Video Underlying Feature Selection And Evaluation Of Correlation Analusis With The Audience

These years, information technology has developed quickly, in people’s daily lives all kinds of video information can be seen everywhere, the current movie trailers, video ads, etc. often previously served in a number of online social networks. Video design and production process, how to design video more able to attract the audience’s attention, have a positive effect on the mood of the audience, thereby improving such video viewers reputation and sales, generate economic benefits? This is an issue related to cultural enterprises, video director, editor, very concerned about.To discover the influence from the commercial videos’ low-level features to the popularity of the videos, the feature selection method should be used to extract the features influenced the videos’ evaluation mostly after getting the source data and the audiences’ evaluations of the videos. This paper improved the Correlation-Based Feature Selection(CFS) which is widely used and proposed an algorithm named CFS-Spearmen which combined the Spearmen correlation coefficient and the classical CFS to select features. The 3 data sets in UCI machine learning data base were employed as the experiment data and the results were compared with traditional CFS and Minimum Redundancy and Maximum Relevance(mRMR). The SVM was used to test the method in this paper. Finally, the proposed method was used in commercial videos’ feature selection and the most influential feature set was chosen. With KNN and SVM methods classifying characteristics and verification experiments, the results prove that choose the movement of the average variance, brightness, saturation, average three video underlying characteristics will cause certain influence to the audience’s evaluation, thus the prediction model has been built. CFS is verified by the ROC curve.CFS-Spearman method combined with the SVM is better than that of CFS-Spearman method combined with the KNN to see from the experimental results of CFS-Spearman method combined with the SVM is more suitable for this research, the research results laid a foundation for further research.