Embedded Image Processing Development
Machine Vision Systems Development
Custom image processing software
FPGA / GPU / NPP -based image processing development
Mixed-signal PCB design and layout (incl. low-noise analog and high-speed digital)
Image processing capabilities
- Anti-aliasing / edge-enhancement
- Color transform / correction
- Screening/halftoning
- Gamma/contrast adjustment/correction
- Auto exposure algorithms
- Computer vision algorithm development – including deep analysis of imagery and video and feature set development for surfacing of actionable intelligence
- Complex pixel reordering for fast lookup operations requiring DDR, QDR, RLDRAM memories.
- Pixel level and image sensor level analysis
- Linear and logarithmic high dynamic range hardware designs and software / firmware algorithms for capture and display
- Alternate color filter patterns and color processing algorithms for low-light color sensing
- Digital image fusion system using VIS-NIR, MWIR and LWIR imagers
- Pixel-level image correction for sensor dark frame variation
- Pixel-level image correction for linear output
- FPGA-based image processing
- GPU-based image processing
- NNP-based (Neural Network Processor) image processing
- False color, Laplacian Pyramid Fusion system
- Automatic gain/exposure correction
- Defect variation image correction
- FPGA-based H.264 compression cores
- Lossy and proprietary lossless compression algorithms
- Gamma & color correction
- Graphics overlay
- Image zoom/pan
- Real-time image rotation
- Image sensor board design – low noise, precision analog design with high-speed serial links
- Integration of broad range of CMOS and CCD image sensors as well as proprietary sensor technologies
- Low noise, low light analog (IT-EMCCD) and digital (CMOS) imaging systems for night vision
AI / Deep Learning Capabilities
- Applied artificial intelligence (image and video) – classification, localization, segmentation
- Implementation of state-of-the-art neural network architectures and deep learning algorithms
- Custom dataset capture process guidance and development
- Custom AI model development and training
- Deployment of custom AI models on the edge – utilizing various hardware platforms
- Inference engine tuning for critical product timing requirements
Supporting capabilities
- FPGA-based external memory interfaces (including DDR4 @ 2400 MT/s)
- SFP, SFP+, QSFP, QSFP+ small form factor pluggable optical transceivers
- Image sensor interfaces – LVDS, HiSPi and MIPI (C PHY and D PHY).
- Mixed-signal PCB design and layout
- Embedded system product design
