CAJ | 학술논문

深入研究视频序列的基本特性，并在此基础上提出一种改进的DCT（离散余弦变换）域的分块视频水印方案。提出的方案中，通过结合人类视觉感知系统和运动目标检测技术，从视频图像序列中实时地发现并提取运动目标，确定运动目标所在区域，选取视频图像中的运动区域和高细节纹理复杂度区域作为视频水印的嵌入区域，并在水印的嵌入的过程中，通过细分像素块来减少像素分类带来的分块效应。通过仿真实验结果显示，该算法具有较好的不可感知性和鲁棒性。
심입연구시빈서렬적기본특성，병재차기출상제출일충개진적DCT（리산여현변환）역적분괴시빈수인방안。제출적방안중，통과결합인류시각감지계통화운동목표검측기술，종시빈도상서렬중실시지발현병제취운동목표，학정운동목표소재구역，선취시빈도상중적운동구역화고세절문리복잡도구역작위시빈수인적감입구역，병재수인적감입적과정중，통과세분상소괴래감소상소분류대래적분괴효응。통과방진실험결과현시，해산법구유교호적불가감지성화로봉성。
We make deep research on basic characteristics of the video sequence,and propose an improved block video watermarking scheme of discrete cosine transform (DCT)domain on the basis of the research.In the scheme,by combining the human visual system with the moving object detection technology,the moving object is found timely from a sequence of video images and is extracted,the area where the moving object located is determined,the moving areas and the areas with highly detailed texture complexity in video image set are chosen as the embedding area of video watermarks.Moreover,in the process of watermarks embedment,the scheme uses the sub-pixel blocks to reduce the block effects of pixel classification.Simulation experimental results show that the algorithm proposed in the paper has better imperceptibility and robustness.