Emotional Semantic Analysis Of Video And Its Application In Sensitive Video Recognition

With the rapid development of network information construction and the rapid rise of the communication technology industry,a large number of video resources have sprung up on the Internet.At the same time,video has become one of the main carriers of information transmission in today's era.A large number of video resources on the Internet are mixed with a lot of harmful information,such as horror factors in horror videos,bad temptations in pornographic videos and military sanctions in violent videos,etc.These harmful information not only pollute the growth environment of teenagers,but also have adverse effects on adults.If the emotional semantics of the video is regarded as the high-level semantics of the video,the sensitive video application can be effectively identified through reasonable utilization.After thinking about the above problems,this paper chooses to research on sensitive video recognition from the perspective of emotional semantic analysis of video.Fear is a special emotion type in emotion semantics.This emotion is related to the recognition of terrorist videos in sensitive videos,and the filtering of terrorist videos in sensitive videos is still at an early stage.At the same time,the relationship between multiple features in horror videos is rarely discussed.Therefore,this paper studies the fusion of multiple video features and proposes a feature fusion algorithm that can highlight the features of terrorist emotions.The experimental results on the horror emotion video show that the feature fusion method can effectively fuse the visual,audio and color emotion features in the video,give full play to the correlation between various features in the video and highlight the emotion semantics of the video,and has achieved good performance in the classification of horror videos.In addition,video is a kind of structural data.At present,the average processing method is mostly used to process video features.On the one hand,this will weaken the main emotional features of the video,on the other hand,it will also lose the structural features of the video.To solve this problem,this paper introduces a multi-instance learning algorithm to analyze the semantic of video emotion,so as to realize the recognition of video emotion.The experimental study of multi-instance algorithm on emotion data set shows that the multi-instance algorithm has certain performance effect in video recognition based on video emotion semantic analysis,and the analysis of experimental results also provides a reasonable research direction for further work improvement.