Live attack detection method and training method of live attack detection model
By extracting texture features and transforming color space in facial images, and combining these with a liveness detection model for feature fusion, the problem of facial recognition systems being vulnerable to liveness attacks has been solved, achieving higher security and accuracy.
CN115862157BActive Publication Date: 2026-05-29ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
View PDF 1 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
- Filing Date
- 2022-12-01
- Publication Date
- 2026-05-29
AI Technical Summary
Technical Problem
Existing facial recognition systems are vulnerable to liveness detection attacks, such as identity spoofing through photos, mobile phone screens, and masks, leading to security issues.
Method used
By extracting texture features and transforming the color space of the target face image, texture images and color space transformed images are generated. These are then input into a liveness attack detection model for feature extraction and feature fusion to obtain classification features, thereby determining whether the image is a liveness attack image.
Benefits of technology
This improves the security of facial recognition systems, effectively identifying liveness attacks and preventing identity spoofing.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN115862157B_ABST
Abstract
Embodiments of the present specification disclose a live attack detection method, a training method and device of a live attack detection model, a storage medium and an electronic device. Texture feature extraction and color space transformation are performed on a target face image to obtain a texture image and a space transformation image of the target face image. The target face image, the texture image and the space transformation image are input into a live attack detection model. Feature extraction and feature fusion are performed on the target face image, the texture image and the space transformation image by the live attack detection model to obtain a first image feature of the target face image, a second image feature of the texture image and a third image feature of the space transformation image. The first image feature, the second image feature and the third image feature are fused by the live attack detection model to obtain a classification feature of the target face image. Whether the target face image is a live attack image is determined based on the classification feature by the live attack detection model.
Need to check novelty before this filing date? Find Prior Art