| 摘要 |
In a method and an arrangement for generating a depth map, a scene is recorded with plural mutually different video camera focal settings F, scanning stepwise through the depth of the scene, and the depth corresponding to the respective focal setting is allocated to image components respectively recorded in-focus with a focal setting, producing a first depth map. The scene is also recorded repeatedly, respectively with different zoom Z settings, and the depth of the respective image component is calculated from the geometrical changes of image components, producing a second depth map. A combined depth map is formed from the two depth maps, producing an image depth map, e.g. for use in security monitoring or three dimensional (3D) imaging. Depth maps may be combined and averaged using an IIR filter. High confidence ratings may be attributed to locally corresponding image components with similar depths in the first and second depth maps, with low confidence for deviating image features. |
