| Title |
Landmark Application De-identification Technique for Re-identification of Encrypted Objects |
| Authors |
채종혁(Jonghyeok Chae) ; 이동규(Dongkyu Lee) ; 오승훈(Seunghoon Oh) |
| DOI |
https://doi.org/10.5573/ieie.2024.61.6.87 |
| Keywords |
Image segmentation. Image encryption; Image de-identification |
| Abstract |
In this paper, we propose a process that de-identifies personal information and adds face landmarks to the detection target to recycle information from de-identified images due to concerns about personal information leakage. This method extracts and encrypts the pixel cluster of the object by applying a segmentation technique, stores the encryption key separately, transmits the processed image safely to the network using pseudonymization technology, and restores the original image using the encryption key stored in the stored encrypted image. You can. For re-identification of unidentified objects, facial landmarks were added and the accuracy was confirmed to be improved to 0.19 using the Yolov8 model. The proposed method introduces a de-identification method that can secure personal information protection and object detection performance by re-identifying de-identified information. |