http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-103810473-B
Outgoing Links
Predicate | Object |
---|---|
classificationIPCInventive | http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06K9-00 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T7-194 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06K9-66 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06T7-11 |
filingDate | 2014-01-23-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
grantDate | 2016-09-07-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
publicationDate | 2016-09-07-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
publicationNumber | CN-103810473-B |
titleOfInvention | A kind of target identification method of human object based on HMM |
abstract | The invention discloses the target identification method of a kind of human object based on HMM, if human object foreground image and target body object foreground image are divided into stem portion by it, then the HMM after the training obtained according to the color characteristics of mean sequence of the human object foreground image that the extracts color characteristic corresponding with target body object foreground image and the poor characteristic sequence of color standard and the characteristic sequence by each target body object, calculate human object foreground image relative to the posterior probability of hidden Markov model after the training of each target body object, thus realize the identification to human object image;The dimension of the feature of human object foreground image and the feature of target body object foreground image is relatively low, and the HMM obtained make use of the feature of multiple image of same target body object, the change of human body angle and the change of light there is certain robustness, and recognition effect is good, amount of calculation is little, complexity is low. |
priorityDate | 2014-01-23-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
type | http://data.epo.org/linked-data/def/patent/Publication |
Incoming Links
Total number of triples: 16.