CAJ | 학술논문

针对现有的零水印算法对图像的光照、旋转、位置、尺度变化较为敏感的问题，提出一种基于脉冲耦合神经网络（M-PCNN）的零水印算法。该方法模拟生物视觉的感知过程，将图像分解成由若干二值图像组成的认知序列，计算序列中的每幅二值图像的熵作为图像零水印特征。理论与实验结果表明，该方法与现有的零水印算法相比，可以描述图像的全局特征，对图像的光照、旋转、位置等变化有较强的鲁棒性，而且具有较低的维数。
침대현유적령수인산법대도상적광조、선전、위치、척도변화교위민감적문제，제출일충기우맥충우합신경망락（M-PCNN）적령수인산법。해방법모의생물시각적감지과정，장도상분해성유약간이치도상조성적인지서렬，계산서렬중적매폭이치도상적적작위도상령수인특정。이론여실험결과표명，해방법여현유적령수인산법상비，가이묘술도상적전국특정，대도상적광조、선전、위치등변화유교강적로봉성，이차구유교저적유수。
Aiming at the problem of existing zero-watermarking algorithms to be sensitive to the variations of images in illumination, rotation,position and scale,we propose a zero-watermarking algorithm which is based on pulse-coupled neural network (M-PCNN).The method imitates the perception process of biological vision,decomposes the images into a recognition sequence composed of a body of binary images,and then calculates the entropy of each binary image in the sequence and uses them as image ’s zero-watermarking features. Theoretical and experimental results show that the proposed approach can describe the global characteristics of the image and is robust to image’s variations in illumination,rotation and position than the existing zero-watermarking algorithms.It is also has lower dimensions.