Method and system for encoding video data
View Patent ↗Described herein is a method and system for encoding video data. The design comprises a two pass encoding system with bi-direction control to and from a classification engine. In the first pass coder, a future picture is encoded to produce a set of parameters that characterize the future picture. In the second pass coder a current picture is encoded to produce a video output. The classification engine is the center of perceptual enhancement. The classification engine interprets the set of parameters from the first pass coder and classifies the current picture.
1. A method for video encoding a current block within a picture, comprising:
encoding a picture to produce a set of parameters that characterize the picture;
classifying the picture based on the set of parameters to produce a first picture classification; and
encoding the picture according to the first picture classification to produce a video output,
wherein the picture comprises a plurality of macroblocks,
wherein the set of parameters that characterize the picture results from the encoding of a future macroblock in the picture, and
wherein the video output results from the encoding of a current macroblock in the picture.
2. The method of claim 1 , wherein encoding the picture to produce the set of parameters that characterize the picture further comprises:
filtering the picture to produce a preprocessed picture;
measuring spatial activity in the preprocessed picture; and
selecting one or more reference pictures that predict motion relative to the preprocessed picture.
3. The method of claim 2 , wherein a reference picture in the one or more reference pictures is generated from a set of prediction residuals.
4. The method of claim 1 , wherein said method further comprises:
determining a rate control parameter based on the video output;
classifying the video output based on the rate control parameter to produce a second picture classification; and
encoding another picture according to the first picture classification and the second picture classification.
5. A method for video encoding a current block within a picture, comprising:
encoding a picture to produce a set of parameters that characterize the picture;
classifying the picture based on the set of parameters to produce a first picture classification; and
encoding the picture according to the first picture classification to produce a video output,
wherein encoding the picture according to the first picture classification to produce the video output further comprises:
predicting spatial activity in the picture to produce a first prediction;
predicting motion in the picture to produce a second prediction; and
selecting between the first prediction and the second prediction, based on the first picture classification.
6. A system for video encoding, said system comprising:
a coarse motion estimator to encode a picture, to produce a set of parameters that characterize the picture;
a classification engine to classify the picture based on the set of parameters, to produce a first picture classification;
a fine motion predictor to encode the picture according to the first picture classification, to produce a video output;
a spatial predictor to predict spatial activity in the picture, to produce a first prediction;
a fine motion predictor to predict motion in the picture, to produce a second prediction; and
a mode selector to select between the first prediction and the second prediction, based on the first picture classification.
7. The system of claim 6 , wherein the picture is a macroblock.
8. The system of claim 6 , wherein the picture comprises a plurality of macroblocks.
9. The system of claim 6 , wherein said system further comprises:
a filter to filter the picture, to produce a preprocessed picture; and
a statistics collector to measure spatial activity in the preprocessed picture.
10. The system of claim 9 , wherein the coarse motion estimator further selects one or more reference pictures that predict motion relative to the preprocessed picture.
11. The system of claim 6 , wherein said system further comprises:
a rate controller to determine a rate control parameter based on the video output.
12. The system of claim 11 , wherein the classification engine classifies the video output based on the rate control parameter to produce a second picture classification.
13. The system of claim 12 , wherein said system further comprises:
encoding another picture according to the first picture classification and the second picture classification.
14. The system of claim 6 , wherein the system encodes original lossless previous pictures with lossy encoding and wherein the coarse motion estimator motion estimates the picture from the original lossless previous pictures.
15. The system of claim 14 , wherein the fine motion estimator motion estimates from reconstructed lossy encoded previous pictures.
16. The system of claim 14 , wherein the coarse motion estimation uses only integer pixel offsets.
17. The system of claim 14 , wherein the fine motion estimator uses fractional pixel offsets.