Intel RealSense 3D R200 Rear Camera

Original author: Colleen Culbertson
  • Transfer

The Intel RealSense 3D series will soon be replenished with a new model: the R200 rear camera. For 64-bit Windows 10 and Windows 8.1, the beta version of Intel RealSense R3 SDK (version 5.0.3) is already available. Most likely, by the end of this year will be released SDK for Android 5 version.
The R200 camera with a size of about 102 x 9.5 x 3.8 mm will appear in some models of tablets and transformer laptops for the 2015 holiday season. The peripheral version of the R200 camera measuring 130 x 20 x 7 mm with the ability to connect via USB3 will be available here .

The R200 consists of three cameras: one that shoots in RGB (color image) mode, and a double stereoscopic IR camera that provides image depth. Thanks to the laser projector, the camera has a three-dimensional scanning function, which allows you to perform scene analysis and provides new photo processing capabilities. The camera range in the room is approximately 0.5–3.5 m, and on the street - up to 10 m. Note. The range depends on the module used and the lighting. To learn more about the operating ranges when using various algorithms for the F200 and R200 cameras, check out the online documentation.

The R200 camera, unlike the Intel® RealSense ™ F200 front-facing model, is directed away from the user and with the help of the RealSense SDK allows solving the following tasks.
  1. Receive three-dimensional images of various objects, change them, send them to other users and print using 3D printers.
  2. Extend the capabilities of photography. Change lighting, focus and background using 3D filters (delete or replace parts of the image).
  3. Add virtual objects to 3D real-world images using the Scene Perception function (scene analysis). At the same time, the R200 camera recognizes surfaces, objects and movements and creates scenes taking into account its own position.

These use cases are determined by the two functional areas of the R200 cameras.
  • Tracking / location. Real-time assessment of the position and orientation of the camera based on depth, color, and gyro stabilizer data.
  • Three-dimensional recreation of volume and surfaces. Digital three-dimensional representation of the scene captured by the camera in real time.

Certain features of the Intel RealSense Windows SDK for the F200 will be available in the near future, as the R3 SDK for this model is still in beta beta for the R200. The alpha version of the R200 SDK (R2 SDK release for the F200 camera) features face recognition and positioning. The maximum distance at which face recognition works is 2.5 m, the positioning is 1.5 m. In the beta version of R3, tracking of blobs and recognition of reference points on faces from a distance of up to 1 m is also available. For Android devices will be released separate SDK.

The Windows version of the RealSense SDK is the same for the F200 and R200. It contains documentation and samples for both models, but a separate DCM and firmware package is required to use the R200. This is because the R200 supports DCM 2.x series and the F200 supports DCM 1.x series.
One of the main advantages of the R200 is its improved scanning function and a new method for measuring image depth. The R200 has both stereo cameras and an RGB camera. Since the camera is less dependent on the IR module, it is suitable for outdoor shooting. Images of people come from a color camera, and data for algorithmic processing comes from two 3D cameras. In addition, if a gyrostabilizer is provided in the system, the SDK can adjust the effect of gravity on objects that are added to the scene.

Thanks to stereoscopic photography, 3D parameters are calculated using the pixel shift method between two “spaced” cameras. Depth is measured using triangulation. Please note that in this case, parallel planes are taken into account, not the absolute distance to the camera. The SDK provides a special component for this. When initializing, it is important to properly adjust the camera to the horizon line to avoid errors in the orientation of the scan target.

In the R200 model, camera tracking in three-dimensional space is performed with six degrees of freedom: three degrees when moving forward / backward / up / down / left / right and three degrees when yawing / pitch / roll.
To fully realize the capabilities of three-dimensional scanning, consider the following recommendations.
  • Use a 2 m3 field of view. and a resolution of 5000 pixels (640 x 480). To ensure correct tracking, use the following table.
Min distance The size of a rectangular object
30 cm 4.5 cm x 3.5 cm
100 cm 16 cm x 11 cm
180 cm 28 cm x 21 cm
  • Do not block the object by more than 20%.
  • Try not to move the subject (you can change the position of the camera).
  • Set the frame rate to 30 or 60 frames per second (see table below). Set high frame rates to reduce inter-frame motion.
  • Do not use flat surfaces without a clear structure. Due to the uneven light image, the IR projector adds texture to the scene, after which it applies IR filters to the received data. In addition, data from an RGB camera is added to stereoscopic depth calculations.
  • You should not move the camera with quick movements - remember that when shooting at a speed of 60 frames per second, 18 million calculation operations per second are performed.
  • Wait for the camera to initialize (see the lower left corner of the screen). Set the camera on the target between the green lines.

A color camera with a constant focal length and an aspect ratio of 16: 3 allows you to shoot RGBA images (32 bits) with a resolution of 1080p at a speed of 60 frames per second. The field of view of an RGB camera is slightly larger than that of dual cameras, but it is not intended for separate use.
Dual depth cameras have a constant focal length, 4: 3 aspect ratio and a field of view of 70 x 59 x 46 degrees.
The IR module is a Class 1 laser with a range of 850 nm.
Available resolutions:
60 frames per second, depth 320 x 240, color 640 x 480
60 frames per second, depth 480 x 360, color 320 x 240 or 640 x 480
30 frames per second, depth 320 x 240, color 640 x 480, 1280 x 720 or 1920 x 1080
30 frames per second, depth 480 x 360, color 320 x 240, 640 x 480, 1280 x 720 or 1920 x 1080

The RealSense SDK interface allows you to project the image depth into color and vice versa.

The energy consumption level of the R200 camera: from 0 to 100 mW in the idle state and from 1.0 to 1.6 W during operation (depending on the modules). It provides power-saving features, such as working with USB3 data packets.
Unity support for the R200 is embedded in the SDK. Note. Since Unity version 4.x is a 32-bit program, and Unity 5.x is 64-bit, add the appropriate libraries to the project. The personal (unprofessional) version of Unity can be used when signed libraries are added to the project.

Learn more about Intel® RealSense Technology atIntel website

Also popular now: