Method and apparatus for determining representative pictures in a picture set

By calculating the relative and absolute richness of images, a regression model is used to select representative images from the image set. This solves the quality problem caused by the failure to consider the correlation between images in existing technologies, and improves the quality and information content of the selected images.

CN115630182BActive Publication Date: 2026-05-26SHENZHEN XUMI YUNTU SPACE TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SHENZHEN XUMI YUNTU SPACE TECH CO LTD
Filing Date
2022-10-20
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing technologies do not consider the correlation between images when selecting representative images from an image set, resulting in the quality of the selected representative images not being guaranteed.

Method used

By acquiring the training dataset, feature vectors of the training images are extracted using a feature extraction network. The relative richness and absolute richness of each image relative to the base image of its corresponding identifier are calculated. Relative richness and absolute richness regression models are trained, and representative images are selected from the candidate image set using these models.

Benefits of technology

This improved the quality of the selected representative images and ensured that the correlation between images was fully considered, thereby enhancing the information content and quality of the images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115630182B_ABST
    Figure CN115630182B_ABST
Patent Text Reader

Abstract

This disclosure relates to the field of image processing technology, and provides a method and apparatus for determining representative images in an image set. The method includes: calculating the relative richness of each training image relative to its corresponding base image based on the feature vectors of each training image and its associated base image; training a relative richness regression model using the feature vectors of each training image and its associated base image, as well as the relative richness of each training image relative to its associated base image; calculating the absolute richness of each training image based on the feature vectors of each training image and other training images of its associated base image; training an absolute richness regression model using the feature vectors and absolute richness of each training image; and selecting representative images from the candidate image set using the absolute richness regression model and the relative richness regression model.
Need to check novelty before this filing date? Find Prior Art