Computational Sensors Corporation

Biologically-inspired hardware for sophisticated imaging tasks


Company	Products	Technology	Projects	Demos	Home
Management	Customers	Services	Brochure	Careers	Contact Us

Research Areas

Computational Sensors Corp.’s powerful multi-node digital processor system, running under a VxWorks operating environment, allows real-time algorithm completion supporting our high-capacity analog VLSI image convolution chip technology.

Analog VLSI Spatial Filtering and Motion Energy Processing
Computational Sensors Corp.’s VLSI System operational parameters are optimized through training to identify targets and perform clutter discrimination, and they are used to control the programmable hardware to maximize contrast between the target and its background. Analog VLSI image processing hardware can natively perform the necessary spatio- temporal operations on live image sensor inputs in real time.

Sample of a motion energy output computed in real-time using a 60 Hz camera input and a spatio-temporal motion processing unit combining multiple CSC ASIC's performing spatial and temporal filtering under digital supervisory control. The output for the hand motion is sensitive to vertical motion with upwards motion color coded red and downwards motion shown in blue.		Spatio-Temporal Motion Energy Computation
		The unprecedented computational capability of the non-linear spatio-temporal motion energy image processing system based on Computational Sensors Corp.’s analog VLSI image processors provides sophisticated robust discrimination of subtle target characteristics, both spatially and temporally. The hardware realized target analysis capabilities are ideally suited for compact, low power, military applications ranging from remote sensing platforms to micro-unmanned aerial vehicles. Originally developed to assist the Missile Defense Agency (MDA) program in algorithms for boost, mid-course and end-game discriminating segments that flow down to major missile defense systems, the core technology behind our spatio-temporal motion energy processor is designed for easy transition into broader missile defense markets such as military laboratories and aerospace corporations.

Wavelet Technology
Computational Sensors Corp.’s wavelet filter bank architectures are enabling for best-basis search methods for choosing optimal wavelet filters from known target characteristics supporting target discrimination applications. Our analog real-time video processing can extract a more comprehensive generalized set of features from a wavelet filter bank. Wavelet-based filter bank solutions can dramatically benefit from the computational capability of analog silicon retinas in multi-chip architectures, enabling robust discrimination without an over reliance on a priori target information.

Directions & Map | Local Interests | Request Info