On March 17, 2014, CEVA, a leading licensing company for SIP (Silicon Intellectual Property) platform solutions and DSP cores, today enhanced the real-time library of CEVA-CV computer vision to support the simplification of programming and computer vision development processes for a wide range of end markets, including mobile, automotive, surveillance, consumer electronics, and the Internet of Things, with over 750 functions.

CEVA facilitates the integration of CEVA-CV functions into various Android-based application processors by extending the CEVA-CV computer vision library with the CEVA Android Multimedia Framework (AMF). Furthermore, since all computer vision processing for both the CPU and GPU is handled by the CEVA-MM3101, it has not only significantly improved the performance of CV devices but also enabled reduced power consumption.

CEVA demonstrated the expanded libraries of CEVA-CV, the Android Multimedia Framework (AMF), and the latest computer vision applications at the GSMA Mobile World Congress 2014 held in Barcelona, Spain last February.

New features added to CEVA-CV include feature detection kernels and target recognition algorithms such as Harris Corner, Hough Transform, Integral Sum, Fast, LBP, SURF, HOG, SVM, and ORB detection and matching. These features are primarily used in augmented reality applications for smartphones, tablets, wearable devices, Natural User Interfaces (NUI), surveillance, and Advanced Driver Assistance Systems (ADAS).

The new optical flow kernel includes KLT and block matching, which are used for motion detection and target tracking functions required by camera devices. Through this technology, digital video stabilization, augmented reality, and gesture recognition functions of camera devices can be used.

The kernel newly included in CEVA-CV follows the Khronos Group's OpenVX 1.0 standard and is designed to serve as a major standard for cross-platform acceleration for computer vision applications and libraries.

“The CEVA-MM3101 is widely used as a proven computer vision platform in the industry. Numerous leading semiconductor companies and OEM licensees are already using the CEVA-MM3101 for next-generation products and developing related applications through the CEVA-CV capabilities,” said Erez Bar-Niv, CTO of CEVA.

He went on to emphasize, “The addition of over 250 new computer vision features has been highly praised by developers using the CEVA-MM3101. Through this, relevant developers can develop the most advanced CV applications while simplifying complex designs and reducing power consumption of CV-connected devices.”