The Need for Embedded Ultra HDR™ Video in Vision Applications
In the past, developing higher quality video has been accomplished on two axis; higher resolutions for sharper, more detailed video and higher frame rates, based on faster pixel readouts in CMOS sensors to capture motion without blur and provide slow motion playback.
Now, a third axis is possible for creating higher quality video in high contrast situations, solving the problem of capturing enhanced data and detail in deep shadows, while holding highlight detail, with a single sensor. It’s called Ultra HDR™ video from Pinnacle Imaging Systems. Ultra HDR™ has the ability to add value, enhance data acquisition and provide higher quality in many video application segments.
Applicable Ultra HDR™ Segments
Today, signal processing of highly computational intensive tasks, using advanced logic, combined with highly developed algorithms makes it possible to provide dynamic range well beyond what any single sensor can provide. This opens up new opportunities for improvement in existing video applications, capturing more information for recognition and other computational tasks and providing higher quality video.
Pinnacle Imaging Systems’ Ultra HDR technology addresses many of the complexities involved in HDR video capture including:
- Automatic Ghost Removal & Halo Reduction – Compensates for movement between HDR exposures, from minimal camera motion to moving objects between frames
- Locally Adaptive Tone Mapping – Automatically optimizes the tone mapping parameters based on the shadow and highlight areas of each individual video frame to ensure a more natural look to the output video
- Automatic White Balance Controls – Automatically calculates proper white balance settings for any scene or lighting condition
- Automatic Exposure Controls – Real time calculation and adjustment of the sensor’s exposure settings based on an automatic or manually selected region of interest to allow accurate exposure throughout a scene
- Shadow Exposure Bias Option – Ability to bias tone mapping with an additional Shadow Tracking option during the Auto Exposure mode for optimal shadow detail and data preservation, a key requirement for surveillance applications
Ultra HDR™ Dev. Platform Resource Requirements
- Existing FPGA Based Platform (full video path)
- 40k slices, 308 DSP blocks
- 600k byte internal RAM (155 36k bit BRAM, 32 18k bit BRAM)
- 28G bit external DDR3 BW, memory for frame buffers
- Currently running at 148.5MHz clock rate
- Can be ported to low cost Altera (Cyclone) or Xilinx (Artix)
Merge Block Details
- Single input data stream, one frame/exposure at a time.
- IP size is 480k gates coupled with 59k bytes of RAM.
- Requires 19G bit/s bandwidth on an external DDR port for frame buffering.
Tone Mapping Block Details
- IP size is 327k gates coupled with 300k bytes of RAM.
- Requires 9.5G bit/s bandwidth on an external DDR port for frame buffering.
Ultra HDR™ Applicable Segments
Data Recognition Cameras
Traffic and Transportation
Surveillance in Transportation Centers
High Contrast Scene Video and Still Capture
Speed and Red Light Detection and Enforcement
License Plate Recognition
Oncoming Headlight Suppression
Oncoming Headlight Suppression
Adaptive Fog Elimination
Veiling Glare / Lens Flare Elimination
Enhanced Dusk and Dawn Video and Still Capture
Wearable Camera Systems
Police Body Cameras
Consumer UAV Cameras
For more information or if you are interested in purchasing an Ultra HDR™ Reference Platform please call Ron Tussy, Director of Business Development or email: firstname.lastname@example.org