An enteroscope video detection method and device based on quality stability joint evaluation

By adopting a colonoscopy video detection method based on joint quality and stability evaluation, the problems of image quality fluctuation and target instability in colonoscopy video detection are solved, and high-precision and robust colonoscopy video detection results are achieved.

CN122347584APending Publication Date: 2026-07-07HUAZHONG UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Applications(China)
Current Assignee / Owner
HUAZHONG UNIV OF SCI & TECH
Filing Date
2026-05-28
Publication Date
2026-07-07

AI Technical Summary

Technical Problem

Existing colonoscopy video detection technology is difficult to meet the requirements of high precision and strong robustness in complex physical environments. Drastic fluctuations in image quality and short-term instability in target detection lead to frequent missed detections, false detections, and detection frame flickering.

Method used

A colonoscopy video detection method based on joint quality and stability evaluation is adopted. Multi-scale feature extraction and spatiotemporal aggregation are performed through feature extraction and spatiotemporal aggregation modules. Combined with frame-level image quality evaluation and short-term target stability evaluation, the detection strategy is dynamically adjusted to output highly reliable detection results.

Benefits of technology

It effectively filters out interference from low-quality images and unstable detection, significantly suppresses missed detections, false detections, and detection frame flickering, and achieves highly reliable multi-target detection in colonoscopy videos.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122347584A_ABST
    Figure CN122347584A_ABST
Patent Text Reader

Abstract

The application provides an enteroscope video detection method and device based on quality stability joint evaluation, and belongs to the technical field of computer vision. The device comprises: an enhanced feature map is obtained from an enteroscope video frame sequence through a feature extraction and spatio-temporal aggregation module; a frame-level image quality evaluation module performs regression evaluation on frame picture quality based on the enhanced feature map, and outputs a frame-level image quality score; a short-time target stability evaluation module processes the enhanced feature map to obtain a short-time target stability score; a joint perception regulation module fuses the frame-level image quality score and the short-time target stability score into a joint reliability factor, dynamically modulates an initial confidence in a logarithmic space based on the joint reliability factor to obtain a final confidence, and simultaneously adjusts a judgment threshold value according to the joint reliability factor, judges the final confidence based on the adjusted judgment threshold value, thereby effectively filtering out interference introduced by low-quality images and unstable detection.
Need to check novelty before this filing date? Find Prior Art