A method for establishing a tree-shaped video semantic index based on context fusion
An index building and context technology, applied in the field of video retrieval, can solve the problem of not meeting user retrieval needs and other problems
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment Construction
[0066] Please refer to figure 1 , a method for building a context-integrated tree-shaped video index, first extracts the semantic information of the shot in units of shots, then obtains the semantic context of the video shot in a supervised manner, and uses a tree structure to represent the context. Then combine the semantics of shots and their contexts to reason about scene semantics. Finally, the shot semantics and scene semantics are embedded into the tree structure and used as video indexes. details as follows:
[0067] 1. For n training video clips video j Perform shot segmentation to obtain r training video shots. Extract and quantify the visual features of the shot, and construct a visual feature vector v.
[0068] Set the annotation semantic set Semantic={Sem t |t=1,...,e}, artificially mark the semantic Sem appearing in r shots t , added to the lens semantic set of each lens, and then for each type of lens semantic Sem t Construct a shot semantic training set, ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


