Intra-frame flicker reduction in video coding
The embodiments of the present invention provide for methods, devices, and systems adapted to reduce flickers introduced in the encoding process, particularly during the generation of intra-coded pictures. The method provides for the determination of flicker areas within pictures and applying a flicker-control cost function, for example, on those determined flicker areas when selecting an intra-prediction mode.
1. A method of encoding, by an encoder, a video comprising a plurality of received pictures, the method comprising:
generating, for each picture of the plurality of received pictures, an intra-coded picture or an inter-coded picture, each of the intra-coded picture comprising a plurality of picture elements and each of the inter-coded picture comprising a plurality of picture elements;
wherein generating the intra-coded picture comprises:
reconstructing each picture element of the plurality of picture elements; and
if a current reconstructed picture element, of the reconstructed picture elements, is located in a flicker area, performing a flicker-control process adapted to select an intra prediction mode based on a flicker-control cost function comprising:
a product of a weighting factor and a distortion of the current reconstructed picture element; and a product of a complement of the weighting factor and a distortion of the previously reconstructed picture element.
2. The method of claim 1 , wherein the flicker area is based on whether a counter exceeds a defined threshold.
3. The method of clam 2 , wherein the defined threshold is based on a number of pictures in a group of pictures (GOP) associated with the video.
4. The method of claim 1 , further comprising:
if the current reconstructed picture element is not in a flicker area, performing an intra-prediction mode selection process adapted to select an intra prediction mode based on a cost function, the cost function based on the current reconstructed picture element and the original picture element.
5. The method of claim 1 , further comprising the step of:
tracking whether each picture element of the plurality of picture elements of the inter-coded picture is a flicker area based on whether the picture element has a zero motion vector and whether the picture element is a skipped picture element.
6. The method of claim 5 further comprising the step of:
incrementing a counter associated with said each picture element when the picture element is a determined flicker area.
7. The method of claim 5 further comprising the step of:
resetting the counter associated with said each picture element.
8. The method of claim 1 , wherein the method supports H.264 specification.
9. The method of claim 1 , wherein the flicker-control cost function is further represented by the following equation:
cost=α·Distortion i +(1−α)·Distortion i−1 .
10. The method of claim 1 , wherein the distortion of the current reconstructed picture element comprises:
a product of a scaling factor and a sum of absolute difference of the current reconstructed picture element and a current original picture element.
11. The method of claim 10 , wherein the distortion of the current reconstructed picture element is represented by the following equation:
Distortion i =c·SAD ( R i ,O i )
wherein c is a scaling factor to substantially match the Distortion, i−1 with the Distortion i , wherein R i is the current reconstructed picture element, and wherein O i is the at least one picture element.
12. The method of claim 1 , wherein the distortion of the previously reconstructed picture element comprises:
a product of a scaling factor and a sum of absolute difference of the current reconstructed picture element and the previously reconstructed picture element.
13. The method of claim 12 , wherein the distortion of the previously reconstructed picture element is represented by the following equation:
Distortion i−1 =c·SAD ( R i ,R i−1 )
c is a scaling factor to substantially match the Distortion i−1 with the Distortion i , and wherein R i is the current reconstructed picture element, and wherein R i−1 is the previously reconstructed picture element.
14. The method of claim 1 , wherein the flicker-control cost function further comprises a bias, and wherein the bias is a number related to the intra prediction mode and a quantization parameter.
15. The method of claim 1 , wherein the flicker-control cost function is further represented by the following equation:
cost=α·Distrotion i +(1−α)·Distortion i−1 +Bias
wherein the bias is a number related to the intra prediction mode and a quantization parameter.
16. A device adapted to encode a video comprising a plurality of pictures, each picture comprising a plurality of picture elements, the device comprising:
an intra mode selection module adapted to:
select, based on a flicker-control cost function, an intra prediction mode for a current picture element of the plurality of pictures, the flicker-control cost function comprising:
a product of a weighting factor and a distortion of the current reconstructed picture element; and a product of a complement of the weighting factor and a distortion of the previously reconstructed picture element;
an intra prediction module adapted to:
generate a predicted picture element based on the selected intra prediction mode;
a transformation module adapted to transform a residual picture element based on the generated predicted picture element and the current picture element;
a quantization module adapted to perform quantization on the transformed residual picture element;
an entropy encoder module adapted to perform entropy encoding on the quantized transformed residual picture element to generate an encoded bit stream.
17. The device of claim 16 , further comprising:
an inter-prediction module adapted to perform inter prediction;
an inverse quantization module adapted to perform inverse quantization on the residual picture element;
an inverse transformation module adapted to perform inverse transformation on the inverse quantized residual picture element;
a filter adopted to perform de-blocking filter on the inverse transformed inverse quantized residual picture element;
a data store adapted to store the filtered residual picture element.
18. The device of claim 16 , further comprising:
a flicker-tracking module adapted to track flicker areas based on whether a picture element is a flicker area based on whether the picture element has a zero motion vector and is a skipped picture element.
19. The device of claim 16 ,
wherein the intra mode selection module is further adapted to select the intra prediction mode, based on the flicker-control cost function, if the current picture element is a flicker area; and
if the current picture element is not a flicker area, select, based on a cost function, the intra-prediction mode for the current picture element based on the current reconstructed picture element and the original picture element.
20. The device of claim 19 , wherein the flicker area is based on whether a counter exceeds a defined threshold.
21. The device of claim 18 , wherein the flicker-tracking module is further adapted to increment a counter associated with the current picture element.
22. The device of claim 16 , wherein the flicker-control cost function is represented by the following equation:
cost=α·Distrotion i +(1−α)·Distortion i−1 .
23. The device of claim 16 , wherein the distortion of the current reconstructed picture element comprises:
a product of a scaling factor and a sum of absolute difference of the current reconstructed picture element and a current original picture element.
24. The device of claim 23 , wherein the distortion of the current reconstructed picture element is represented by the following equation:
Distortion i =c·SAD ( R i ,O i )
wherein c is a scaling factor to substantially match the Distortion, i−1 with the Distortion, wherein R i is the current reconstructed picture element, and wherein O i is the at least one picture element.
25. The device of claim 16 , wherein the distortion of the previously reconstructed picture element comprises:
a product of a scaling factor and a sum of absolute difference of the current reconstructed picture element and the previously reconstructed picture element.
26. The device of claim 25 , wherein the distortion of the previously reconstructed picture element is represented by the following equation:
Distortion i−1 =c·SAD ( R i ,R i−1 )
wherein c is a scaling factor to substantially match the Distortion i−1 with the Distortion i , and wherein R i is the current reconstructed picture element, and wherein R −1 is the previously reconstructed picture element.
27. The device of claim 16 , wherein the flicker-control cost function further comprises a bias, and wherein the bias is a number related to the intra prediction mode and a quantization parameter.
28. The device of claim 16 , wherein the flicker-control cost function is further represented by the following equation:
cost=α·Distrotion i +(1−α)·Distortion i−1 +Bias
wherein the bias is a number related to the intra prediction mode and a quantization parameter.
29. A system comprising:
an encoding device, operably coupled to a decoder via a transport medium, the encoding device adapted to encode a video comprising a plurality of pictures, each picture comprising a plurality of picture elements, the device comprising:
a flicker-tracking module adapted to track flicker areas based on whether the picture element has a zero motion vector and is a skipped picture element;
an intra mode selection module adapted to:
if a current picture element of the plurality of picture elements is a flicker area, select, based on a flicker-control cost function, an intra prediction mode for the current picture element, the flicker-control cost function comprises: a product of a weighting factor and a distortion of the current reconstructed picture element; and a product of a complement of the weighting factor and a distortion of the previously reconstructed picture element;
if the current picture element is not in a flicker area, select, based on a cost function, the intra prediction mode for the current picture element, the cost function based on the current reconstructed picture element of the current picture and the current picture element;
an intra prediction module adapted to:
generate a predicted picture element based on the selected intra prediction mode;
a transformation module adapted to transform a residual picture element based on the generated predicted picture element and the current picture element;
a quantization module adapted to perform quantization on the transformed residual picture element;
an entropy encoder module adapted to perform entropy encoding on the quantized transformed residual picture element to generate an encoded bit stream;
a decoding device adapted to decode the encoded bit stream.
30. A method of encoding, by an encoder, a video comprising a plurality of pictures, the method comprising the steps of:
receiving the plurality of pictures; and
for each picture of the plurality of pictures, generating an intra-coded picture or an inter-coded picture, each of the intra-coded picture comprising a plurality of picture elements and each of the inter-coded picture comprising a plurality of picture elements;
wherein the step of-generating the intra-coded picture comprises reconstructing each picture element of the plurality of picture elements; and
if a current reconstructed picture element, of the reconstructed picture elements, is located in a flicker area, performing a flicker-control process adapted to select an intra prediction mode based on a flicker-control cost function, the flicker-control cost function based on the reconstructed picture element of a previously reconstructed picture, the current reconstructed picture element, and an original picture element associated with the current reconstructed picture element, and wherein the reconstructed picture element of the previously reconstructed picture, the current reconstructed picture element, and the original picture element are collocated in the same location; and
wherein the flicker-control cost function is represented by the following equation:
cost=α·Distrotion i +(1−α)·Distortion i−1 +Bias
where α is a weighting factor between 0 and 1,
Bias is selected from at least one of the following:
Bias=λ*Rate, where λ is a Lagrange multiplier and Rate is a coding bit rate; and
Bias= number related to the intra prediction mode and a quantization parameter;
where Distortion i−1 of the previously reconstructed picture element is represented by the following equation: Distortion i−1 =c·SAD ( R i ,R i−1 )
where R i is the current reconstructed picture element and R i−1 is the previously reconstructed picture element;
c is a scaling factor to substantially match the Distortion i−1 with the Distortion i ; and
where Distortion i of the current reconstructed picture element is represented by the following equation:
Distortion i =c·SAD ( R i ,O i )
where O i is the at least one picture element.
31. A device adapted to encode a video comprising a plurality of pictures, each picture comprising a plurality of picture elements, the device comprising:
an intra mode selection module adapted to:
select, based on a flicker-control cost function, an intra prediction mode for a current picture element of the plurality of pictures, the flicker-control cost function based on a reconstructed picture element of a previously reconstructed picture, a current reconstructed picture element of the current picture element, and the current picture element, and wherein the reconstructed picture element of the previously reconstructed picture, the current reconstructed picture element, and the current picture element are collocated in the same location;
an intra prediction module adapted to:
generate a predicted picture element based on the selected intra prediction mode;
a transformation module adapted to transform a residual picture element based on the generated predicted picture element and the current picture element;
a quantization module adapted to perform quantization on the transformed residual picture element;
an entropy encoder module adapted to perform entropy encoding on the quantized transformed residual picture element to generate an encoded bit stream; and
wherein the flicker-control cost function is represented by the following equation:
cost=α·Distrotion i +(1−α)·Distortion i−1 +Bias
where α is a weighting factor between 0 and 1,
Bias is selected from at least one of the following:
Bias=λ*Rate, where λ is a Lagrange multiplier and Rate is a coding bit rate; and
Bias=number related to the intra prediction mode and a quantization parameter;
where Distortion i−1 of the previously reconstructed picture element is represented by the following equation:
Distortion i−1 =c·SAD ( R i ,R i−1 )
where R i is the current reconstructed picture element and R i−1 is the previously reconstructed picture element;
c is a scaling factor to substantially match the Distortion i−1 with the Distortion i ; and
where Distortion i of the current reconstructed picture element is represented by the following equation:
Distortion i =c·SAD ( R i ,O i )
where O i is the current picture element.