Djupbild

Djupbild, or depth image, is a two-dimensional array where each pixel stores the distance from an imaging sensor to the scene point projected to that pixel. The depth information may come from active depth sensors such as time-of-flight cameras or structured-light systems, from laser scanners, or be inferred from stereo vision by matching image features across viewpoints. In many modern workflows, a depth image is paired with a color image (RGB-D), allowing simultaneous access to color and depth for the same scene.

Depth data are typically stored as 16-bit integers representing millimeters or as 32-bit floating-point values representing meters. Some formats use disparity values in pixels for stereo systems. Invalid or occluded pixels may be marked with special values such as zero, NaN, or negative depth, depending on the format. For practical use, depth images are often rectified to align with color imagery, requiring intrinsic and extrinsic calibration to relate depth to the camera’s coordinate system.

Applications of djupbild include 3D reconstruction, object recognition, robotics, autonomous navigation, and augmented or mixed reality. They enable surface modeling, obstacle detection, hand and gesture tracking, and general scene understanding. Limitations include sensor noise, especially at longer ranges, reduced accuracy on reflective, transparent, or absorbing surfaces, and challenges in low-texture or low-light conditions. The term is widely used in computer vision and robotics, with standards and representations varying by sensor and software framework.