一种基于边界比特优化的二值描述符学习方法
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING UNIV OF POSTS & TELECOMM
- Filing Date
- 2024-03-12
- Publication Date
- 2026-07-17
AI Technical Summary
Existing binary descriptor learning methods fail to effectively address the dimensionality collapse problem in the real-valued feature extraction stage and the fuzzy bit problem in the binarization stage, resulting in a decrease in binary descriptor performance and failing to fully consider the mutual influence between the two.
A binary descriptor learning method based on boundary bit optimization is designed. By introducing a descriptor gradient optimization module (DGOM) during the gradient optimization process of the neural network, inefficient dimensions are optimized in the real-valued feature extraction stage, and blurred bits are optimized in the binarization stage. Image enhancement and contrast loss calculation are used to adjust the model parameters to improve feature matching accuracy.
It achieves improved image matching accuracy and feature representation capabilities with low computational and storage costs, providing a binary descriptor with richer information capabilities, suitable for computer vision tasks.
Smart Images
Figure CN118015429B_ABST