About the Event
An energy-efficient adaptive image sensor for distributed sensor network and wireless biomedical imaging systems has been implemented. The sensor provides three adaptive functions including energy-adaptive imaging for low-power consumption, illumination-adaptive imaging for high-dynamic range, and object-adaptive imaging for temporal-spatial bandwidth suppression.
For energy/illumination adaptation, we designed a reconfigurable CMOS imager which mostly operates in monitoring mode at 1.36 μW from harvested energy, and upon request changes its mode to capture high-sensitive (24 V/lx-sec) and wide dynamic range (99 dB) images at 867 μW in battery operation. The chip achieved a power FOM of 15pW/pixel·frame in 0.18 µm CMOS technology.
In order to suppress the spatial and temporal bandwidth redundancy, we implemented a motion-triggered and object-adaptive imaging scheme. The sensor wakes up triggered by motion sensing and extracts features from the captured image for the detection of object-of-interest. Full image capturing operation is performed only when the interested objects are found, resulting in significant reduction of total power consumption at the sensor node. The object detection result shows a 94.5 % success rate for human recognition.