基于残差约束的颜色补偿模型的训练方法及应用
By using a color compensation model based on residual constraints, combined with regional heterogeneity feature maps and absorption peak sensitive weight maps, pixel-level adaptive color compensation in digital pathological imaging was achieved, solving the problem of missing band information and improving the diagnostic consistency of pathological images and the accuracy of AI analysis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN SHENGQIANG TECH
- Filing Date
- 2026-05-18
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies in digital pathological imaging suffer from the loss or significant attenuation of specific band information, leading to color shifts, blurred boundaries, and loss of local tissue layers, which affect the analysis and interpretation of pathological images. Furthermore, global mapping compensation methods are prone to overcompensation or undercompensation, neglecting the requirements for diagnostic consistency.
A residual-constrained color compensation model is adopted, and a gating mechanism is used to combine regional heterogeneity feature maps and absorption peak sensitive weight maps to achieve pixel-level adaptive color compensation, accurately restoring the color differences and hierarchical relationships of missing bands.
It improves the accuracy and targeting of color compensation, avoids ineffective compensation for insensitive pixels, and ensures the diagnostic consistency of pathological images and the analytical accuracy of AI models.
Smart Images
Figure CN122199301B_ABST