Research And System Implementation Of Image Annotation Based On LDA Topic Model

In this paper,we focus on the problem of automatic annotation of images,and study the multi-objective image annotation method based on LDA topic model,and use the spatial proximity information to optimize the topic model to improve the accuracy of annotation.The specific research work can be reflected in the following aspects:We study the single-objective image annotation based on the LDA topic model.Using the LDA topic model,the topic probability distribution of the task image is calculated.By calculating the spatial position information of the visual feature in the topic,then it realize the segmentation of the region of object and annotation for the single-objective image.The experimental results show that this method has a good effect on the region of object and annotation for the single-objective image.In this paper,we study the problem of multi-objective image annotation,and use LDA topic model to realize the annotation of image.The topic probability distribution of the image is calculated by the generated topic model.The spatial proximity information of the visual feature is introduced to optimize the topic model.By using the filtered topic model,image is marked.The experimental results show that the addition of spatial proximity information to the topic model can improve the accuracy of annotation.This paper implements an image annotation system based on LDA topic model.Through the system,the user can not only achieve image text annotation,but also can achieve the region of object of the image segmentation.