Non-convex graphics processing method, device, storage medium and computer equipment

By identifying and processing the non-convex graphics of tubular non-rigid objects in the image, and using the hinge prediction network to predict endpoint information, the problem of difficulty in accurately detecting tubular non-rigid objects in the prior art is solved, and higher detection accuracy and segmentation accuracy are achieved.

CN114359165BActive Publication Date: 2025-03-25ALIBABA (CHINA) CO LTD
2 Cites 0 Cited by

Patent Information

Application Number
CN202111510898.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-10
Publication Date
2025-03-25
Estimated Expiration
2041-12-10

AI Technical Summary

Technical Problem

The prior art has difficulty in accurately detecting tubular non-rigid objects, especially in distinguishing multiple examples and improving segmentation accuracy.

Method used

By identifying the non-convex figure of the tubular non-rigid object in the image, the endpoint sets containing the endpoint joint circles are detected, and the relevant image features are input into the hinge prediction network to predict the information of the next predicted endpoint.

Benefits of technology

The detection accuracy of tubular non-rigid objects is improved, and the detection accuracy of tubular objects can be more accurately identified and distinguished, which enhances the segmentation accuracy.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention discloses a method, device, storage medium and computer equipment for processing non-convex graphics. The method comprises: identifying a tubular non-rigid object from an image, obtaining a non-convex graphic of the tubular non-rigid object; detecting the non-convex graphic, identifying an endpoint set including at least one endpoint joint circle, wherein the endpoint joint circle represents the intersection of the pipeline end of the tubular non-rigid object in the image; extracting image features associated with the endpoint joint circle from a feature map of the image according to the position of the endpoint joint circle in the image; inputting the image features associated with the endpoint joint circle into a hinge prediction network, and predicting the endpoint information of the next predicted endpoint associated with the endpoint joint circle. The present invention solves the technical problem in the related art that it is difficult to accurately detect tubular non-rigid objects.
Need to check novelty before this filing date? Find Prior Art