Abstract: Accurate acquisition of users' objective needs is one of the urgent problems to be solved in intelligent personality customization. In traditional acquisition, there are problems such as misplacement of multi-attribute user groups' needs and strong dependence on experience designers. Therefore, a cluster user preference inference method based on image-product classification under the mechanism of brain visual cognition is proposed. Firstly, the sample domain and image domain are selected by statistical methods to reduce manual intervention;Then, similar user clusters are constructed based on k-modes clustering user cognitive feature elements;Intelligently identify the clusters of subjects through the K-nearest neighbor (KNN) algorithm, and carry out eye movements combined with electroencephalogram(EEG) experiments to extract key subjective and objective cognitive indicators in the clusters;Finally, technique for order preference by similarity to an ideal solution intelligence is used to classify the corresponding image-product group of cluster users and deduce the cognitive preferences of new similar users. Taking unmanned aerial vehicle (UAV) users as an example, the image-UAV classification of cluster users was established under the mechanism of brain visual cognition, and the demand preference of new users is quickly deduced. The results showed that it met the needs of users, and verified the effectiveness of the method.This study using the existing cluster the user's physiological and cognitive objective preference data to predict new users, improve the utilization rate of physical experiment, provide custom pattern swarm intelligence personality design rational objective perspective.

Key words: personalized needs, image recognition, the cluster theory, potential preference