Image Privacy Regions Encryption Scheme Based on SIFT Features and Face Detection

Privacy protection of cloud images is widely concerned, while their usability is often neglected. Therefore, an image privacy regions encryption scheme based on SIFT features and face detection is proposed. Firstly, the algorithm detects the faces of the image and obtains the face regions. Then the SIFT feature points of the image are extracted, and the image is divided into several blocks. Meanwhile, the image blocks with more SIFT feature points are selected as SIFT feature regions. The face regions and the SIFT feature regions are viewed as the regions of interest (ROIs) that require privacy protection. In the scrambling encryption step, the ROIs are combined to improve the scrambling effect. Each ROI adopts a different chaotic sequence in the diffusible encryption step to enhance chaos complexity. Experiments show that the encryption scheme saves running time, and the usability of the encrypted image is preserved. This scheme is suitable for the encryption of images such as ID cards and invoices.