Low Power Depth and Velocity from a Passive Moving Sensor
E. Alexander, Sanjeev Jagannatha Koppal, Todd E. Zickler · 2015
We present an opportunity for the visual sensing of depth and 3D velocity using a passive sensor that has extremely low power requirements. This opportunity comes from a new mathematical constraint, which we derive, that relates depth and velocity to spatial and temporal derivatives of image values captured by a coded-aperture camera that observes a moving scene. The constraint exploits the fact that there are two causes of brightness change in this situation: features move across the image due to motion, and contrast changes because of time-varying optical blur. The sensor that could be realized from this constraint is called a focal flow sensor. We analytically characterize the working volume of such a sensor in relation to its size, and we provide simulation results that affirm its viability.