Embedded Vision HDR Cameras Capture Contrasts Up to 120 dB

Related Vendors

FAULHABER Drive Systems

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

SIMTOS (Seoul International Machine Tool Show)

Innodisk focuses on HDR camera technology for industrial imaging and edge AI applications. The camera models are designed to capture details in bright and dark areas simultaneously, supporting AI systems under changing lighting conditions.

Cameras are no longer used solely in controlled environments for industrial applications. They monitor outdoor areas, support autonomous mobile robots, capture vehicles in parking systems, or provide image data for AI-powered inspection applications. This increases the demands on sensors: relevant image information must be preserved even in conditions with backlighting, reflections, or harsh shadows.

This is exactly where HDR comes in. High Dynamic Range extends a camera's dynamic range and ensures that bright and dark areas of an image remain simultaneously analyzable. For AI systems, this is less about image aesthetics and more about data quality. If the sensor loses details in overexposed or overly dark areas, it can affect recognition accuracy.

The models introduced by embedded manufacturer Innodisk include EV2U-LOM1-RHCF, EV2F-OOM3-RHCF, and EV3F-ZSM1-RXCF. According to the company, they achieve a dynamic range of up to 120 dB or approximately 20 stops. This allows brightness differences of more than 1,000,000:1 to be captured. The cameras have a resolution between 2 and 3 MP and a frame rate of 30 to 60 fps. Access to the data stream is provided via USB 2.0 or GMSL2.

In HDR processes, Innodisk differentiates between Multi-Exposure HDR and Single-Frame HDR. In the Multi-Exposure process, multiple images with different exposure times are captured and then combined. This can deliver high image quality but is primarily suitable for static scenes. In dynamic applications, such as conveyor belts, vehicles, or robots, multiple exposures can result in motion artifacts.

Single-Frame HDR, on the other hand, captures bright and dark image information within a single shot. Special sensor architectures utilize pixel areas with varying sensitivities for this purpose. After capturing, the image signal processor handles further processing, including tone mapping. This creates an image with an extended dynamic range without the delays or artifacts caused by multiple exposures.

For integration into embedded systems, Innodisk claims to support MIPI and GMSL2 interfaces. Sensors, lenses, and image signal processors can be adapted to different application scenarios. The company lists parking management, autonomous mobile robots in agriculture, self-checkout systems, and outdoor cameras as application fields.

This positions Innodisk HDR not as a mere camera feature but as a building block for industrial edge AI systems. The key point is whether the camera provides sufficient usable information under changing lighting conditions, allowing downstream algorithms to reliably recognize objects, people, license plates, or products. 

Link: Learn more on the topic at Innodisk