GAN Face Recognition Deblurring with Key Point Encoder
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current generative adversarial network (GAN) based face deblurring methods improve image clarity but neglect unique features of blurry faces, resulting in low accuracy for face recognition.
Innovation Solution
A face recognition method using a trained generative adversarial network that incorporates a key point detection encoder and decoder, coupled with a feature extraction network, to deblur images while retaining facial features, and matches these features with a preset database for accurate recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current GAN based face deblurring methods are used to improve image clarity, then the clarity and intuitive experience of the face image is improved, but the unique features of the blurry face are neglected resulting in low accuracy for face recognition
Solution Approach 1:
The generator is segmented into two independent encoders (a first encoder for global feature extraction and a second encoder for key point detection) and a decoder. This segmentation allows the system to separately preserve global facial features and key point information during deblurring, preventing the loss of unique facial features while improving image clarity.
Solution Approach 2:
A feedback mechanism is introduced where the discriminator provides guidance to the generator through adversarial training. The discriminator evaluates whether the generated clear face image preserves authentic facial features, and this feedback is used to iteratively improve the generator's ability to retain unique facial characteristics while removing blur.
2Reliability
If traditional deblurring methods are used to enhance image clarity, then the visual quality of the face image is improved, but the recognition accuracy remains low due to feature neglect
Solution Approach 1:
The generator is designed with multi-functionality, serving both as an image deblurring device and as a feature preservation mechanism. By integrating key point detection and global feature extraction within the same generator structure, the system achieves reliable face recognition without requiring separate complex processing pipelines.
Solution Approach 2:
The network architecture employs a nested structure where the second encoder (for key point detection) and first encoder (for global features) are nested within the generator, which itself is nested within the overall GAN framework. This nested design allows complex functionality to be organized in a manageable hierarchy, improving reliability while controlling complexity.
Data Source
AI summary
The present disclosure provides a method and an apparatus for face recognition and a computer readable storage medium. The method includes: inputting a to-be-recognized blurry face image into a generator of a trained generative adversarial network to obtain a to-be-recognized clear face image; inputting the to-be-recognized clear face image to the feature extraction network to obtain a facial feature of the to-be-recognized clear face image; matching the facial feature of the to-be-recognized clear face image with each user facial feature in a preset facial feature database to determine the user facial feature best matching the to-be-recognized clear face image as a target user facial feature; and determining a user associated with the target user facial feature as a recognition result. Through this solution, the accuracy of the recognition of blurry faces can be improved.