CN108765256A

CN108765256A - Data waterprint embedded method based on a kind of dct transform that human vision is covered

Info

Publication number: CN108765256A
Application number: CN201810761387.0A
Authority: CN
Inventors: 王慧泉; 韩旭; 赵喆; 王金海
Original assignee: Tianjin Polytechnic University
Current assignee: Tianjin Polytechnic University
Priority date: 2018-07-10
Filing date: 2018-07-10
Publication date: 2018-11-06

Abstract

The invention discloses a kind of digital image watermark embedding methods for the dct transform covered based on human vision, include the following steps：Two-dimensional image is expressed as one-dimensional 0,1 sequence by the first step to the binary bitmap w (i, j) of m*n dimensions by sequential scan from left to right, from top to bottom；Original image is divided into 8*8 blocks, calculates variance by second step, the larger image block of M*N variance before obtaining；Third walks, and to this m*n image block through DCT direct transforms, while carrying out difference processing to DCT coefficient；4th step, compares the magnitude relationship of Mid Frequency DCT coefficient by each piece in the block for this m*n DCT, to determine that watermark picture point value embedded in the block is " 0 " or " 1 "；5th step, will be from one-dimensional 0 made of watermarking images conversion, and 1 sequence is embedded in original image, finally by DCT inverse transformations, forms the image containing watermark.

Description

Data waterprint embedded method based on a kind of dct transform that human vision is covered

Technical field

The content of present invention is related to dct transform on the basis of human vision covers characteristic, belongs to image watermark embedded technology Original image copyright is especially protected in field, also belongs to Information Hiding Techniques field.

Background technology

Internet exchange of today has become a not erasable part in life, weight of the multimedia as information interchange Medium is wanted to play increasingly important role, digital product is also various informative, emerges one after another.So more and more people select Network delivers oneself outstanding copyright or carries out net and casts net trade, but every have opposite, numerous pirate infringements Works are consequently increased, for example are not had copyrighted image, video display much etc. and utilized by people's various duplications of taking away, have copyright not award The works of power are even more that propagation or violated sale can be spread unchecked by infringer, these all seriously endanger the knowledge for having arrived copyright owner Property right, secondly as the characteristic of copyright is abundant, one of them is exactly that the duplication of digital product has no effect on works itself Quality, and cost is extremely low, so being easy to recycle by renovation.We need to protect the legal power of these works owner Benefit enables the development of its industry higher speed.The center of gravity for solving the problems, such as this is exactly Information of Development concealing technology, and digital watermarking Core technology exactly in this technical field.

Application No. is a kind of 201210165687.5 entitled " digital figure watermark sides of view-based access control model attention mechanism Two big characteristics of digital watermarking system are mentioned in the patent of method "：Robustness and invisibility, wherein invisibility depend on the mankind Target watermark is embedded into the marking area of image in this patent by visual characteristic from the beginning by the characteristic of visual attention location, To have preferable robustness to the watermark of extraction, but visual attention location can be caused by both of which：One is objective content is driven Dynamic bottom-up concern model；Another kind is the word top of subjective order guidance and ShiShimonoseki injection molding type, no matter it is any all too according to Rely the psychological activity in people, the judgement in notable area can inevitably be had an impact.

Application No. is the patents of 200910021564.2 entitled " being based on DCT algorithms digital image watermark embedding method " In algorithm be black white image to be embedded into the texture area of original image, however why do not mention will be embedded into original image Texture area, and in telescopiny not as good as directly using watermark pixel in black white image it is non-zero be 1 the characteristics of carry out it is embedding Enter.

Application No. is the patents of 201410479555.9 entitled " a kind of digital watermarking algorithm of image content-based " The embedded mobile GIS of a high power capacity is accomplished, but it is necessary to accomplish to the greatest extent may be used in the copyright of original image for protecting merely Insertion more than energy needs to ensure that hole is aggressive high, has as many as 22 to the extraction of DCT intermediate frequency coefficients in the patent, also to take into account It is cumbersome to carry out some for image texture, edge and smooth three characteristics.

Shown in sum up, covering image texture region in characteristic based on human vision, there are larger brightness and direction to become Change, human eye declines the resolution ratio of the area information, i.e., texture is covered, we are embedded in watermarking images to the texture area of host image It can successfully allow human eye that cannot identify, and leave in original image the mark of protection copyright, while we are to watermark The selective positioning of image on bianry image, this let us particularly simple completed in algorithmic rule the insertion of watermark with it is original The protection of image copyright.

Invention content

The content of present invention is to protect the copyright of original image to provide a kind of number for the dct transform covered based on human vision Word Image Watermarking, while making algorithm simple and effective as far as possible, and the picture of embedded watermark is made to have certain robust Property.

To achieve the above object, the technical solution of use includes the following steps：

1) bianry image that will be embedded in carries out the conversion of one-dimensional sequence, makes the bianry image of m*n by from left to right, from upper It is expressed as 0,1 sequence to lower scanning；

2) original image is subjected to 8*8 piecemeals, calculating variance is carried out by following formula of variance, and these variances press from Arrive greatly it is small be ranked sequentially, find the maximum m*n image block of variance；

Wherein, in every piece of m picture point average gray, f (i, j) it is corresponding exactly each gray value.

3) DCT direct transforms, following formula are carried out to these m*n image block；

In above-mentioned formula：As u=0 or v=0,When the N-1 that works as u, v=1,2 ... ...,.

4) each piece in the block for this m*n DCT, compare the magnitude relationship of two position DCT coefficients of Mid Frequency, simultaneously Make difference processing, i.e., larger coefficient makees the reduction processing of original error value half, and smaller coefficient makees the increase processing of original error value half, with Determine that watermark picture point value embedded in the block is " 0 " or " 1 "；

5) will from one-dimensional 0,1 sequence insertion original image made of watermarking images conversion, finally by DCT inverse transformations, The image containing watermark is formd, DCT reconstructed formulas are as follows.

Compared with the method for other insertion watermarks, equally there is certain attack tolerant, white Gaussian noise can be resisted, shears, JPEG compression etc. different types of attack in various degree, but telescopiny is simpler, has height to protection original image copyright Imitate quick advantage.

Description of the drawings

Fig. 1 is the flow diagram of the content of present invention；

Fig. 2 is the original image in an example of the content of present invention；

Fig. 3 is the two-value picture to be embedded in an example of the content of present invention；

Fig. 4 be the content of present invention an example in be embedded in the picture of watermark.

Specific implementation mode：

The content of present invention is described in further detail below in conjunction with attached drawing embodiment.

Embodiment one

Referring to the flow of Fig. 1, the bianry image of the Chinese character " rising sun " of a 32*32 is embedded into as watermarking images In the original image of the classical lina of 512*512

First, the bianry image Chinese character " rising sun " that will be embedded in carries out the conversion of one-dimensional sequence, make the bianry image of 32*32 by From left to right, scanning is expressed as 0,1 sequence from top to bottom；Then original image lina is subjected to 8*8 piecemeals, by calculating side Difference, and these variances are found 32*32 image block of variance maximum by being ranked sequentially from big to small；These 32*32 is schemed As block carries out DCT direct transforms, wherein having selected these to be the two as the dct coefficients of (4,3) in block and (5,2) two positions Number makees difference processing, i.e., larger coefficient makees the reduction processing of original error value half, and smaller coefficient makees the increase processing of original error value half, And for 1 guarantee, the former coefficient is less than the latter by picture point value, by picture point value, for 0 guarantee, the former coefficient is more than the latter；Later will From one-dimensional 0 made of watermarking images conversion, 1 sequence is embedded in original image lina, finally by DCT inverse transformations, is formd and is contained There are the lina images of watermark.

Claims

1. the data waterprint embedded method based on a kind of dct transform that human vision is covered, feature include the following steps：1) right Two-dimensional image is expressed as one-dimensional by the binary bitmap w (i, j) of m*n dimensions by sequential scan from left to right, from top to bottom 0,1 sequence；2) original image is divided into 8*8 blocks, calculates variance, the larger image block of M*N variance before obtaining；3) to this M*n image block carries out difference processing through DCT direct transforms, while to DCT coefficient；4) in the block each for this m*n DCT Block compares the magnitude relationship of Mid Frequency DCT coefficient, to determine that watermark picture point value embedded in the block is " 0 " or " 1 "；5) will From one-dimensional 0 made of watermarking images conversion, 1 sequence is embedded in original image, finally by DCT inverse transformations, is formd containing water The image of print.

2. the method as described in claim 1, it is characterised in that：The picture point value of binary map is utilized there was only 0 and 1 spy in step 1) Point, which facilitates, is quickly embedded into it in original image.

3. the method as described in claim 1, it is characterised in that：It is locked using this characteristic that human vision is covered in step 2) The region to be embedded in is the texture part of original image, while ensureing that the texture part size of selection and the watermark to be embedded in are big It is small consistent.

4. the method as described in claim 1, it is characterised in that：Difference processing is done using DCT coefficient avoid choosing in step 3) Degradation problems caused by the difference for the two intermediate frequency fragment positions selected is excessive or too small.

5. the method as described in claim 1, it is characterised in that：It can be quick by comparing Mid Frequency DCT coefficient in step 4) The embedded watermark picture point value of record is 0 or 1, facilitates extraction.