CAJ | 학술논문

提出一种基于双重水印的音频篡改检测算法，利用二值图像进行降维处理作为水印信号，按照适当的分段方式，平均嵌入到音频中生成内容认证水印，利用段序号的二值数字串生成完整性水印，实现音频的版权保护和内容认证。通过提取水印并与原始水印进行比对，判断音频的完整性，并精确定位篡改区域。仿真实验表明，算法具有很好的不可感知性，且对音频篡改检测的准确性高。
제출일충기우쌍중수인적음빈찬개검측산법，이용이치도상진행강유처리작위수인신호，안조괄당적분단방식，평균감입도음빈중생성내용인증수인，이용단서호적이치수자천생성완정성수인，실현음빈적판권보호화내용인증。통과제취수인병여원시수인진행비대，판단음빈적완정성，병정학정위찬개구역。방진실험표명，산법구유흔호적불가감지성，차대음빈찬개검측적준학성고。
It proposes an audio tamper detection algorithm based on dual watermark. It uses binary image as the watermark signal by reducing the image dimension. By appropriate subparagraph, the signals are embedded into the audio averagely so to generate content authentication watermark. Using the binary digital string of section number, the integrity watermark is generated so to implement audio content authentication and copyright protection. By extracting the watermark and comparing with the original watermark, it determines the audio integrity and locates the tamper region precisely. Simulation results show that this algorithm has good imperceptibility and high accuracy to audio tamper detection.