CN104103031A

CN104103031A - Normalization-based vector spatial data blind watermark method

Info

Publication number: CN104103031A
Application number: CN201410376222.3A
Authority: CN
Inventors: 张黎明; 闫浩文; 齐建勋; 张永忠; 李文德
Original assignee: Lanzhou Jiaotong University
Current assignee: Lanzhou Jiaotong University
Priority date: 2014-08-03
Filing date: 2014-08-03
Publication date: 2014-10-15
Anticipated expiration: 2034-08-03
Also published as: CN104103031B

Abstract

For a Robust vector spatial data watermark technology, a geometric transformation attack is difficult to deal with. A traditional geometric transformation resistance algorithm is difficult to resist a vertex attack. The invention provides a normalization-based vector spatial data blind watermark method through a data Normalization thought. Before a watermark is embedded, a coordinate value of spatial data is subjected to normalization processing to realize the invariance of translating and zooming. A normalization value of vertex coordinate data is revised to embed the watermark, and the watermark is embedded for multiple times to realize watermark blind extraction. A test result indicates that the method exhibits good robustness for attacks, such as the translating, the zooming, point adding and deleting, cutting, compression, factor ordering, data format conversion and the like, and meanwhile, spatial data errors caused by watermark embedding can be controlled.

Description

Based on normalized Vector spatial data blind watermark method

Technical field

The invention belongs to cartography and Geographical Information Sciences technical field, is the blind watermarking algorithm of a kind of vectorial geographical spatial data based on method for normalizing.

Background technology

Vectorial geographical spatial data is the important strategic information resource of country; it is the basic data of economy, military affairs, national defense construction and social development; obtaining conventionally will be by means of expensive professional equipment and a large amount of human and material resources of cost, so its copyright protection is most important.Vectorial geographical spatial data is preserved with digitized form, when facilitating data copy and propagating, also makes piracy become extremely easy.At present, in the urgent need to the reliable safety of technical guarantee geographical spatial data.

Along with the development of computing machine and network technology, copyright protection of digital product causes that people study widely and pay close attention to.Digital watermark, at digital content security, copyright protection and authenticated connection, is considered to the most promising technology at present.Digital watermark develops into video watermark, audio frequency watermark, Text Watermarking, database water mark, spatial data watermark etc. from initial image watermark.With respect to other digital watermarks, the development of Vector spatial data digital watermark relatively lags behind.

In general, according to the embedded location of watermark, Vector spatial data watermarking algorithm can be divided into: spatial domain watermark algorithm and frequency domain water mark algorithm.Yet the common algorithm of spatial domain algorithm is simple, easy to operate, watermark capacity is larger, can resist and increase point, deletion point, cutting and attacked by noise, but opposing geometric attack aspect effect is poor; And although frequency domain algorithm can be resisted geometric attack, poor robustness aspect summit increase, deletion.So existing algorithm is difficult to resist summit and attacks and geometric attack simultaneously.

Summary of the invention

For above-mentioned situation, the present invention proposes a kind of " based on normalized Vector spatial data blind watermark method ", this algorithm is normalized the coordinate figure of spatial data before embed watermark, to realize the unchangeability to Pan and Zoom, by revising the normalized value of vertex coordinates data, carry out embed watermark.Watermark is repeatedly embedded, and has realized the blind extraction of watermark; Solve the problem that Vector spatial data watermarking algorithm is attacked by geometric attack and summit simultaneously, improved the robustness of watermarking algorithm.

Choose the geometric apex coordinate embed watermark of Vector spatial data herein.In order to make watermark information embed all Vector spatial datas, with resisting cropping attack, watermark information should evenly embed X, the Y coordinate on all summits of Vector spatial data, to guarantee the anti-attack ability of watermark.Before watermark embeds, first the coordinate figure on Drawing Object summit is normalized, then embed watermark information in normalized value.Like this, when implement, after geometric transformation, not need geometric transformation to proofread and correct containing watermark spatial data, just can extract watermark.Meanwhile, in order to eliminate the correlativity between watermarking images pixel, strengthen the security of watermark, before watermarking images is embedding, application Logistic chaos algorithm scramble.The initial value of chaos transformation can be used as the key that watermark information extracts.

The inventive method comprises: the embedding of watermark information and the extraction of watermark information.

Watermark information embeds and to refer to watermark information is embedded into and in original vectorial geographical spatial data, obtains vectorial geographical spatial data with watermarked information.Step is as follows: first, read Vector spatial data, extract all coordinate points of Drawing Object, construct X ₀, Y ₀set; Secondly, to X ₀, Y ₀carry out respectively data normalization and amplify 10 ⁿdoubly; Secondly, in the coordinate points after amplification, use quantization method embed watermark; Finally, the coordinate points after embed watermark is dwindled to 10 ⁿdoubly, and renormalization, afterwards the spatial data that contains watermark is preserved to output.

The leaching process of watermark is the inverse process of telescopiny.First, read testing data, data are normalized and amplify 10 ⁿdoubly, n gets n value identical while embedding with same watermark; Secondly, by Hash () function, calculate i (i is the position of watermark); Secondly, extract the value of watermark bit W (i) by QIM quantization method, R water intaking prints the quantized value while embedding; Finally, to the one dimension watermark sequence extracting, carry out liter the dimension unrest of processing and be inverted, obtain final watermarking images.

Accompanying drawing explanation

Fig. 1 is that watermark embeds process flow diagram

Fig. 2 is watermark extraction process

Fig. 3 (a) is original watermark

Fig. 3 (b) is watermark after scramble

Fig. 4 is error distribution histogram

Fig. 5 (a) is stack before and after embed watermark

Fig. 5 (b) is partial enlarged drawing after stack

Fig. 6 (a) is data after cutting

Fig. 6 (b) is the watermark of extracting

Table 1 is n value and maximum error, RMSE relation

Table 2 is robustnesss of geometric attack

Table 3 is to increase, delete a little and the robustness of modifying point attack

Table 4 is that compression, key element are deleted the robustness of attacking

Embodiment

Effect in order to describe technology contents of the present invention, structural attitude, the object realizing in detail and to reach, describes in detail below in conjunction with embodiment.

Implementation step of the present invention may be summarized to be two parts: watermark information embeds and watermark information extracts.Below each implementation step is further elaborated.

Watermark information embeds and to refer to watermark information is embedded into and in original vectorial geographical spatial data, obtains vectorial geographical spatial data with watermarked information.

This algorithm be take vector graphics object as unit embed watermark.Set V for the summit of Drawing Object ₀be expressed as:

V ₀={v _i}; v _i=(x _i,y _i) i=1,2,…,N

X ₀={x _i}; Y ₀={y _i} i=1,2,…,N

V wherein _irepresent each summit, (x _i, y _i) represent two coordinate figures on summit, X ₀represent the set of abscissa value, Y ₀represent the set of ordinate value, the number that N is summit.

The idiographic flow that watermark embeds is as follows:

(1) read Vector spatial data, extract all coordinate points of Drawing Object, construct X ₀, Y ₀set;

(2) to X ₀, Y ₀carry out respectively data normalization and amplify 10 ⁿdoubly, be designated as ;

(3) exist the middle quantization method embed watermark that adopts respectively, concrete embedding grammar is as follows:

1. calculate the value i of Hash (x), x is watermark data to be embedded, and Hash (x) function is the Hash mapping function between x value and watermark bit;

2. extract watermark bit w[i to be embedded] (1 ), w is the watermark after scramble, the length that M is watermark;

3. apply QIM method, embed watermark in x, through type (3) calculates the data x ' after embed watermark, and R is quantized value;

(3)

(4) to after embed watermark dwindle 10 ⁿtimes, and renormalization;

(5) finally will preserve containing the output of watermark spatial data.

In watermark embed process, use Hash function to set up the higher significance bit part of normalized value (this part not can embed watermark), and the mapping relations between watermark bit (1 ~ M).By normalization data is amplified to 10 ⁿdoubly, will make watermark be embedded into numerical value compared with low order part, will greatly reduce like this watermark and embed the error causing.Data after embed watermark also need to use extreme value renormalization, in order to reduce as far as possible data error, do not affect the extraction of watermark, can not embed watermark in extreme value data.

The leaching process of watermark is the inverse process of telescopiny, as shown in Figure 3.Watermark extracting detailed process is as follows:

1) read testing data, data are normalized and amplify 10 ⁿdoubly, n gets n value identical while embedding with same watermark;

2) by Hash () function, calculate i (i is the position of watermark);

3) by QIM quantization method, extract the value of watermark bit W (i), R water intaking prints the quantized value while embedding;

4) to the one dimension watermark sequence extracting, carry out liter the dimension unrest of processing and be inverted, obtain final watermarking images.

In watermark embeds, watermark may repeatedly be embedded, and therefore adopts Voting principle to determine watermark information.The method of calculating is: define an integer sequence isometric with watermark sequence B (i)=0, i=1 ..., M}, M is watermark length.Single watermark bit =1, and-1}, in the leaching process of same watermark position, use formula B (i)=B (i)+ count the majority of watermark information value-1 and 1, if " 1 " is most, ; Then according to formula (4), reconstruct binary bitmap.

(4)

Claims

1. before watermark embeds, first the coordinate figure on Drawing Object summit is normalized, then embed watermark information in normalized value; Like this, when implement, after geometric transformation, not need geometric transformation to proofread and correct containing watermark spatial data, just can extract watermark.

2. the embedding of watermark: first, read Vector spatial data, extract all coordinate points of Drawing Object, construct X ₀, Y ₀set; Secondly, to X ₀, Y ₀carry out respectively data normalization and amplify 10 ⁿdoubly; Again, in the coordinate points after amplification, use quantization method embed watermark; Finally, the coordinate points after embed watermark is dwindled to 10 ⁿdoubly, and renormalization, afterwards the spatial data that contains watermark is preserved to output.

3. the extraction of watermark: first, read testing data, data are normalized and amplify 10 ⁿdoubly, n gets n value identical while embedding with same watermark; Secondly, by Hash () function, calculate i (i is the position of watermark); Again, extract the value of watermark bit W (i) by QIM quantization method, R water intaking prints the quantized value while embedding; Finally, to the one dimension watermark sequence extracting, carry out liter the dimension unrest of processing and be inverted, obtain final watermarking images.