http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-113012254-B
Outgoing Links
Predicate | Object |
---|---|
classificationCPCAdditional | http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T2207-20221 |
classificationCPCInventive | http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T11-001 http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06T5-50 http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06N3-04 http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06N3-08 |
classificationIPCInventive | http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T11-00 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T5-50 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06N3-08 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06N3-04 |
filingDate | 2021-04-20-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
grantDate | 2022-04-26-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
publicationDate | 2022-04-26-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
publicationNumber | CN-113012254-B |
titleOfInvention | Underwater image synthesis method based on pixel-level self-supervision training |
abstract | The invention belongs to the field of computer vision, and particularly relates to an underwater image synthesis method based on pixel-level self-supervision training, aiming at solving the problem that the prior art cannot realize stable high-quality underwater image synthesis. The invention comprises the following steps: acquiring a natural light image to be synthesized, corresponding image depth information, image depth information and a randomly generated noise vector; attenuating the natural light image to be synthesized through an attenuation module of the underwater image synthesis model, scattering image depth information through a scattering module of the underwater image synthesis model, and connecting the attenuation image and the scattering information; and halo addition is carried out through the inverse operation of the virtual light correction module of the underwater image synthesis model, so as to obtain a synthetic underwater image corresponding to the natural light image to be synthesized. The method realizes stable high-quality underwater image synthesis, and has high model training efficiency and low training data acquisition difficulty. |
priorityDate | 2021-04-20-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
type | http://data.epo.org/linked-data/def/patent/Publication |
Incoming Links
Predicate | Subject |
---|---|
isDiscussedBy | http://rdf.ncbi.nlm.nih.gov/pubchem/substance/SID419512635 http://rdf.ncbi.nlm.nih.gov/pubchem/compound/CID962 |
Total number of triples: 19.