一种基于明暗通道先验的投影融合补偿方法及系统
By employing a projection fusion compensation method based on the prior knowledge of the light and dark channels, and utilizing an embedded module and a cross-attention module for reflected light density and intensity information, the problem of poor compensation accuracy caused by mutual reflection of light in multiple projection devices is solved, thus achieving efficient and accurate improvement in projection imaging quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SICHUAN UNIV
- Filing Date
- 2024-01-10
- Publication Date
- 2026-07-17
AI Technical Summary
Existing multi-projection devices struggle to achieve high-precision compensation results in projection compensation, exhibiting poor accuracy and long processing times, especially with limited compensation effects for high-resolution images.
A projection fusion compensation method based on prior knowledge of light and dark channels is adopted. The prior knowledge of light and dark channels of light mutual reflection images is extracted by designing a compensation network. Feature interaction and reconstruction are carried out by using a reflected light density position embedding module, a reflected light intensity cross attention module and a cross-dimensional attention fusion module. A dataset is constructed and the compensation network is trained to approximate the target compensation function.
It achieves high-precision color and brightness compensation, reduces compensation time, improves image detail retention, and obtains more realistic projection imaging quality.
Smart Images
Figure CN117939092B_ABST