DMD Development Kit Setup and Issues 1.1 2008/04/15 17:38:42.285 GMT-5 2008/04/29 05:14:13.590 GMT-5 J. Ryan Stinnett jryans@rice.edu Jennifer Ann Gillenwater jgillenw@rice.edu J. Ryan Stinnett jryans@rice.edu Jennifer Ann Gillenwater jgillenw@rice.edu compressed compressive imaging lab After running into many dead ends trying to create a projector-based compressive imaging NIR camera, we did some work with a stand-alone TI board, in a lab table setup. Before we began working on this project, a similar setup already existed, as shown in . Our lab table setup is very similar, with the exception that we have black cardboard barriers to shield the photodiode from the light source. Our photodiode, light source, and target object are also of course different from those shown in . , , and show what the new setup looks like.

Original Lab-table Setup Original lab-table version of a compressive sensing camera .

Camera Front End Light source, target object, focusing lens, and DMD

Camera Top View TI 1100 Discovery Board and light source

Camera Back End Focusing lenses and photodiode

With this setup, we have direct control of the DMD mirrors; the TI 1100 Discovery Board has a high-speed interface that we can control using ALP (Application Programming Interface) software. To capture diode data, we use VILogger, then clean up the resulting stream in MATLAB. Thus, it seems probable that with such a setup, achieving a fast capture speed is within reach.However, we continued to experience problems on the NIR lighting and sensing ends of the camera. The light bulb burnt most substances we tried to mount it on, and had to be positioned just right in order to reflect enough light off the target object. Further, the diode data showed a lot of noise when the signal was low, as can be seen in . Over a long capture period, we also often observed a drift in the base signal, as shown in .

Noisy Diode Data Each set of pulses should be flat, as each is a set of the same random frame sent out repeatedly; the diode signal is clearly noisy

Drifing Diode Baseline Over a capture period of several minutes, the base diode signal decreases significantly

http://www.dsp.ece.rice.edu/cs/cscamera/ October 2007