Low-altitude detection method for non-fixed changing target
Through the low-altitude patrol change detection model with style transfer and differential attention mechanism, the problem of natural change mis-check in low-altitude patrol is solved, and efficient detection of non-fixed change targets is achieved.
CN120259911APending Publication Date: 2025-07-04XI'AN UNIVERSITY OF ARCHITECTURE AND TECHNOLOGY
0 Cites 0 Cited by
Patent Information
- Application Number
- CN202510168775.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-17
- Publication Date
- 2025-07-04
AI Technical Summary
Technical Problem
The existing low-altitude inspection methods cannot effectively distinguish between natural changes and non-natural changes, resulting in a decrease in detection accuracy and robustness.
Method used
A low-altitude patrol change detection model based on style transfer is adopted, the style characteristics and content characteristics of the image are extracted through the encoder, the style differences are analyzed using the differential attention mechanism, and the image distribution is unified through the adaptive normalized parameter replacement module to generate a changing image.
Benefits of technology
Improve the accuracy and robustness of change detection, reduce the interference of natural changes, and accurately identify non-fixed change targets.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention discloses a low-altitude detection method for a non-fixed change target, relates to the technical field of change detection, and particularly aims to solve the problem that an existing detection method cannot effectively distinguish natural change from non-natural change, so that false detection is generated for natural change with a higher proportion in low-altitude inspection. According to the method for low-altitude detection of the non-fixed changing target, style features and content features of an image are extracted through an encoder, the difference of the image style in the channel dimension is analyzed through a difference attention mechanism, and then the low-altitude detection of the non-fixed changing target is achieved through a self-adaptive normalization parameter replacement module. Natural condition distribution of the image is unified in a decoding stage so as to obtain a generated image, and a detection module performs normalized mapping on reconstruction loss based on the generated image so as to obtain a changed image. According to the method, the content feature difference of the change area can be highlighted, so that the accuracy and robustness of change detection are improved.
Need to check novelty before this filing date? Find Prior Art