The methods of analyzing user's relative sensitivities to visual features and applying them to content-based image retrieval

摘要

Information present invention apply a relative sensitivity to the user that by using a user feedback (feedback) apply the sensitivity analyzes the relative sensitivity analyzes for each time characteristic of the user to search for later visual features for search results based image retrieval relates to a method.; Such content-based image retrieval method according to the invention, the second stage from the first stage, and a user to study by analyzing the relative sensitivity of the user for each visual characteristics by using the search results for any query given the query information, , by inputting each time the characteristic value of the query to the neural network to the equation for obtaining the distance from each time feature a third step that takes the relative sensitivity for the feature between two images the relative sensitivity is applied for each visual feature space by applying the query and the search image, according to the fourth step of calculating a distance in feature space between the relative sensitivity applied to the query and the search image, and the distance in the feature space between the query and the search image the query and a fifth step of providing a similar image search to the user.