Research Of Image Retrieval On Cultural Attractions Based On Deep Convolution Feature Aggregation

At present,China's domestic cultural tourism resources accumulated for thousands of years are very rich.In the combination of cultural tourism and modern technology,although most of the cultural content has been converted to digital storage,its combination with big data is still very few.Facing the contradiction of rich digital content and less user income,it is urgent to mark and organize the cultural content data set and the corresponding big data algorithm to accurately solve the growing cultural needs of the masses through big data technology.Therefore,the research on content-based image retrieval related to Chinese cultural content is of great significance.In this paper,under the topic of cultural tourism,in the current image retrieval field,there is no current situation of Chinese culture-related image data sets,and to deal with the lack of content-based image retrieval algorithms under the characteristics of the data set.More than 3,000 images related to the content of Chinese cultural attractions,constructed an image dataset of Chinese cultural attractions,and conducted experiments using the current image retrieval algorithm.The design and implementation of a feature aggregation algorithm based on the center weighting of some feature maps are summarized and given.Excellent performance has been achieved on the image data sets of cultural attractions and other standard data sets,and content-based image retrieval of Chinese cultural content images has been achieved,and the specific standards for image collection and query under realistic scenarios related to the subject have been concluded.