Across industries, real-time image processing has the potential to revolutionize how we engage with technology.
More and more, electronics across technology platforms are beginning to rely on high-speed processing to capture and interpret an image, enabling a system to respond immediately with an appropriate action. From security systems to manufacturing and remote sensing to multimedia, the ability to perform image processing quickly is hardly a perk—it’s become an expectation.
Nextchip's Apache4 SoC pairing with CEVA's XM4 image processing platform is a step towards making this technology more common, especially in the blossoming field of autonomous vehicles and its reliance on ADAS.
Applications of Real-Time Image Processing
Though it is becoming more common, there are a few applications for real-time image processing that promise big advancements. Researchers at the University of Texas at Austin, the University of Texas at San Antonio, and the University of Texas MD Anderson Cancer Center are working on a supercomputer that performs minimally invasive laser treatment without a surgeon.
Using an adaptive-feedback control system based on mathematical and computational models that use magnetic resonance temperature imaging to determine the heat transferred by a laser to the tissue and then the tissue’s response, researchers hope to revolutionize cancer surgery.
Magnetic resonance temperature imaging used in thermal therapy. Image courtesy of David Fuentes, University of Texas MD Anderson Cancer Center
Another application for real-time image processing collision avoidance. By building infrastructure enhancements in "smart cities" and passive safety features into vehicles, collision avoidance is greatly enhanced, leaving both pedestrians and drivers safer when they’re on the road.
In biometrics, real-time image recognition can also be helpful. Fingerprint detection that responds quickly can help users gain timely access to secure systems, while facial recognition has the potential to help law enforcement officers scan large amounts of footage for subjects in a short amount of time.
There are uses in agriculture, motion detection, and even traffic control—real-time image processing shows no signs of stopping.
Nextchip, CEVA, and ADAS
Nextchip Co., a fabless company which specializes in embedded vision applications for ADAS systems, recently announced that they’ve licensed the CEVA-XM4 imaging and vision platform for its APACHE4 vision-based pre-processor, geared toward real-time ADAS vision systems.
In an attempt to provide an affordable ADAS application, the company incorporated CEVA’s programmable vision platform into APACHE4 along with its differentiated image processing accelerators.
“We developed the APACHE4 to provide affordable and scalable ADAS systems for the mass market. CEVA’s industry-leading vision platform adds a high degree of flexibility to our solution to enable differentiated, machine vision-related ADAS products,” says Kyoungsoo Kim, Nextchip CEO.
Image courtesy of Nextchip.
APACHE4 is a vision-based pre-processor SoC, which targets next-generation ADAS systems and has a dedicated sub-system of image processing accelerators and accompanying software. Nextchip claims that it significantly reduces the primary ECU’s workload while allowing detection algorithms to operate simultaneously. It also has a detection engine dedicated to pedestrians, vehicles, lane detection, and moving object detection. They also boast that, with an embedded CEVA-XM4 imaging and vision platform, designers will be able to develop differentiated ADAS applications with advanced software programming.
Most machine learning and vision applications used across industries require a high level of processing and low power constraints, which CEVA’s most recent imaging and vision DSP platforms address. The DSP-based platforms rely on both scalar and vector DSP processors and all-inclusive ADK to unify software deployment. CEVA ADK includes CEVA-LInk for software integration with a host processor, a variety of commonly used software algorithms, CEVA Deep Neural Network software framework, and development and debugging tools.