Authentication of images for 3D cameras: Reversibly embedding information using intelligent approaches

摘要

In this work, a reversible watermarking approach for authentication of 3D cameras based on computational intelligence is presented. Two intelligent techniques based on differential evolution (DE) and hybrid DE are employed to optimize the tradeoff between watermark imperceptibility and capacity. The proposed approach is suitable for images of 3D cameras. These cameras generally work on the concept of time-of-flight and not only produce the 2D image but also generate the corresponding depth map. In this approach, the depth map is considered as secret information and is hidden in the integer wavelet transform of the corresponding 2D image. The proposed technique is prospective for authenticating 3D camera images and allows the secure transmission of its depth map. It has the advantage of the lossless recovery of original 2D image as and when needed. The watermarking of the 2D images is based on integer wavelet transform and threshold optimization. The threshold map thus obtained using the intelligent optimization approaches is not only used for watermark embedding, but is also utilized for authentication purpose by correlating it with the corresponding 2D transformed image. Experiments conducted on images and depth maps obtained using 3D camera validate the proposed concept.