Research on Key Technologies of Street View Video Privacy Information Processing
Qianqian Zhu, Yingnan Liu, Zihao Zhao · Journal of Physics Conference Series · 2022
Abstract As image and video data are recorded, stored or applied worldwide, data privacy protection is receiving increasing attention. Before data application, it is necessary to hide the information that can be personally identified in the video, for example, the human face that can be directly located to the personal identity, the license plate number that can be indirectly located to the personal identity and other data, while retaining the appropriate degree of data availability. Taking the actually collected European road street view video as an example, and taking the face and license plate as the privacy information, this paper proposes a method to automatically detect the face and license plate area in the street view video based on the YOLO5s model, and hide the detected object area with the Gaussian blur in the image processing technology. The experiment shows that the method is scientific, efficient and feasible.