Depth information aided image coding

In conventional video systems only a flat representation of reality is retained. This makes it hard to identify background patterns in images, which are generally of a minor significance to visual perception, as opposite to the main objects in the scene, which are made context-relevant by focusing through camera optical lens regulations. According to such a visual criterion, the separation of image contents into foreground and background areas is proposed through the analysis of the depth field associated with the paired views of a stereoscopic system; by the provision of a camera-originated signal varying with the shooting focal distance, positions of objects can be evaluated with reference to the plane of focus, thus enabling a priority scheme for bit assignment when digitally coding images for broadcast, transmission over wide-band networks or storage purposes. This process can be straightforwardly applied to 3DTV, as well as extended to the flat compatible images of a conventional single-channel video system suitably provided with an auxiliary stereo camera.<
>