Skip to content
Search

Range imaging

lastmod 2026-07-19created 2026-03-092 edits

Range imaging is the name for a collection of techniques used to produce a 2D image showing the distance to points in a scene from a specific point, normally associated with some type of sensor device.

The resulting range image has pixel values that correspond to the distance. If the sensor used to produce the range image is properly calibrated, the pixel values can be given directly in physical units, such as meters.

Depth Estimation

Depth estimation is a technique that uses visual information such as images or video to measure how far objects or points in a scene are from the observation point (distance). It plays an important role in fields that need to understand space, such as robotics, augmented reality (AR), and autonomous driving.

Depth Camera

Broadly divided into active and passive methods, but today many hybrid forms combining the two are also used.

Active methods

  • Principle:
    • Measures depth by artificially projecting light (IR or EV) or a signal from outside.
  • Technologies:
    • Structured Light
    • Time of Flight (ToF)
    • LIght Detection And Ranges (LIDAR).
  • Pros:
    • Works well even in dark environments.
    • Provides reliable depth information even on surfaces lacking texture.
  • Cons:
    • Sensitive to external light (especially strong sunlight), which can degrade performance.
      • Natural light contains light of many frequencies, so when using IR, strong IR in natural light becomes a problem.
      • When other strong light sources (natural light, ambient light) are present, the light or signal for depth measurement must be emitted more strongly.
    • Requires additional hardware (projectors, lasers, etc.), which can increase cost.

Passive methods

  • Principle:
    • Extracts depth using natural or ambient light in the environment.
    • Captures an object from different angles using two or more cameras,
    • and computes depth by analyzing the disparity between the two images.
  • Technologies:
    • Stereo Vision
    • Multi-view Stereo
  • Pros:
    • No additional light source or projector is needed, so hardware is simple and relatively cheap.
    • Works well even outdoors.
  • Cons:
    • Performance degrades in dark environments.
    • On surfaces lacking texture, finding correspondences is hard, which can lower accuracy.

Main Depth Camera Methods

Current major depth cameras are divided as follows.

ToF (Time of Flight) camera:

  • A ToF camera emits light (or ultrasound) and calculates distance by measuring the time it takes for that light (or ultrasound) to reflect off an object and return.
  • This technology can obtain depth data quickly, so it is used in various application fields.
  • It can use IR, ultrasound, LASER, etc.; among these, the LASER case is called LIDAR.
    • When using IR, it is vulnerable to strong ambient light.
  • Also, the ToF method generally produces a dense depth map, while LIDAR outputs results as a sparse point cloud.
  • Representative products
    • Azure Kinect: integrates with the vision service of Azuer Cloud. IR-based (940nm) ToF.
    • ToF sensors in some smartphones

Structured Light Camera:

  • This technology projects a predefined pattern of light (usually dots, lines, grid, etc.) onto an object, captures the pattern with a camera, and extracts depth: it is one of the following two main approaches.
    • Computes depth from the image of the pattern formed on the object, using one or more cameras.
      • 2024.07.18 - (Programming/DIP) - (CV) Depth Camera (or Active Sensor): Structured Light
      • With this approach, depth can be extracted with just one camera.
        • Put simply, it is passive stereo with one camera replaced by a projector.
      • Major commercial products:
        • Microsoft Kinect v1: the early model, using one RGB camera and an IR projector.
        • Intel RealSense Depth Camera D415: one of Intel's RealSense series, using one camera and a structured light source.
        • Apple TrueDepth: consists of an infrared (IR) projector, an infrared (IR) camera, and a regular camera. The IR projector projects a pattern of thousands of dots onto the face, and the IR camera captures how the dot pattern is distorted after being projected onto the face.It then analyzes the positions of the distorted dot pattern to compute the face's depth information.
    • Captures the pattern projected by the projector with two cameras and computes depth using stereo vision.
      • Also called Active Stereo Vision.
      • Requires two or more cameras.
      • Major commercial products:
        • Microsoft Kinect v2: the successor to Kinect v1, using two IR cameras and one IR projector to extract more precise depth.
        • Intel RealSense Depth Camera D435: uses two IR cameras and a structured light source.

Stereo Vision Camera:

  • Captures an object from different poses using two or more cameras,
  • and extracts depth by comparing these images.
  • Similar to the human visual system.
  • Mainly used in passive methods, but also used in active methods.

LIDAR (Light Detection and Ranging):

  • Uses light (laser): reflects a laser off an object and calculates distance by measuring the reflection time.
  • In some respects very similar to ToF, but LIDAR is used to obtain accurate depth at longer distances (compared with other depth cameras).
    • It obtains more accurate depth than other technologies,
    • but differs in that it produces a sparse point cloud (other technologies provide a dense depth map)
  • When radio waves (λ from cm to m) are used, it is called RADAR (RAdio Detection And Ranging).
    • For LIDAR, the wavelength is on the order of hundreds of nm to a few micrometers.
    • RADAR measures long distances of several km or more but has lower precision than LIDAR.
    • LIDAR can measure distances of hundreds of meters with mm-to-cm precision.
  • Differs in that data is obtained as a point cloud.
  • LIDAR provides very high accuracy and is widely used in autonomous vehicles, terrain surveying, and so on.
    • Spatial resolution tends to be low.
  • Major commercial products:
    • Velodyne Puck: measures 300,000 points per second. Depth range of 100m.

Multi-baseline Stereo:

  • A method that uses multiple cameras to compute depth more accurately.
  • Can produce a more precise depth map based on images from many cameras.

See also

Favorite site

author Zer0 sitemap