基于仿海马体记忆机制的图像处理方法及其相关设备
By employing an image processing method based on a simulated hippocampal memory mechanism, utilizing deep autoencoders and generative models for adversarial training, and combining meta-learning and clustering algorithms, the problem of insufficient recognition by deep learning algorithms under small sample data conditions is solved, achieving efficient image recognition and detection capabilities.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING LINJIN SPACE AIRCRAFT SYST ENG INST
- Filing Date
- 2022-12-03
- Publication Date
- 2026-07-17
AI Technical Summary
Existing deep learning algorithms are difficult to train effectively with small sample data, resulting in insufficient recognition capabilities in data-scarce application scenarios such as industrial product defect detection and medical image recognition.
An image processing method based on a hippocampal memory mechanism is adopted. Feature maps are extracted through a deep autoencoder to build a background knowledge graph. Adversarial training is carried out using generative and discriminative models. Combined with meta-learning and clustering algorithms, real-time target detection under sparse sample conditions is achieved.
It significantly improves image recognition capabilities under small sample constraints, achieving real-time target detection under sparse sample conditions of 10-102, thereby enhancing the accuracy of image classification and recognition and the performance of artificial intelligence algorithms.
Smart Images
Figure CN115908934B_ABST