Method, apparatus or system for image processing
AssigneeNokia Technologies Oy
InventorsJian Ma, Ye Tian, Xiaogang Yang, Wendong Wang
Life of this patent
Estimating the term…
Abstract
A method, apparatus, system for image processing is provided that creates an object note for a picture. The method, apparatus or system includes: automatically obtaining picture description information from a picture source, wherein the picture description information at least partly have been formed automatically; obtaining annotation information from at least two object sources, wherein the two object sources are different from the picture sources, automatically fusing the said annotation information from the two object sources to form fused annotation information, and attaching said fused annotation information to the picture to create the object note for the picture.
Claims20
A method comprising:
- obtaining picture description information from a picture source, wherein the picture description information having been formed by adding at least one of a time, a location, or user information to the picture description information,
- obtaining annotation information from at least two object sources, wherein the at least two object sources are different from said picture source,
- fusing said annotation information from said at least two object sources to form fused annotation information, wherein fusing the annotation information includes:
- selecting content for annotation from the annotation information from the at least two object sources by parsing the annotation information into a text format and selecting relevant text,
- assigning a score value to each sentence from the selected relevant text based upon relevancy and redundancy in comparison with other sentences from the selected relevant text;
- filtering, using assigned score values for the selected content, to reduce irrelevant or redundant information by removing stigma words and repetitive information, wherein repetitive information comprises selected content that overlaps in subject matter, and
- enhancing the cohesion or coherence of the content by at least arranging the text with topically related themes to reduce non-fluency, and
- attaching said fused annotation information to a picture to create an object note for said picture.
An apparatus comprising at least one processor, and memory including computer program code, the memory and the computer program code configured to, with the at least one processor, cause the apparatus to perform at least the following:
- obtain picture description information from a picture source, wherein the picture description information having been formed by adding at least one of a time, a location, or user information to the picture description information,
- obtain annotation information from at least two object sources, wherein the at least two object sources are different from said picture source,
- fuse said annotation information from said at least two object sources to form fused annotation information, wherein fusing the annotation information includes:
- select content for annotation from the annotation information from the at least two object sources by parsing the annotation information into a text format and selecting relevant text,
- assign a score value to each sentence from the selected relevant text based upon relevancy and redundancy in comparison with other sentences from the selected relevant text;
- filter, using assigned score values for the selected content, to reduce irrelevant or redundant information by removing stigma words and repetitive information, wherein repetitive information comprises selected content that overlaps in subject matter, and
- enhance the cohesion or coherence of the content by at least arranging the text with topically related themes to reduce non-fluency, and
- attach said fused annotation information to a picture to create an object note for said picture.
A computer program product comprising at least one non-transitory computer readable medium having computer executable program code portions stored thereon, the computer executable program code portions comprising program instructions configured for:
- obtaining picture description information from a picture source, wherein the picture description information having been formed by adding at least one of a time, a location, or user information to the picture description information,
- obtaining annotation information from at least two object sources, wherein the at least two object sources are different from said picture source,
- fusing said annotation information from said at least two object sources to form fused annotation information, wherein fusing the annotation information includes:
- selecting content for annotation from the annotation information from the at least two object sources by parsing the annotation information into a text format and selecting relevant text,
- assigning a score value to each sentence from the selected relevant text based upon relevancy and redundancy in comparison with other sentences from the selected relevant text;
- filtering, using assigned score values for the selected content, to reduce irrelevant or redundant information by removing stigma words and repetitive information, wherein repetitive information comprises selected content that overlaps in subject matter, and
- enhancing the cohesion or coherence of the content by at least arranging the text with topically related themes to reduce non-fluency, and
- attaching said fused annotation information to a picture to create an object note for said picture.
A network service comprising at least one server and at least one user device providing to a user at least the following:
- picture description information from a picture source, wherein the picture description information having been formed by adding at least one of a time, a location, or user information to the picture description information,
- annotation information from at least two object sources, wherein the at least two object sources are different from said picture source,
- fused annotation information formed by fusing said annotation information from said at least two object sources, wherein fusing the annotation information includes:
- selecting content for annotation from the annotation information from the at least two object sources by parsing the annotation information into a text format and selecting relevant text,
- assigning a score value to each sentence from the selected relevant text based upon relevancy and redundancy in comparison with other sentences from the selected relevant text;
- filtering, using assigned score values for the selected content, to reduce irrelevant or redundant information by removing stigma words and repetitive information, wherein repetitive information comprises selected content that overlaps in subject matter, and
- enhancing the cohesion or coherence of the content by at least arranging the text with topically related themes to reduce non-fluency, and
- an object note for a picture by attaching said fused annotation information to said picture.
Dependencies are inferred from the claim text.