Deliver superior results in extremely low-light conditions, surpassing conventional computer vision technologies
SHANGHAI, China--BUSINESS WIRE--
VeriSilicon (688521.SH) today unveiled its ISP9000 series Image Signal Processing (ISP) IP, a next-generation AI-embedded ISP solution designed to address the evolving demands of intelligent vision applications. Built on a flexible AI-optimized architecture, ISP9000 delivers exceptional image quality, low-latency multi-sensor management, and seamless AI integration, making it ideal for advanced use cases, such as intelligent machines, surveillance cameras, and AI PCs.
With its AI-powered ISP capabilities, VeriSilicon’s ISP9000 series IP delivers superior image quality. It integrates advanced AI noise reduction (AI NR) algorithms, combining multi-scale 2D and 3D noise reduction with YUV-domain chroma noise reduction (CNR) in a multi-domain noise reduction architecture. This effectively minimizes noise while preserving fine details, particularly in extremely low-light conditions. Its support for triple-exposure High Dynamic Range (HDR) processing and dynamic range compression, with a 20-bit pipeline, ensures that high dynamic range scenes retain critical details in both bright and dark areas. Beyond that, the ISP9000’s 3A capabilities, including auto exposure (AE), auto focus (AF), and auto white balance (AWB), support up to 25 configurable regions of interest (ROI), which enables AI-assisted object detection and recognition. Additionally, users can integrate third-party 3A libraries, providing the flexibility to tailor 3A algorithms for specific application needs.
The ISP9000 supports multi-sensor configurations and leverages hardware-accelerated multi-context management (MCM) and frame slicing mechanisms to enable efficient data stream switching. This ensures the stability of multi-sensor data streams while providing low latency and cost efficiency. Supporting up to 16 sensors, the ISP9000 integrates VeriSilicon’s VI200 video interface IP for seamless connectivity with mainstream MIPI Rx interfaces. Its scalable multi-core architecture supports high-performance processing with up to 8K@30fps and 4K@120fps. Through VeriSilicon’s FLEXA SBI interface, the ISP9000 enables optimized data transmission to video encoders, Neural Network Processing Units (NPUs), or display processors, enabling optimized subsystem solutions.
Furthermore, the ISP9000 is equipped with VeriSilicon’s AcuityPercept, an AI-powered automatic ISP tuning system, to enhance its intelligent vision capabilities. Unlike traditional tuning focused on human-perceived image quality, AcuityPercept optimizes the ISP settings for superior object recognition in the AI/NPU pipeline, adapting to specific AI algorithms and application requirements.
“As AI becomes increasingly embedded in devices across a wide range of applications, and with the rapid advancement of robotics, the demand for next-generation ISPs to meet evolving product requirements continues to grow,” said Weijin Dai, Chief Strategy Officer, Executive Vice President, and General Manager of the IP Division at VeriSilicon. “In addition to delivering high image quality and low power consumption, next-generation ISPs must efficiently support multiple sensors with fast context switching and low latency, and optimize output for both NPUs and human vision. They should also enable automatic tuning in conjunction with target perception NPUs. The ISP9000 is architected with these requirements in mind and has been developed through close collaboration with leading customers. By embedding AI technology, ISP9000 achieves exceptional image quality in extremely low-light conditions, surpassing the capabilities of traditional computer vision technologies.”
About VeriSilicon
VeriSilicon is committed to providing customers with platform-based, all-around, one-stop custom silicon services and semiconductor IP licensing services leveraging its in-house semiconductor IP. For more information, please visit: www.verisilicon.com
View source version on businesswire.com: https://www.businesswire.com/news/home/20250401360621/en/
Contact details:
Media Contact: [email protected]
