Research Of Emotional Image Retrieval Based On EEG Signal

With the development of high-speed computers, Internet and the popularity of the family of digital devices, a large number of digital photos, videos and other multimedia information are increasing rapidly. How quickly and accurately search in the multimedia database becomes an increasingly critical issue. The content-based image retrieval technology which analyzes images directly (such as feature extraction) and breaks through the traditional limitations of keyword-based retrieval is currently popular and common used.However, the image retrieval method based on image content is essentially a computer understanding of the image. The semantic information an image contains is much richer than these low-level characteristics can express. As the human brain wave signals not only can directly reflect the human thinking and emotional biological information, but also is easy to collect and analyze, in this paper, EEG signal analysis techniques, and image feature extraction techniques, we proposed an emotional image retrieval method based on EEG signal by combining the brain wave signal analysis and the image feature extraction techniques.The main contents of this article are as follows:(1) The paper systematically expounds the underlying characteristics of the reflected image of the human emotional information. And we also discuss the emotional image recognition problem from the image's color, texture, shape and other image features and the human ECG, EEG and other physiological signals.(2) Specifically address the EEG signal acquisition, feature extraction and analysis, etc., and discuss a resolution from the EEG signal in the way of human emotional information for our model of the proposed retrieval system foundation.(3) The emotional image retrieval model is proposed based on EEG. For improving the retrieval of semantic precision, the model combines the emotional information reflected by EEG and image information of the underlying. The model also takes advantage of SVM (support vector machine) algorithm, using machine learning to infer information about the link between the two features (EEG and image information of the underlying), to improve the retrieval accuracy.(4) Implements an emotional image retrieval prototype system based on EEG, base on which experimental verification and analysis were carried out. Experimental results show the good effectiveness and accuracy of the system, and further verify the feasibility of our retrieval model.