujitsu Labs pesented an interesting application to measure pulse by using video of a human face in real time.

The new technology detects pulse based on color elements of video by exploiting the fact that hemoglobin in blood absorbs green light and detecting color change on a facial surface caused by blood flow. An influence of facial and body movements (such as talking on the phone) is automatically removed.

"At the fastest rate, it can measure pulse in five seconds," said Yoshinori Yaginuma, director, Human Solutions Laboratory, Human Centric Computing Laboratories, Fujitsu Laboratories. "The maximum margin of error is about ±3 heart beats per minute."

"The required resolution is VGA or so," he said. "Twenty frames per second are enough. So, a camera of a personal computer, tablet computer or smartphone is sufficient."

The company is considering using the technology for simple health checkup at home, monitoring elderly people, automatic health problem detection in offices and detection of sick people and suspicious individuals in airports or at gates at event sites.


  1. Looks like the Eulerian Video Magnification for Revealing Subtle Changes in the World out of MIT:


  2. I would say the limiting factor here is noise, not spatial or temporal resolution. If my eyes cannot detect pulsing colors on someone's face, I highly doubt a smartphone's camera is sufficient in this task.

  3. Anonymous,

    As the MIT work shows, your eyes are not a good benchmark.


  4. Philips Vital Signs App for iPhone already does this. Available for $0.99.

  5. It seems that more people look into this kind of application. Another interesting approach is here:

  6. There was also this interesting news from fujitsu: they are merging their semiconductor division with panasonic LSI division.

    Areas of focus include ... 2) Visual and imaging solutions.
    It looks like this includes image sensors from panasonic. Anyone knows more about this?


All comments are moderated to avoid spam.