SHANGHAI -- VeriSilicon (688521.SH) announced the launch of its latest AI-based AI-NR and AI-SR series of image processing IPs, including the AINR1000 and AINR2000 for intelligent noise reduction, and the AISR1000 and AISR2000 for advanced super resolution. These series IPs provide cost-effective, highly efficient, and PPA (performance, power efficiency, area) optimized implementations with high flexibility and scalability, making them ideal for a wide range of applications, including automotive, surveillance, cloud gaming, consumer electronics, and more.

VeriSilicon’s AI-NR and AI-SR series IPs leverage the company’s proprietary AI pixel processing algorithms to deliver advanced image processing capabilities. The AINR1000 and AINR2000 IPs effectively suppress noise while preserving intricate details and color accuracy, delivering exceptional performance in low-light or complex lighting conditions. These IPs effectively reduce noise in both static and dynamic scenes, ensuring consistently clear and smooth visuals. The AINR1000 supports a resolution of 2560x1440 at 30fps, making it well-suited for low-resolution, cost-sensitive applications such as entry-level Internet Protocol Cameras (IPCs), while the AINR2000 delivers superior performance with a higher resolution of 5600x4208 at 60fps, catering to high-end applications such as advanced surveillance systems and sports cameras.

The AISR1000 and AISR2000 IPs deliver advanced AI-based super resolution, enabling high-quality upscaling from Standard Definition (SD), High Definition (HD), and Full High Definition (FHD) to 4K across various input sources, including real-time video, cloud gaming, and virtual cloud desktops. These IPs leverage innovative operator-layer combinations to enhance fine details and eliminate jagged edges, resulting in sharper, more natural high-resolution images. The AISR1000 supports up to 4K at 60fps, while the AISR2000 leverages dual-engine architecture to achieve 4K at up to 144fps, optimizing both performance and power efficiency. With support for flexible scaling ratios up to five times, the AI-SR series IPs offer superior visual quality and scalability, making it ideal for applications requiring real-time, high-resolution image enhancement.

Leveraging VeriSilicon’s innovative FLEXA interface communication technology, these IPs can be seamlessly integrated with VeriSilicon’s Image Signal Processing (ISP) IP, Video Processing Unit (VPU) IP, and Display Processing IP, enabling efficient and flexible incorporation into existing imaging pipelines and enhancing overall system performance. Additionally, these IPs can also be integrated into third-party pixel pipelines, offering flexibility for a wide range of use cases.

“Our AI-NR and AI-SR series IPs are designed to significantly enhance image quality by leveraging the latest AI technology, with a strong emphasis on low power consumption and silicon efficiency. These IPs have been successfully adopted by multiple customers across a wide range of devices, including battery-powered cameras and mobile application processors. Additionally, our AI-SR series IPs have been deployed in data centers to enhance video transcoding for AI-driven video applications,” said Weijin Dai, Chief Strategy Officer, Executive Vice President and General Manager of IP Division of VeriSilicon. “Our solutions integrate seamlessly to deliver comprehensive image enhancement, enabling our customers to accelerate the development of next-generation products with superior performance. Looking ahead, we are expanding our portfolio of AI-based image pre-processing (Image Reconstruction) and post-processing (Image Enhancement) technologies, continuously introducing new capabilities to further improve image quality and efficiency.”

The AI-NR and AI-SR series IPs are now available for licensing and integration.