OKI (News - Alert) Electric Industry has announced the release of a face recognition middleware suitable for the surveillance market. The “FSE (Face Sensing Engine) Ver.4” for embedded systems will be able to detect faces as well as identify specific human beings from surveillance videos.

OKI has said it will launch an embedded video recognition module which includes a human tracking function soon. When the face is detected from a video the system will continue to track the specific person through other features even if the person’s face cannot be seen in the video recorded by the camera. This is made possible when the video recognition module is combined with the system.

The FSE Ver.4 has the capability to reduce the network load from surveillance systems as it has always posed problems when large volumes of video data are sent. Now it only necessitates sending parts of the video or transmits specific images over the network instead of sending all the video data.

The combination of FSE Ver.4 with an object tracking function enables a system to monitor the behavior of an identified person. Besides, with the digital video recorder, data capacity can be reduced at the same time retaining images of suspicious figures to be present in the recorded data.

“OKI’s award-winning FSE has been implemented in various applications including digital cameras for auto focusing functions, mobile phones for security and for entertainment-use, etc.,” said Seiji Inoue, General Manager of Human Sensing Venture Unit at OKI.

“We are pleased to introduce the FSE Ver. 4, featuring improvements made in response to the needs of the surveillance market. We believe the product will help the market offer more efficient surveillance systems utilizing biometric technology. "

Basic features now include the FSE Ver.4 which can detect and capture faces from broader angles enabling face detection from cameras fixed on the ceiling. This reduces network load and the system can select only relevant videos or images for transmission across the network.

And besides the conventional API, FSE Ver.4 includes a new API for expandability. This results in better functions.

OKI has also increased extracting feature points which enable more accurate identification of individuals. The feature points are specific points of the face like the location of eyes, mouth and facial contour in an image. The movement and status of these facial parts can be detected by detecting the moves of these points.