Stereoscopic Image Digital Watermark Technique And Its Application Research

Three-dimensional (3D) multimedia services can provide more realistic and immersive experiencesto end users when compared to traditional two-dimensional (2D) multimedia services, which isconsidered to be the next generation of image/video system. However, the stereoscopicimage/video is likely to encounter copyright protection, content authentication and visual distortion,etc. In recent years, digital watermarking technology has been used for digital products copyrightprotection, content authentication and quality assessment. In this paper, the concrete researchcontents are as follows:1) In order to protect the copyright of stereoscopic images in the condition of the image quality,a zero-watermarking stereoscopic image algorithm is presented based on hyper-chaotic system.Disparity zero-watermark is constructed according to the stabilities of disparity between lowfrequency bands of wavelet decomposition of the left and right views and direct coefficients ofdiscrete cosine transform. The relationship between security and watermark capacity is alsoanalyzed. Experimental results show that the proposed algorithm is strong robust asymmetrical andsymmetrical attack.2) In the past decade, various fragile digital watermarking techniques have been proposed formono-scopic image authentication. In this paper, a novel binocular visual characteristics basedpixel-wise fragile watermarking scheme for stereoscopic image authentication and tamperdetection is proposed. In the watermarking embedding process, the binocular just noticeabledifference (BJND) model is used for guiding watermarking embedding. In the tampering detectionprocess, a probability theory and an algorithm of binocular disparity are employed to improvepreviously obtained detection results to enhance authentication accuracy. In this paper, we alsopropose a new watermarking algorithm for stereoscopic image tamper detection and self-recovery.Both of matching blocks in the left and right views have similar pixel values, so that fewer bits areallocated for recovery watermark generation, which can increase the quality of watermarkedstereoscopic image. A hierarchical tamper detection strategy with a four levels check up ispresented to improve the accuracy of tamper localization.3) Most of digital watermarking algorithms are designed for only single purpose. In this paper, anew stereoscopic image watermarking algorithm with multipurpose is proposed with stereomatching technique, which can be used for copyright protection, content authentication and tamperdetection. Specifically, the chaotic feature watermarks are generated according to the stability of low frequency coefficients and largest singular value. Each block of the two views is classified intomatchable or non-matchable block to embed digital watermarking with different purpose. At thereceiving side, using stereo marching technique as a bridge, the robust and fragile watermarks canbe blindly extracted without access to the host stereo image.4) In order to enrich and expand the theoretical connotation of digital watermarking technology,this paper puts forward two kinds digital watermarking based reduced-reference stereoscopicimage quality evaluation model.The first type: Through analyzing the image structure direction characteristics and thestereoscopic perception of human visual system, and combining semi-fragile digital watermarkingand support vector regression (SVR), a reduced-reference stereoscopic image quality assessment(RR-SIQA) model based on visual perception and zero-watermark is proposed. In this model, theview zero-watermark is constructed by judging the relation of the horizontal and vertical waveletcoefficients. Meanwhile, the disparity zero-watermark that reflects the stereoscopic perceptionquality is constructed by using the disparity between the left and right views, and then the relativityof two watermark-recovering rates and subjective quality scores can be learned by the trainingprocedure of SVR, finally, stereoscopic image quality was predicted by trained SVR.The second type: a novel RR-SIQA algorithm is proposed by means of an unconventional use ofwatermarking. In the proposed RR-SIQA algorithm, the features of stereoscopic image areextracted, and then embedded into the stereoscopic image as invisible hidden information by usingWatson contrast sensitivity masking and JPEG quantization table based perfectly blind dithermodulation (DM) watermarking method. In order to improve the reliability of the watermark, somechannel coding techniques are applied before the process of embedding watermark. At the receiver,the users decode the hidden watermarking information and use them to measure the quality of thedistorted stereoscopic image. The proposed algorithm overcomes the limitations of some currentRR-IQA methods that require an auxiliary channel.