Image processing method and vehicle
By generating a snowflake mask and reconstructing the image using a physical scattering model, the problem of poor image quality in streaming rearview mirrors under snowy conditions was solved, achieving efficient and robust image restoration and improving driving safety in snowy weather.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GREAT WALL MOTOR CO LTD
- Filing Date
- 2026-03-05
- Publication Date
- 2026-06-05
AI Technical Summary
In snowy conditions, the streaming rearview mirror image is interfered with by snowflakes, resulting in a blurred and washed-out image, making it difficult to identify the outline of vehicles behind and affecting driving safety. The physical scattering model in the existing technology fails under high-density snow conditions, and the deep learning method has high computational complexity and limited generalization ability, which cannot meet the requirements of real-time performance and environmental adaptability.
By generating a snowflake mask, determining the ambient light parameters and target transmittance distribution map, and using a physical scattering model for image reconstruction, combined with a lightweight mask prediction model and motion estimation technology, the spatiotemporal consistency of the snowflake mask is optimized to achieve accurate localization and removal of snowflake interference.
It improves image quality in snowy conditions, ensures the naturalness and reliability of image restoration, meets the comprehensive requirements of automotive-grade systems for processing speed, environmental adaptability and computing resources, and enhances driving safety.
Smart Images

Figure CN122156379A_ABST