Back to results
US 10,834,407 B2Granted
SDR bit depth enhancement via codeword range amplification in a codec with inverse display management
AssigneeDolby Laboratories Licensing Corporation
InventorGuan-Ming Su
Life of this patent
Estimating the term…
Abstract
Novel methods and systems for encoding standard dynamic range video to improve the final quality after converting standard dynamic range video into enhanced dynamic range video are disclosed. A dual layer codec structure that amplifies certain codeword ranges can be used to send enhanced information to the decoder in order to achieve an enhanced (higher bit depth) image signal. The enhanced standard dynamic range signal can then be up-converted to enhanced dynamic range video without banding artifacts in the areas corresponding to those certain codeword ranges.
Claims14
A method to enhance the bit depth of a video signal, said method comprising:
- receiving a first standard dynamic range (SDR) video signal;
- receiving a second SDR video signal derived from the first SDR video signal, the second SDR video signal having a lower bit-depth than the first SDR video signal;
- determining a low codeword threshold SI and/or a high codeword threshold Sh that bounds an entire codeword range of an area of interest within the second SDR video signal, wherein determining the low codeword threshold SI and/or the high codeword threshold Sh comprises detecting and connecting smooth areas in the second SDR video signal, wherein connecting the smooth areas comprises using at least one morphological operator, calculating a histogram of non-zero pixels in the smooth areas, and collecting clusters in the histogram;
- generating a predicted SDR video signal from the second SDR video signal, wherein all codewords within the determined codeword range are set to a constant codeword value based on the determined low codeword threshold SI and/or the determined high codeword threshold Sh;
- generating a mask that selects the codeword values below the low codeword threshold SI and/or above the high codeword threshold Sh;
- encoding the difference between the first SDR video signal and the predicted SDR video signal, masked with the generated mask, into an enhancement layer;
- generating interlayer predictor metadata from the second SDR video signal;
- and
- generating a base layer from the second SDR video signal;
- and
- thus providing the enhancement layer, the interlayer predictor metadata, and the base layer for decoding.
8 more elements
A method for producing an enhanced video signal from a standard dynamic range (SDR) video signal, said method comprising:
- producing a lower bit depth SDR video signal from the SDR video signal;
- producing a predicted SDR video signal by clipping the lower bit depth SDR video signal at a determined pixel location, wherein the step of clipping comprises clipping codewords in the lower bit depth SDR below a low codeword threshold SI to a first constant codeword value and/or clipping codewords in the lower bit depth SDR exceeding a high codeword threshold Sh to a second constant codeword value, wherein determining the determined pixel location comprises detecting and connecting smooth areas in the lower bit depth SDR video signal, wherein connecting the smooth areas comprises using at least one morphological operator, calculating a histogram of non-zero pixels in the smooth areas, and collecting clusters in the histogram;
- forming a residual mask with respect to the determined pixel location;
- applying the residual mask to a residual of the SDR video signal and the predicted SDR video signal, producing a masked residual;
- and
- producing an enhancement layer by quantizing the masked residual;
- and
- producing the enhanced video signal by combining the predicted SDR video signal with the masked residual.
5 more elements
A system for enhancing the bit depth of an input SDR video signal, the system comprising:
- an encoder taking as input the input SDR video signal, and providing an enhancement layer, a base layer, and interlayer predictor metadata;
- and
- a decoder taking as input the enhancement layer, the base layer, and the interlayer predictor metadata, and providing an enhanced bit depth SDR video signal, and applying SDR to EDR conversion of the enhanced bit depth SDR signal, wherein the SDR video signal is derived from a higher bit-depth SDR video signal, and wherein the encoder is configured to:
- determining a low codeword threshold SI and/or a high codeword threshold Sh that bounds an entire codeword range of an area of interest within the input SDR video signal, wherein determining the low codeword threshold SI and/or the high codeword threshold Sh comprises detecting and connecting smooth areas in the SDR video signal, wherein connecting the smooth areas comprises using at least one morphological operator, calculating a histogram of non-zero pixels in the smooth areas, and collecting clusters in the histogram;
- generating a predicted SDR video signal from the input SDR video signal, wherein all codewords within the determined codeword range are set to a constant codeword value based on the determined low codeword threshold SI and/or the determined high codeword threshold Sh;
- generating a mask that selects the codeword values below the low codeword threshold SI and/or above the high codeword threshold Sh;
- encoding the difference between the higher bit-depth SDR video signal and the predicted SDR video signal, masked with the generated mask, into the enhancement layer;
- generating the interlayer predictor metadata from the input SDR video signal;
- and
- generating the base layer from the input SDR video signal.
7 more elements
Dependencies are inferred from the claim text.