Infrared small target detection method and device deeply fusing edge details and deep features
By constructing an infrared small target detection network that deeply integrates edge details and deep features, and utilizing void spatial pyramid pooling and decoder feature reconstruction modules, the problems of low detection accuracy and imprecise segmentation of infrared small targets are solved, achieving higher detection accuracy and robustness.
CN116468980BActive Publication Date: 2026-05-29NAT UNIV OF DEFENSE TECH
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NAT UNIV OF DEFENSE TECH
- Filing Date
- 2023-03-31
- Publication Date
- 2026-05-29
Smart Images
Figure CN116468980B_ABST
Abstract
The application discloses an infrared small target detection method and device deeply fusing edge details and deep features, and the method comprises the following steps: S1, acquiring a data training sample set and performing data expansion and data set division; S2, constructing an initial infrared small target detection network containing a loss function based on ResNet-18; S3, minimizing the loss function of the infrared small target detection network in a supervised manner based on the data training sample set through a back propagation algorithm to obtain an optimized infrared small target detection network; and S4, inputting an image to be identified into the optimized infrared small target detection network to output an infrared small target detection result of the image to be identified. The device is used to implement the above method. The application has the advantages of simple principle, easy implementation, wide application range and improved detection precision.
Need to check novelty before this filing date? Find Prior Art