A multi-modal affective semantic perception computing and transmission method
By using feature mapping of multimodal data and a continuous-time state-space model, the problems of geometric distortion and temporal asynchrony in multimodal semantic communication are solved, achieving robust and accurate transmission under complex channel conditions and improving the reliability of sentiment analysis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHONGQING UNIV OF POSTS & TELECOMM
- Filing Date
- 2026-04-14
- Publication Date
- 2026-06-26
AI Technical Summary
Existing technologies in multimodal semantic communication suffer from problems such as geometric distortion, temporal asynchrony, and insufficient robustness to channel fading, leading to unstable feature extraction, information loss, and semantic information distortion.
By using feature mapping, geometric consistency correction, continuous-time state-space model, and structural consistency mapping of multimodal data, we achieve accurate alignment and robust transmission of multimodal features, including feature extraction, fusion, and channel symbol mapping of text, visual, and audio modalities.
It achieves robustness and transmission efficiency of multimodal semantic communication in wireless scenarios with limited bandwidth and changing environments, ensuring the accuracy and integrity of sentiment analysis.
Smart Images

Figure CN122294147A_ABST