Image processing method and image processing apparatus
View Patent ↗An image processing method with which occurrence of an error is suppressed is an image processing method for processing an image signal generated by encoding a plurality of pictures. The method includes: obtaining, from the image signal, a parameter indicating restriction on a reference relation between at least one of the plurality of pictures and an other one of the plurality of pictures in the encoding; and performing restriction alleviation processing for alleviating the restriction on the reference relation indicated by the parameter.
1. An image processing method for processing an image signal generated by encoding a plurality of pictures, the image processing method comprising:
obtaining a parameter from the image signal, the parameter indicating restriction on a reference relation between at least one of the plurality of pictures and an other one of the plurality of pictures in the encoding; and
performing restriction alleviation processing for alleviating the restriction on the reference relation indicated by the parameter, wherein
the image signal includes a plurality of network abstraction layer (NAL) units as signal units, and each of pictures corresponding to a different one of the plurality of NAL units is classified into any one of a plurality of layers, and encoded without referring to a picture which belongs to a layer of a higher level than a layer to which the picture belongs,
in the performing,
a target NAL unit of the signal units is a target for the restriction alleviation processing, such that even when a first NAL unit type included in the image signal is incorrect, a second NAL unit type, included in the image signal, in which restriction on a reference relation is alleviated, is applied to the target unit of the signal units, and
in the obtaining,
a NAL unit type included in the target NAL unit to be processed, which is any one of the plurality of NAL units, is obtained as the parameter, the NAL unit type being included in a NAL unit corresponding to a target picture to be processed among the plurality of pictures, and defining forbidding a reference relation in which the target picture refers to a specific picture which belongs to a same layer as the target picture,
the image processing method further comprising: in the performing,
when the NAL unit type is the first NAL unit type, the second NAL unit type, which is not included in the target NAL unit, is applied to the target NAL unit instead of the first NAL unit type, the second NAL unit type defining alleviated restriction which is less strict than restriction on a reference relation defined by the first NAL unit type,
when the first NAL unit type defines that a picture corresponding to the target NAL unit is: a random access skipped leading (RASL) picture that (i) is a leading picture which is subsequent in decoding order to a random access point (RAP) picture that is a starting point of the random access reproduction and which precedes the RAP picture in display order and (ii) has a reference relation in which decoding is impossible when the random access reproduction is performed,
the second NAL unit type, which is not included in the target NAL unit, is applied to the target NAL unit instead of the first NAL unit type, and the picture is decoded, the second NAL unit type defining that the picture is a random access decodable leading (RADL) picture that is the leading picture and has a reference relation in which decoding is possible even when the random access reproduction is performed, and
when the picture is decoded,
reference picture information corresponding to the picture is decoded,
when a reference picture identified by the decoded reference picture information is stored in a buffer, the picture is decoded by referring to the reference picture, and
when the reference picture is not stored in the buffer, an alternative picture, which is not stored in the buffer, is separately generated, and the picture is decoded by referring to the alternative picture as the reference picture.
2. The image processing method according to claim 1 , further comprising:
wherein in the performing,
when the first NAL unit type defines forbidding a reference relation in which a subsequent picture subsequent in decoding order to a broken link access (BLA) picture which is a starting point of random access reproduction and corresponds to the target NAL unit refers, as a leading picture which precedes the BLA picture in display order, to an other picture,
the second NAL unit type which defines permitting the reference relation is applied to the target NAL unit, instead of the first NAL unit type.
3. The image processing method according to claim 2 , further comprising:
wherein in the performing,
when the first NAL unit type defines forbidding a reference relation in which the subsequent picture refers, as the leading picture, to a picture which precedes the BLA picture in decoding order,
the second NAL unit type which defines permitting the reference relation is applied to the target NAL unit, instead of the first NAL unit type.
4. The image processing method according to claim 1 , further comprising:
wherein in the performing,
when the first NAL unit type defines forbidding a reference relation in which a subsequent picture subsequent in decoding order to an instantaneous decoding refresh (IDR) picture which is a starting point of random access reproduction and corresponds to the target NAL unit refers, as a leading picture which precedes the IDR picture in display order, to an other picture,
the second NAL unit type which defines permitting the reference relation is applied to the target NAL unit, instead of the first NAL unit type.
5. The image processing method according to claim 1 , further comprising:
wherein in the performing,
a NAL unit including the second NAL unit type is generated by replacing the first NAL unit type included in the target NAL unit with the second NAL unit type.
6. An image processing apparatus which processes an image signal generated by encoding a plurality of pictures, the image processing apparatus comprising:
control circuitry; and
storage accessible from the circuitry,
wherein the control circuitry performs, using the storage,
obtaining a parameter from the image signal, the parameter indicating restriction on a reference relation between at least one of the plurality of pictures and an other one of the plurality of pictures in the encoding; and
performing restriction alleviation processing for alleviating the restriction indicated by the parameter, wherein
the image signal includes a plurality of network abstraction layer (NAL) units as signal units, and each of pictures corresponding to a different one of the plurality of NAL units is classified into any one of a plurality of layers, and encoded without referring to a picture which belongs to a layer of a higher level than a layer to which the picture belongs,
in the performing,
a target NAL unit of the signal units is a target for the restriction alleviation processing, such that even when a first NAL unit type included in the image signal is incorrect, a second NAL unit type, included in the image signal, in which restriction on a reference relation is alleviated, is applied to the target unit of the signal units, and
in the obtaining,
a NAL unit type included in the target NAL unit to be processed, which is any one of the plurality of NAL units, is obtained as the parameter, the NAL unit type being included in a NAL unit corresponding to a target picture to be processed among the plurality of pictures, and defining forbidding a reference relation in which the target picture refers to a specific picture which belongs to a same layer as the target picture,
the image processing method further comprising: in the performing,
when the NAL unit type is the first NAL unit type, the second NAL unit type, which is not included in the target NAL unit, is applied to the target NAL unit instead of the first NAL unit type, the second NAL unit type defining alleviated restriction which is less strict than restriction on a reference relation defined by the first NAL unit type,
when the first NAL unit type defines that a picture corresponding to the target NAL unit is: a random access skipped leading (RASL) picture that (i) is a leading picture which is subsequent in decoding order to a random access point (RAP) picture that is a starting point of the random access reproduction and which precedes the RAP picture in display order and (ii) has a reference relation in which decoding is impossible when the random access reproduction is performed,
the second NAL unit type, which is not included in the target NAL unit, is applied to the target NAL unit instead of the first NAL unit type, and the picture is decoded, the second NAL unit type defining that the picture is a random access decodable leading (RADL) picture that is the leading picture and has a reference relation in which decoding is possible even when the random access reproduction is performed, and
when the picture is decoded,
reference picture information corresponding to the picture is decoded,
when a reference picture identified by the decoded reference picture information is stored in a buffer, the picture is decoded by referring to the reference picture, and
when the reference picture is not stored in the buffer, an alternative picture, which is not stored in the buffer, is separately generated, and the picture is decoded by referring to the alternative picture as the reference picture.
7. The image processing method according to claim 1 , wherein
the alternative picture is a gray image having zero chromaticity.