Fuzzy Logic-Based MicroBattery for MEMS Sensors

Not Available Force protection can greatly benefit image understanding technologies. In most battlefield scenarios, it is valuable to be able to assess the area of operation prior to deployment of troops. Video sensors deployed in the field provide valuable visual information that can be used to develop strategies prior to sending in troops. A major limitation of existing video sensors is field of view. It is highly desirable to have sensors that provide complete coverage after deployment. From this perspective, omni-directional video cameras present major advantages over conventional ones. The 360 degree coverage provided by such a sensor enables a remote user to look around and assess the battlefield. Today, the main drawback of omni-directional cameras is their limited resolution. Since a complete hemispherical fields is projected onto a single CCD detector, far away objects are only imaged onto a small numbers of pixels. For instance, when a standard NTSC video camera is used in conjunction with omni-directional imaging optics, a human 50 meters away will appear within just 5-10 pixels in a typical omni-directional image. This proposal is geared towards the development of high-resolution omni-directional cameras. The implementation of such sensors requires not only the use of a high-definition video camera but also the design and fabrication of accompanying optics to generate the desired field of view. Our optical design will be a catadioptric one, where mirrors are used in conjunction with lenses to project the large field of view onto a single planar image detector. This design procedure will require the use of sophisticated optical design techniques. In addition to the design of a high-definition omni-directional sensor, we propose the implementation of a series of low-level image processing and image understanding algorithms. These algorithms will include real-time generation of perspective video, real-time tracking and monitoring of multiple moving objects, fast ego-motion estimation for applications where the omni-directional sensor is mounted on a moving platform, as well as higher-level algorithms for activity and object recognition.