VD55G4

预览

产品尚在设计阶段

超低功耗 800×700 单色全局快门 CMOS 图像传感器，专为常开视觉应用而优化

描述

The VDx5G4 is a unique ultracompact and low-power image sensor in the ST BrightSense portfolio. It has been developed to enable the next generation of smart and power-efficient systems. Two versions are available: monochrome and RGB Bayer. Leveraging the world’s first 2.16 μm global shutter pixel, the sensor highlights maximum compactness. It has a tiny 2.73 x 2.16 mm² footprint that fits into the most size-constrained devices such as smart glasses, XR and wearables, personal electronics, or laptops.

VDx5G4 is optimized for low power consumption across a wide frame‑rate range, ideal for battery‑operated AI systems. A dedicated low frame‑rate autoexposure and auto‑white‑balance mode precompute exposure and color parameters directly on‑sensor before each frame. By offloading these iterative calculations from the host, it accelerates convergence, cuts end‑to‑end latency, and reduces host CPU or NPU activity. This architecture significantly lowers overall system power, while maintaining robust image quality even in challenging or dynamic lighting conditions.

An embedded auto wake‑up function further enables ultralow‑power “sleepy” monitoring, instantly switching to full streaming when an event is detected. In addition to this low operating power, the sensor embeds an auto wake-up feature. This enables the sensor to operate in an ultralow power sleepy mode when no change or event is detected. As soon as an event is detected, the sensor automatically sends a trigger to its host processor to switch into full streaming mode to capture the corresponding event. Thanks to this clever feature, vision system makers can finally get rid of the power-hungry, ineffective 24/7 operation of systems such as security cameras, auto-ID, or in-line product inspection.

Beyond power optimization, the VDx5G4 embeds a complete toolbox of on-chip features for optimizing image quality, adjusting frame rate, or improving data efficiency. Autoexposure, noise reduction, and various image corrections help to get stunning images directly from the sensor. This sensor is tailored for computer vision applications with dedicated features. The frame statistics can be output, allowing fast and low computer vision application. This sensor can capture event frames in an event-like mode, enabling synchronous output of event information. A background subtraction mode can also be activated to seamlessly differentiate the foreground.