Research On Reversible Visible Watermarking Algorithms

Visible watermarking technology is an effective active copyright protection mechanism,which superimposes the owner's copyright identification information in a translucent manner to a specific location of a digital image or video frame,visually displaying copyright information,thus effectively protecting digital intangible assets.Reversible visible watermark is watermarking technique in which the watermark pattern can be completely erased from the carrier image,and the original carrier image can be restored without distortion.There is no doubt that the development of reversible visible watermarking will further expand the application of visible watermarking to meet the needs of high-fidelity images in medical imaging,military intelligence and other fields.In this paper,digital images as the information carrier of visual watermark,two new visual watermark embedding algorithms are proposed:(1):In order to improve the robustness,visibility and transparency of visual watermarking technology,a method of adaptive embedding of visual watermark is proposed.First,the significant area detection of the host image is carried out based on multi-layer image segmentation.This aspect causes the visual watermark pattern to avoid the key objects in the host image and does not damage the value of the host image itself.On the other hand,the embedded area of the visual watermark pattern changes with the location of the host image,which increases the difficulty of the visual watermark pattern being maliciously removed,especially to resist the attack of automatic removal of the visual watermark pattern in bulk;Finally,taking into account the JND coefficient,grayscale distribution and image texture complexity of the visual watermark pattern area,the intensity of watermark embedding is calculated adaptively,so as to obtain the ideal visual watermark pattern transparency.The experimental results show that the algorithm has better performance than the existing algorithm.(2):In the research of the existing reversible watermarking algorithm,the original watermark image has a very important auxiliary reference function,and some algorithms entirely rely on the original watermark image to achieve reversible recovery of the carrier image.Therefore,in order to realize the semi-blind reversible watermarking algorithm,a kind of Visible watermark image embedding algorithm based on vector compression is proposed in this paper.Firstly,the watermarking image pattern is processed by edge detection,and the discrete vertex set ? of vector contour curve is obtained.Then,the discrete points in the set ?are simplified by the improved Douglas-Peucker algorithm,which gets the compressed image contour information of watermarking image.At the same time,we propose a reversible visible watermarking algorithm based on Convolution filter relief pattern and difference expansion,which realizes reversible embedding of visible watermarking image and lossless restoration of carrier image.The experimental results show that the proposed vectorization compression method has more advantages than the traditional image compression algorithm,which greatly reduces the storage space and thus reduces the distortion of the carrier image.In addition,the embedded watermarking pattern has translucent stereo relief effect,which makes the fusion of carrier image and watermarking image more natural.