Today, due to the emergence of powerful, low-cost, energy-efficient processors, it is possible to build vision into a wide range of embedded systems. However, there are two big challenges to implementing embedded vision systems: complex signal processing algorithms and system optimization requirements.
Complex signal processing algorithms
Signal processing algorithms form the heart of embedded vision functionality. Signal processing is used:
- At the pixel level, to extract objects from images
- At the object level, to track, evaluate, “see” and “understand” the behavior of objects
Successful embedded vision development requires a command of both algorithm and implementation aspects—and the interplay between the two.
System optimization requirements
Embedded vision algorithms are extremely computationally intensive. Fitting these “big algorithms” into low-cost implementations is tricky, requiring creative, multi-level hardware and software optimization strategies.
How BDTI Can Help
If would like to build vision capability into your product but don’t have the expertise in-house to meet the challenges of vision processing, BDTI can help. BDTI provides a range of engineering services for design and implementation of vision applications. Our team of engineers has years of experience building complex, reliable, and low-cost signal processing systems, including video analytics and computer vision applications. We can work with you to understand your application, identify your requirements, select the best technology—hardware and software—and then design and build your system. We can work closely with your engineering team, handling only the video analytics/computer vision portion of the design, or we can execute a complete “turn-key” project, delivering a complete, finished system. In addition, we can leverage our own existing intellectual property (e.g., object tracking algorithms) or develop custom IP for your application.
Example BDTI Embedded Vision Projects
Some recent examples of BDTI video analytics, computer vision, and related projects include:
- For a global processor vendor, a complete demonstration system for a new processor targeting industrial machine vision applications. The demo shows the usefulness of a new, unique processor architecture that incorporates a configurable vision processing pipeline. (Read about the demo.)
- For a major semiconductor company we prototyped computer-vision-based tripwire and object recognition algorithms, using a combination of an FPGA and a processor. (In the image at right, the software identifies a pedestrian crossing a line into a “danger zone” and marks him with a red box. A pedestrian outside the “danger zone” is indicated by a green box.)
- For a scientific instrumentation company we developed a mathematical model for a vision-based feedback control system, shaving weeks off the company’s development time and enabling the company to improve throughput of its equipment.
- For testing high-level synthesis tools, we designed and implemented a complex, demanding optical flow algorithm, used to detect apparent motion in an HD video stream.
- For a consumer electronics company we created a real-time FPGA-based prototype of a proprietary video display enhancement algorithm.
How to Engage BDTI
BDTI can add computer vision and video analytics to your products faster and with less risk than you may have thought possible. And that translates to faster time to market and increased revenue! Contact us today to discuss how video analytics and computer vision can enhance your products.
For a confidential discussion of your needs, call BDTI +1 (925) 954-1411 or use the form under “Contact” to the right to contact us.