Focus-aware compression and image quality metric for 3D displays

3D displays are capable of immersive 3D content presentation without glasses and headsets. The displays project different views depending on the user’s viewing angle. Multiple views are visible to the user at once to simulate the 3D perception. The visual blending of the simultaneously projected views creates out-of-focus areas in the scene. The scene can be displayed with only one focusing distance, where the objects appear sharp. This paper utilizes this effect in data compression. First, several metrics are proposed for automatic visual quality assessment of the results on the 3D display. Then, automatic detection of the focusing distance in the scene, based on the input views, is proposed. Based on this detection, the high spatial frequencies in out-of-focus areas can be eliminated using the depth-of-field effect, or such areas can be compressed with higher compression ratio, depending on the use case. The paper compares the proposals. A user study is conducted to obtain human evaluation of the results. The results obtained identified the optimal visual quality metric for the 3D display. The proposed compression proved to be beneficial for 3D displays and capable of reaching a higher compression ratio than standard methods without a perceivable quality loss.