To solve the problem of detecting when a person is in a certain room we have decided to implement a 2 to 3 stage detection array. Each of the detection stages have a inherent flaw that prevents them from successfully detecting a person as a stand-alone system. By using 2 or 3 of these detection methods in parallel we should be able to eliminate those flaws. The methods of detection are:
1. Passive InfraRed (PIR) sensor: detects changes in ambient IR landscape and outputs a single clock pulse at every detected change. However, if a person is still for a period of time, or isn't moving fast enough, the PIR sensor will not produce a pulse. This could be our "first alert" detector placed near entrances.
2. Microphone detection: subtract pre-recorded samples of ambient noise from a live microphone to detect footsteps, breathing, etc
3. Video camera: subtract a digital image of the empty room from images produced by a live camera to detect "blobs" of data indicative of a person. Foreseeable problems include changes in the room's layout or objects in the room (kids' toys, etc.).
By using these three elements and the DE2 to make logical decisions we can potentially develop a "smart" room capable of detecting a person despite the aforementioned difficulties.
Video recognition has been eliminated as a possible detection method due to the lack of an Altera core. This core would interface the TV decoder with the NIOS II in the same way that the audio CODEC is interfaced. Since no such core exists and time is running short, we have decided to drop the video method of detection.
In order to make the system as effective without video we will employ IR sensors like those used in garage door systems. These sensors, though expensive, may be placed at entrances to rooms for an additional method of detection. In concert with the audio and PIR sensors, the system will be able to successfully identify if people are a given room.
Finalized Detection Method:
The three types of sensors that we were able to get working in concert are listed below. Sensor distribution for a basic 2-door room can be seen on the test room page.
1. PIR Sensor: one sensor for every room.
2. Audio detection: one microphone for every room. A sample is compared to a scalar value representative of an empty room to determine if extra noise is present. No pre-recorded baseline is necessary.
3. Laser/CDS beam: a laser is focused on a CDS cell. When the beam is broken, a high voltage signal is generated by a TTL "OR" gate.
MATLAB Image Processing
Sensor Design and Construction
VGA Sensor Display
Conclusions and Future Work