Research On 3D Image Watermarking Based On DIBR Rendering

3D video is becoming the main trend of digital media development with its strong three-dimensional perception and new visual experience.In current 2D-3D image/video conversion technology,depth-based drawing(DIBR)technology is due to its small amount of data.The advantages of compatibility,flexibility and adaptability are widely used to draw 3D views.The 3D image may encounter multimedia security issues such as illegal copying,distribution,and authenticity and integrity authentication of content during transmission and application.This topic mainly studies the characteristics of 3D image rendering based on depth map in the practical application process.The security issues faced by it,combined with the traditional digital image watermarking technology,studied the 3D image robust watermarking technology based on depth map rendering,thus achieving the purpose of copyright protection and content protection for 3D images.In view of the unique properties and characteristics of 3D images and the 3D image generation process and application requirements,this paper proposes a 3D image digital watermarking algorithm based on feature region and a 3D image digital watermarking algorithm based on geometric correction.In the feature area-based digital watermarking method,a meaningful binary image is selected as the watermark image to be embedded,the image is scrambled,and a pseudorandom sequence is used to spread the spectrum,and the original SIFT algorithm is used to find the original watermark image.The image feature points are filtered and a series of feature areas that do not overlap each other are formed according to the distance constraint method,and watermark embedding is performed in the frequency domain of the feature area.In the digital watermarking method based on geometric correction,the original image is subjected to three-layer wavelet transform,and the watermark information is embedded in the third layer transformed high-frequency coefficients.When the watermarked image is subjected to geometric attacks,the original image and the image being attacked are subjected to Feature point matching,image rotation and scaling correction parameters are obtained according to the relative relationship between the feature points,and the watermark is extracted after the image is corrected.This article has carried on the experiment to the method proposed,from the experimental process and the result analysis data to be able to see,the proposed method has the good robustness to the common signal distortion,also has the good to the geometry attack such as the rotation,the zooming and so on.Robustness.