Method for mining community social relations on basis of video information

Abstract

The invention provides a method for mining community social relations on the basis of video information. The method includes steps of firstly, preprocessing character images in video; secondly, computing correlations between every two persons, drawing social relation graphs and mining community relations corresponding to feature sub-graphs; thirdly, analyzing and predicting group relations; fourthly, analyzing action expression and states of every single person. The method has the advantages that data can be mined from video data materials with abundant contents, practical problems of multi-person character complicated relations in existing video can be solved, and accordingly multi-person relations in various scenes can be effectively quantified and evaluated; the method supports real-time analysis on various scenes of practical life in the future, and accordingly powerful methodological support can be provided for analyzing character relations in multi-person complicated scenes.

Description

Technical field: [0001] The invention proposes a method for mining group social relations based on video data. It involves face detection, recognition, network structure analysis, calculation of social relations, etc., and belongs to the field of data mining. Background technique: [0002] Social relationship mining is a fast-growing field and an interdisciplinary subject that deeply integrates computer science and social science. Generally, the relationship between people is difficult to obtain through social survey methods such as direct questionnaires. It is necessary to use the current big data methods to indirectly mine social relationships. Social relationship mining has a wide range of applications in technology management, business intelligence, emergency management, anti-terrorism and anti-riot and other aspects. [0003] "Social connections are essential to finding meaning in life". By establishing and maintaining connections with others, individuals can achieve...

AI Technical Summary

Problems solved by technology

[0011] To sum up, the current research and algorithms are mostly based on data types such as quantitative characteristics or language analysis, and carry out indirect social relationship mining and analysis, but the level of social connections reflected in these data is relatively superficial, and most of them are medium and long-term social relationships. The accumulation of static relationships (such as address book data), or the embodiment of similarity in life patterns (such as track data), or the social interaction under a certain interest (such as Weibo or blog), is difficult to reflect in a multi-type, multi- Deep features of social relationships in groups of interests and multiple behavioral patterns

Images