SceneScan: Real-Time Depth-Sensing Through Stereo Vision

SceneScan Stereo Vision Sensor

SceneScan is Nerian's latest 3D depth sensing solution, and the successor to our popular SP1 stereo vision sensor. SceneScan relies on stereo vision and hardware-accelerated image processing to gain a 3D understanding of an observed area. Unlike conventional depth sensing methods, stereo vision is a passive technology that does not require the emittance of visible or invisible light. This is why SceneScan provides accurate 3D depth measurements even in challenging situations, such as in bright daylight, far-range measurements, multiple overlapping sensors, or even under water.

SceneScan connects to our Karmin2 stereo camera, or to two conventional USB3 Vision industrial cameras. The camera images are processed in real-time using a powerful FPGA and our state-of-the-art stereo matching algorithm. The processing result is a sub-pixel accurate disparity map (an inverse depth map), which is streamed through gigabit ethernet to a connected computer or embedded system. Using our open-source and cross-platform API, this disparity map can be converted into a dense 3D point cloud. Through a built-in auto-recalibration, long-term operability of the system is ensured, even if the cameras encounter mechanical stresses.

SceneScan with attached stereo camera

By choosing a suitable camera setup, very different measurement ranges can be covered. The user can further choose between a high frame rate, or a high image or depth resolution, with frame rates up to 100 fps and resolutions up to 3 megapixels being possible. This renders SceneScan the most flexible 3D depth sensing solution on the market. Nerian's support staff will assist with choosing the optimal configuration for each customer project.

The FPGA design that drives SceneScan can also be licensed as an IP core.

Contact Sales

Depth Measurement Examples

Example left camera image Example depth measurements
Close
Example left camera image
Example left camera image.
Close
Example depth measurements
Depth measurements (disparity map) computed by SceneScan.
Example left camera image Example depth measurements
Close
Example left camera image
Example left camera image. Taken from the Middlebury 2006 dataset.
Close
Example depth measurements
Depth measurements (disparity map) computed by SceneScan.
Example left camera image Example depth measurements
Close
Example left camera image
Example left camera image.
Close
Example depth measurements
Depth measurements (disparity map) computed by SceneScan.

Models

Two different models are available: SceneScan and SceneScan Pro. Both models provide identical features but differ in their processing capabilities. SceneScan Pro achieves the highest possible processing performance, while SceneScan has been cost optimized for less demanding applications.

SceneScan
SceneScan Pro
Max. image resolution: 800 × 800 pixels 1856 × 1856 pixels
Max. disparity: 128 pixels 256 pixels
Max. frame rate: 45 fps 100 fps

The maximum frame rate depends on the selected image resolution and depth resolution (the disparity range). We recommend using one of the configurations listed below, but other configurations are possible to meet specific application requirements.

Model Disparity Image Resolution
Range 640 × 480 800 × 592 1280 × 960 1600 × 1200
SceneScan 64 pixels 45 fps 30 fps n/a n/a
128 pixels 30 fps 20 fps n/a n/a
SceneScan Pro 128 pixels 100 fps 65 fps 24 fps 15 fps
256 pixels 70 fps 45 fps 15 fps 10 fps

Videos

SceneScan Specifications

Hardware Details
Power supply: 11 - 14 V DC
Power consumption: Less than 10 W without supplying camera power
Up to 20 W with supplying camera power
Dimensions: 104.5 × 105.5 × 45 mm without mounting brackets
104.5 × 130 × 45 mm with mounting brackets
Weight: 400 g
I/O: 2× USB 3.0 host, gigabit ethernet, GPIO
Conformity: CE, FCC, RoHS
Stereo Matching
Stereo algorithm: Variation of Semi-Global Matching (SGM)
Max. image resolution:
SceneScan:
SceneScan Pro: 
800 × 800 pixels
1856 × 1856 pixels
Supported pixel formats: Mono8, Mono12, Mono12p, Mono12Packed
Disparity range:
SceneScan:
SceneScan Pro: 
64 to 128 pixels
96 to 256 pixels (32 pixel increment)
Max. frame rate:
SceneScan:
SceneScan Pro: 
30 fps
100 fps
Sub-pixel resolution: 4 bits (1/16 pixel)
Post-processing: Erroneous disparity removal, noise removal, speckle filtering
Image Rectification
Horizontal displacement: -39 to +39 pixels
Vertical displacement: -39 to +39 pixels
Interpolation: Bilinear

Supported Cameras

SceneScan supports the USB3 Vision standard. Hence it can connect to any USB3 Vision compatible camera with a grayscale image sensor. In order to guarantee compatibility, however, we recommend using our own Karmin2 stereo camera or cameras by one of the following certified manufacturers: FLIR (formerly Point Grey), Basler, SVS-Vistek, The Imaging Source.

Supported Platforms and Standards

ROS HALCON MATLAB MATLAB GenICam GenTL

Downloads

Resources

See also