Developed the world's first back-illuminated CMOS camera optimized for near-infrared (NIR) response and is ideally suited for low light NIR imaging applications including microscopy, hyperspectarl imaging, biometrics, medical imaging, and surveillance.
· 1280 x 1024, 30 fps
· Flexible ROI settings
· CameraLink data port
A complete unit design was performed including:
- Hardware architecture development,
- Full set of real-time video processing algorithms: 3D adaptive noise reduction (DNR), piece-wise Gamma, and automatic self-adjustable histogram equalizer.
- Verilog modules for large FPGA device using ActiveHDL, Synplify, Quartus (including Nios sort core processor),
- Analog front-end including anti-aliasing filters, DC restore circuit, 16-bits ADC,
- CameraLink video transfer interface and USB for ROI definition,
- Win32 application for ROI definition with user-friendly GUI.
Developed a LED-based video display processor including general concept, hardware architecture, distributed network fundamentals; and finally a pure Verilog synthesisable ASIC code was delivered.
In progress of development a few innovations were discovered: highly effective video processing algorithm, adaptive Manchester decoding algorithm, and sophisticated automatic recovery circuit.
The development was including:
- Image visualization by distributed pixels network
- Image processing for outdoor video visualization
- Noise immune communication network including adaptive filtration and synchronization methods
- Image processing for system testing and calibration.
Developed real-time FPGA-based video processing system including general concept, hardware architecture, software model, Verilog synthesisable code development, ORCAD schematic drawing, working prototype demonstration and full project documentation release.
In progress of development a few innovations were discovered:
- Objects detection for very hard outdoor condition (wind, cloud, rain, fog, etc.) by using 3-D adaptive filtration methods.
- Objects tracking, by objects color tone, texture, and shape analysis.
- Objects recognition (human-animal-vehicle) was based on collecting information (properties) for each object, building objects properties data-base, data-base records analysis including objects cross-influence.
Developed real-time FPGA-based video processing system for large size LCD/Plasma displays including:
- Highly effective Adaptive 3-D noise reduction algorithm;
- MPEG/H.xxx compression artifacts reduction by adaptive non-linear 3-D image processing;
- Adaptive local and integral non-linear transform: gamma, histogram equalization;
- Sharpness enhancement without "ringing effect";
- full real-time FPGA/ASIC based approach for 720p, 1080i, and 1080p video modes.
Development was included a general concept, hardware architecture, software model, Verilog synthesisable code for Altera, Xilinx FPGAs, and ASIC, OEM board development and production, and full project documentation release.
HDTV real-time video improvement processor
Smart video surveillance system
LEDs-based video display processor ASIC
High definition low light NIR camera
Piece-wise non-linear video transform
Pipelined 3-D adaptive video filter
Highly effective algorithm for real-time FPGA/ASIC platform including SDRAM-based temporal (frame-by-frame) recursive (IIR) video filter with tunable recursive coefficient, and 2-D Finite Impulse Response Gaussian filters.