Depth codec for 3D-video recording and streaming applications
This invention provides a codec for depth compression and decompression in 3D-video streaming, recording and visualization applications. PROBLEM: Use of 3D-video imaging is relevant to a vast of data to be recorded into storage media and streamed in real-time through communication media with bandwidth limitations. Reduction of 3D-video data amount can be achieved by using variety of compression methods. Some conventional methods of 2D-video processing are applicable also for 3D-video streams. However, significant reduction of video data amount is relevant to use of lossy processing methods, operation-greedy algorithms, and challenges for visual quality of the processed 3D-videos. SOLUTION: The codec for 3D-video depth information compression and decompression, by employing depth similarities between neighboring rows of a single depth frame, and capabilities of rANS-algorithm to compress sets of values that have highly irregular distribution. Further, the codec comprises a step of dynamic depth interlacing at pixel level, according to the detected movements in different regions of a pixel-depth frame, throughout the sequence of said frames. The codec enables both lossless and lossy compression-decompression techniques which allow to maintain efficiently and predictably the trade-off between visual quality of 3D-view and compression rate, and allows prediction of processing resources.
1. A method of data encoding and decoding in 3D-video processing applications, said applications comprising stages, at least, of:
3D-scene acquisition by one or more 3D-video stream, wherein said 3D-video stream comprises a sequence of 2D-picture frames and the corresponding pixel-depth data frames,
filtering of said 3D-stream,
encoding said 3D-stream,
transmitting said 3D-stream via communication media of limited bandwidth,
recording said 3D-stream into a digital storage,
decoding said 3D-stream,
rendering and visualization of 3D-scene from said 3D-stream,
characterized in that the method comprises steps of pixel-depth frame data encoding and decoding, wherein
depth similarities between the neighboring rows of a single pixel-depth frame are employed to reduce amount of data,
capabilities of rANS-algorithm is employed to compress sets of pixel-depth values which have extremely uneven distribution, from a single pixel-depth frame,
dynamic pixel-depth value interlacing at pixel level is employed to compress sets of pixel-depth values, according to movements detected in different regions of the frame, said movements detected by comparing the neighboring pixel-depth frames.
2. The method according to claim 1 , characterized in that the depth range data encoding and decoding steps have controllable and predictable amount of data processing operations.
3. The method according to claim 1 , characterized in that the depth range data encoding and decoding steps provide controllable and predictable volume of said pixel-depth data.
4. The method according to claim 1 , characterized in that the depth range data encoding and decoding steps provide a controllable visual quality of said depth range data.
5. The method according to claim 1 , characterized in that said encoding and decoding steps allow lossless encoding and decoding of the depth range data.
6. The method according to claim 1 , characterized in that said encoding and decoding steps allow lossy encoding and decoding of the depth range data.
7. The method according to claim 1 , characterized in that said method is applicable for applications with real-time streaming of 3D-videos over a communication media with a limited bandwidth.
8. The method according to claim 1 , characterized in that said method is applicable for applications with recording of 3D-videos into a digital storage with a limited space.
9. The method according to claim 1 , characterized in that said 3D-video encoding and decoding stages comprise at least one embodiment on the basis of the standardized 2D-video encoding-decoding framework, at least, H.264.