How Wisenet 9 Enhances Video Surveillance with Edge AI and Dual NPUs
Wisenet 9 is a next-generation System on a Chip (SoC) developed by Hanwha Vision, designed to support advanced video surveillance applications. It features a dual Neural Processing Unit (NPU) architecture, with one NPU dedicated to image processing and the other to analytics. This separation allows for consistent performance in both areas.
AI-Based Image Processing
The SoC includes an AI-based Noise Reduction (NR) system that operates independently from other processing tasks. This system reduces noise caused by sensor sensitivity and low-light conditions, helping to preserve image detail and minimize motion artifacts. The Image Signal Processor (ISP) then applies further enhancements to optimize texture, color, and tone.
Low-Light and Dynamic Range Performance
Wisenet 9 is designed to improve video clarity in environments with challenging lighting. It enhances visibility in both bright and dark areas of a scene, supporting identification of key visual elements such as faces and license plates. The system also includes multi-frame extreme Wide Dynamic Range (WDR) technology, which adjusts tone and contrast to improve clarity in high-contrast scenes. AI-based enhancements further refine exposure and reduce motion artifacts.
Bandwidth Optimization
The SoC incorporates WiseStream, an AI-based compression technology that uses object detection to reduce bandwidth usage. When combined with H.265 compression, this approach maintains video quality while optimizing storage and transmission efficiency.
Customizable AI Capabilities
Wisenet 9 supports the development and integration of custom AI applications. This flexibility allows the system to be adapted for specific industry requirements, enabling tailored analytics and functionality.