一种基于深度学习的显微图像无缝拼接与增强重建方法
A seamless stitching and enhancement reconstruction method for microscopic images, constructed using deep learning, solves the problems of structural misalignment and detail blurring in microscopic image stitching. It achieves adaptive registration, seamless fusion, and detail enhancement of images, thereby improving the overall quality and structural continuity of the images.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- DINGCHANG MEDICAL TECHNOLOGY (SUZHOU) CO LTD
- Filing Date
- 2025-10-16
- Publication Date
- 2026-07-17
AI Technical Summary
Existing microscopic image stitching methods are prone to structural misalignment, abrupt gap boundaries, brightness jumps, and loss of detail under complex sample conditions. Furthermore, reconstruction algorithms struggle to effectively recover the fine textures of tissue structures and lack a collaborative optimization mechanism under an end-to-end unified architecture.
We construct a deep learning-based method for seamless stitching and enhanced reconstruction of microscopic images, including a structure-aware feature extraction network, an image registration module, a boundary attention-driven stitching network, and a residual hierarchical reconstruction network. Through deformable convolution, affine parameter estimation, boundary saliency detection, and multi-scale reconstruction branches, we achieve adaptive registration, seamless fusion, and detail enhancement of images.
It significantly improves the overall structural continuity, natural stitching, and detail clarity of images, and solves the problems of image misalignment, boundary breakage, and detail blurring in traditional methods. It has strong adaptability and generalization ability.
Smart Images
Figure CN121353070B_ABST