Hidden structure learning-based image digest generation method

Abstract

The invention discloses a hidden structure learning-based image digest generation method. The method comprises the following steps: (1) extracting the HSV (Hue, Saturation and Value) color histogram characteristic, visual word characteristic and orientation histogram characteristic of a picture; (2) performing normalization preprocessing on the three characteristics extracted in the previous step and combining the three characteristics into a characteristic vector after normalization; (3) constructing a structural support vector machine with a hidden variable, selecting training sets from a database for many times, and performing weight coefficient learning on picture sets related to different subjects in a training set; (4) selecting picture sets related to different subjects from the database by using the weight coefficient obtained by the learning in the previous step, predicting hidden characteristic selection preferences of the picture sets and generating a digest picture set corresponding to the hidden characteristic selection preferences. The method has higher information coverage and lower redundancy, can implicitly learn the different preferences of the picture sets related to the different subjects in characteristic selection and has a better effect compared with the conventional method.

Description

technical field [0001] The invention relates to image summary generation, in particular to an image summary generation method based on hidden structure learning. Background technique [0002] At present, with the development of storage technology and network technology, a large number of image files are uploaded on the Internet every day. On the one hand, users can access more and more image data, on the other hand, the increasingly complex structure and redundancy of the data itself also bring great difficulties to users to obtain useful information. For example, users can use search engines to search a large number of pictures through keywords, but these pictures contain a large number of heavily similar pictures, which hinders the overall presentation of information, and a summary that takes into account information coverage and redundancy becomes more reasonable. Another example is that there are a large number of personal photo albums on the photo sharing website Flick...

AI Technical Summary

Problems solved by technology

Third, the differences in topics, that is, the information contained and preferred by different topic-related image sets are different

Because many traditional summarization methods often consider each picture in the summation independently, and do not comprehensively consider the complementary information and redundant information contained in each picture in the summarization picture set, so they cannot comprehensively analyze the original picture set and get a good result. summary of

In addition, although some summary generation methods proposed in the past consider the collection of summary pictures as a whole structure, they do not take into account the differences of related topics, so they cannot well adapt to user needs.

Images