CN116757909B

CN116757909B - BIM data robust watermarking method, device and medium

Info

Publication number: CN116757909B
Application number: CN202311049686.9A
Authority: CN
Inventors: 徐建新; 周倩雯; 蒋星祥; 张磊; 胡娈运; 佘迎晨; 任娜; 朱贤姝; 边婉平; 朱长青
Original assignee: Jiangsu Provincial Archives Of Surveying And Mapping Data; Nanjing Normal University
Current assignee: Jiangsu Provincial Archives Of Surveying And Mapping Data; Nanjing Normal University
Priority date: 2023-08-21
Filing date: 2023-08-21
Publication date: 2023-11-03
Anticipated expiration: 2043-08-21
Also published as: CN116757909A

Abstract

The invention belongs to the technical field of watermark information security, and particularly discloses a BIM data robust watermarking method, device and medium. Experimental results show that the method is high in safety and strong in robustness to geometric attacks, can resist digital-analog separation attacks common to BIM+GIS scenes, can meet actual use requirements, and provides a feasible scheme for BIM data copyright protection.

Description

BIM data robust watermarking method, device and medium

Technical Field

The invention belongs to the technical field of watermark security, and particularly relates to a BIM data robust watermarking method, device and medium.

Background

In recent years, with the vigorous development of smart city construction, it is an important fundamental work to develop a novel city three-dimensional modeling technology to better digitally describe urban geographic entities. The BIM (Building Information Modeling) data is widely focused on being capable of expressing the full factor information of the city, providing basic data for the smart city, and being helpful for the description of geographic entities and analysis of microscopic space of the city. However, BIM data has the problems of high production cost, high precision, more secret involved, infringement, secret leakage and the like in the data distribution and use process, and not only damages the interests of legal owners of the data, but also is more likely to cause serious losses to national interests and safety, so that the safety protection of the BIM data has important significance. The digital watermarking technology is used as a leading-edge information security technology, and hides watermark information such as copyrights, users and the like in carrier data, so that copyright identification, infringement tracing, content authentication, additional information hiding and the like are carried out, and a feasible solving approach is provided for BIM copyright and security protection.

At present, the BIM model robust watermarking algorithm has few research results, the three-dimensional grid and point cloud model and BIM data are visual three-dimensional models, the robust watermarking algorithm has more research results, and the BIM data watermarking algorithm has reference significance. Three-dimensional model robust watermarking algorithms can be divided into three classes: a robust watermarking algorithm based on a transform domain, a robust watermarking algorithm based on global features and based on local features.

The robust watermarking algorithm based on the transform domain generally adopts frequency analysis methods such as discrete cosine transform, wavelet analysis, manifold harmonic basis and the like, and embeds the watermark by using coefficients obtained by model change. Watermark synchronization is established with the center of the grid and salient points, as based on visual saliency associated with the wavelet coefficient vector, and the watermark is inserted into the three-dimensional grid by quantizing the wavelet coefficients (Mohamed Hamidi, aladine Chetouani, mohamed El Haziti, mohammed El Hassouni, hocine cherifei, blind Robust 3D Mesh Watermarking Based on Mesh Saliency and Wavelet Transform for Copyright Protection[J ]. Information,2019,10 (2.) ]). After applying a 3-level Haar transform on the three-dimensional model, the optimized parameters and the Arnold transform scrambled watermark are embedded into DCT coefficients, realizing watermark embedding ([ Tamane SC, deshmukh R R, jadhavpatil V. Optimization of Blind 3.3 3D Model Watermarking Using Wavelets and DCT. IEEE Computer Society, 2013. ]). The transform domain-based algorithm has better invisibility, but the frequency domain analysis needs to save additional information, so that the transform domain-based algorithm is suitable for a three-dimensional model with a large number of vertexes, and is difficult to apply to BIM data with low data redundancy and fewer primitives.

The robust watermarking algorithm based on global features is to embed watermarking information into geometric domain global feature quantities such as three-dimensional model histogram offset, vertex-to-model center distance vertex norms, region error expansion and the like. For example, model vertices are clustered based on an ad hoc graph SOM, and watermark information is embedded in a three-dimensional model by modifying vertex norm distribution. Three-dimensional data is divided into groups based on a Shape Diameter Function (SDF), and watermarks are embedded into all groups using a histogram mapping function ([ Soliman M, hassanien a E, onsi H m. Robust watermarking approach for 3D triangular mesh using self organization map[C ]// Computer Engineering & Systems (ICCES), 2013 8th International Conference on.2013 ]. In general, due to the high stability of global statistical features, the method can resist more types of malicious attacks, but feature invariant extraction is limited by data types, BIM data has outstanding layering characteristics, and watermark embedding relations are difficult to construct by using global geometric features such as histogram statistics and the like.

The robust watermarking algorithm based on local features realizes watermark information embedding by adjusting the relative positions, vertex sequences, vertex coordinates and the like of adjacent points of the three-dimensional model. A distortion function based on vertex normals is designed as in the prior art and bit information is embedded in the bit plane of the vertex coordinates ([ Hang methou, kejiang Chen, weiming Zhang, yuanzhi Yao, nenghai Yu. Distortion Design for Secure Adaptive-D Mesh Steganography ] [ J ]. IEEE trans. And converting each three vertexes from a three-dimensional domain to a two-dimensional domain in the three-dimensional model by using corresponding transformation matrixes, and realizing watermark information embedding by modifying vertex coordinates. The local feature-based method has higher flexibility and capability of resisting attacks such as clipping, and has lower requirements on the vertex number of carrier data ([ Khalil o.h., elhadad Ahmed, ghareb a.. A blind proposed 3D mesh watermarking technique for copyright protection[J ]. The Imaging Science Journal,2020,68 (2)) ]. However, because of strict spatial topology of the BIM data primitives, partial primitive position changes may cause internal model distortions, resulting in model errors, and thus such methods have not been well used for BIM data.

In summary, the watermark algorithm based on the transform domain has better invisibility, but has higher requirements on the number of vertexes of the carrier data, the watermark algorithm based on the global feature can resist more types of attacks, but is difficult to keep watermark information intact under the data clipping attack, the global feature structure is limited by the original data format, and the watermark algorithm based on the local feature has lower requirements on the number of vertexes of the carrier data compared with the transform domain method, and has higher flexibility and robustness. However, since the BIM data and the three-dimensional grid and point cloud model data are greatly different in organization form, the method cannot be applied to the BIM data at present.

Disclosure of Invention

The present invention has been made to solve the above-mentioned problems occurring in the prior art. Therefore, there is a need for a robust watermarking method, device and medium for BIM data, specifically, a robust watermarking method is provided for BIM data by combining spatial position and attribute information in local features of primitives, the spatial position of the primitives of the BIM data is expressed by using a central point of a bounding box, the movable primitives perform spatial disturbance based on the spatial position features, and the immovable primitives perform invisible character writing based on the attribute information features, so as to realize multiple embedding of watermark information. Experiments show that the method has strong robustness to translation, rotation, detail degree change, graphic element deletion and digital-analog separation attack, and can effectively protect copyright information of a model in BIM data use and distribution processes.

According to a first technical scheme of the present invention, there is provided a robust watermarking method for BIM data, the method including:

watermark information generation:

obtaining a keyOriginal watermark sequence read by scanning +.>Defining scrambling function +.>As shown in formula (3):

（3）

wherein L is Logistic chaotic system for watermark decorrelation whenWhen the Logistic mapping is in a complete chaotic state, the calculation formula is shown as formula (4):

（4）

in the middle of，/>And->By key->Setting.

Watermark embedding:

based on BIM data primitive addition and deletion, the main body and the structural diagram element with stable BIM data are screened as watermark information carriers, and the central point of a primitive bounding box is calculatedJoin set->Expressed as->，/>，/>The total number of the watermark carrier primitives; selecting the high position of the decimal point of the XYZ coordinates, synchronizing the watermark bit of the graphic primitive, performing movement test on the graphic primitive, and adding the movable graphic primitive into the set +.>Non-movable element center adding set +.>；

Traversing collectionsAnd embedding watermark information into the low positions of X coordinates and Y coordinates of the central point of the graphic primitive. Taking X coordinate as an example, the disturbance method is shown as a formula (5), when the calculated value of the coordinate setting position and the corresponding watermark information +.>When the two primitives are not matched, the corresponding primitive is disturbed to realize watermark information embedding;

（5）

In the middle ofRepresentation->Middle->Center point coordinates of each graphic element embedded with watermark information, < >>Embedding a position for watermark information->A watermark value representing the mapped bits;

based on keys before invisible character embeddingSetting invisible character code to be embedded +.>Attribute fields carrying watermark information with the graphic elements; when watermark is embedded, the set is traversed>Each non-movable element when said non-movable element corresponds to watermark bit information +.>1 and the primitive attribute field contains invisible characters +.>At this time, character embedding is performed according to the formula (6):

（6）

in the middle ofFor the collection->Middle->Primitive attribute field of individual primitive,/->To embed attribute fields after invisible characters.

Further, the primitive bounding box center point is calculated by the following formulaCoordinates of (c):

（1）

in the method, in the process of the invention,representing the center point of the bounding box of the graphic element>Is>Represents the bounding box abscissa maximum, +.>Representing the minimum value of the abscissa of the bounding box, +.>Representing the center point of the bounding box of the graphic element>Ordinate of>Representing the maximum value of the ordinate of the bounding box,/>Representing the minimum value of the ordinate of the bounding box, +.>Representing the center point of the bounding box of the graphic element>Vertical coordinates of>Representing bounding box vertical coordinate maximum,/-)>Representing bounding box vertical coordinate minima.

Further, watermark bits are calculated by the following formula：

（2）

Wherein ⌊ ⌋ denotes a rounding down,Pis the least significant bit after the decimal point of the coordinates,modindicating that the remainder is taken,Sthe number of significant bits is mapped for the watermark,Drepresents the coordinate value of the axis of the element center point X, Y, Z,representing the watermark length.

Further, the method further comprises watermark information detection, the watermark information detection comprising:

BIM data is acquired, and is preprocessed to obtain a movable element setAnd non-movable element set +.>；

Traversing to a movable primitive setPrimitive based on key->Setting watermark information carrying bit, extracting watermark information as shown in formula (7), and updating detected watermark value>And increasing the number of times of writing watermark values in the corresponding positions;

（7）

traversing a set of non-movable primitivesPrimitive based on key->Invisible character of the setting->Extracting watermark information from the primitive attribute field as in formula (8), and updating the detected watermark value>And increasing the number of times of writing watermark values in the corresponding positions;

（8）

traversing all watermark bits based on majority principle and updating watermark information as in formula (9) to obtain watermark sequenceAnd +.>The dimension is increased and the scrambling is reversed, so as to obtain a BIM data watermark information image;

（9）

For the sequence->Position->Number of times of mapping into watermark->Indicate->Bit watermark information->。

According to a second aspect of the present invention, there is provided a BIM data robust watermarking method apparatus, the apparatus comprising:

a watermark information generation module configured to generate watermark information:

（3）

（4）

in the middle of，/>And->By key->Setting.

A watermark embedding module configured to embed a watermark:

based on BIM data primitive addition and deletion, the main body and the structural diagram element with stable BIM data are screened as watermark information carriers, and the central point of a primitive bounding box is calculatedJoin set->Expressed as->，/>，/>The total number of the watermark carrier primitives; selecting the high position of the decimal point of the XYZ coordinates, synchronizing the watermark bit of the graphic primitive, and carrying out graphic primitiveMoving test, movable element added set +.>Non-movable element center adding set +.>；

Traversing collectionsAnd embedding watermark information into the low positions of X coordinates and Y coordinates of the central point of the graphic primitive. Taking X coordinate as an example, the disturbance method is shown as a formula (5), when the calculated value of the coordinate setting position and the corresponding watermark information +. >When the two primitives are not matched, the corresponding primitive is disturbed to realize watermark information embedding;

（5）

（6）

Further, the watermark information generation module is further configured to calculate the primitive bounding box center point by the following formulaCoordinates of (c):

（1）

in the method, in the process of the invention,representing the center point of the bounding box of the graphic element>Is>Represents the bounding box abscissa maximum, +.>Representing the minimum value of the abscissa of the bounding box, +.>Representing the center point of the bounding box of the graphic element>Ordinate of>Representing the maximum value of the ordinate of the bounding box,/>Representing the minimum value of the ordinate of the bounding box, +. >Representing the center point of the bounding box of the graphic element>Vertical coordinates of>Representing bounding box vertical coordinate maximum,/-)>Representing bounding box vertical coordinate minima.

Further, the watermark information generation module is further configured to calculate watermark bits by the following formula：

（2）

Further, the apparatus further comprises a watermark information detection module configured to:

（7）

（8）

（9）

for the sequence->Position->The number of times the watermark is mapped in,/>indicate->Bit watermark information->。

According to a third aspect of the present invention, there is provided a readable storage medium storing one or more programs executable by one or more processors to implement the method as described above.

The invention has at least the following beneficial effects:

aiming at the BIM data copyright protection requirement, the invention analyzes the data characteristics of the BIM common format RVT, and aims at the characteristics of small BIM data redundancy, strict topological relation of the graphic elements and the like, and a digital watermarking method is provided based on the local characteristics of the model, so that the movable graphic element disturbance and the invisible character embedding of the immovable graphic element are combined, the number of watermark carriers is expanded, and the repeated embedding of digital watermarking information is realized. Experiments show that the invention has good imperceptibility, small redundancy added by embedding invisible characters, and strong robustness to translation, primitive deletion and detail degree change attack, and can resist digital-analog separation attack. Meanwhile, the watermark detection process is blind detection, and the participation of the original BIM data is not needed, so that the watermark detection method has strong practicability. The invention can provide a feasible solution for BIM data copyright protection.

Drawings

FIG. 1 is a schematic diagram of primitive disturbance;

FIG. 2 is a schematic diagram of a watermark synchronization mechanism;

fig. 3 is a watermark image before and after scrambling, where (a) is the watermark image before scrambling and (b) is the watermark image after scrambling;

FIG. 4 is a watermark information embedding flow diagram;

FIG. 5 is a flow chart of watermark information detection;

figure 6 is a schematic view of RVT format experimental data, wherein (a) represents an office building, (b) represents a gym, and (c) represents a structural model;

FIG. 7 is a comparison of model details before and after watermark embedding;

FIG. 8 is a graph of translation attack results;

FIG. 9 is a schematic diagram of a primitive attack section data, wherein (a) represents a deletion ratio of 40%, (b) represents a deletion ratio of 60%, (c) represents an increase ratio of 60%;

FIG. 10 (a) is a schematic diagram of the result of a primitive deletion attack;

FIG. 10 (b) is a diagram showing the result of an element addition attack;

FIG. 11 is a diagram of a detailed level change attack result;

FIG. 12 (a) is a graph showing the result of a 100 m translation attack;

FIG. 12 (b) is a schematic diagram of the detailed level change attack result;

FIG. 12 (c) is a diagram showing the result of 20% attack by primitive deletion;

FIG. 12 (d) is a graph showing the 20% increase in the number of primitives;

FIG. 12 (e) is a diagram showing the result of deriving the FBX data;

Fig. 12 (f) is a schematic diagram of the result of exporting JSON data.

Detailed Description

The present invention will be described in detail below with reference to the drawings and detailed description to enable those skilled in the art to better understand the technical scheme of the present invention. Embodiments of the present invention will be described in further detail below with reference to the drawings and specific examples, but not by way of limitation. The order in which the steps are described herein by way of example should not be construed as limiting if there is no necessity for a relationship between each other, and it should be understood by those skilled in the art that the steps may be sequentially modified without disrupting the logic of each other so that the overall process is not realized.

The embodiment of the invention provides a BIM data robust watermarking method based on primitive disturbance and invisible characters, which has the key point of improving the capability of watermark attack resistance on the premise of keeping a model space topological relation. The method based on the local features has high flexibility and stronger robustness to attacks such as primitive deletion and the like, and BIM data local features are selected for watermark information embedding and extraction. Designing a related watermark scheme needs to solve two problems: (1) selecting local features and embedding watermark information; (2) synchronizing watermark information.

Because BIM data has graphic elements constrained by space and watermark embedding can not be realized through position movement, graphic element attributes are introduced, a watermark embedding method is designed for the non-movable graphic elements, and the number of watermark information carriers is increased. In addition, in order to enhance the robustness of the watermark, a watermark bit mapping relation is established by using the high position of the coordinates of the graphic element, so that the graphic element relation of each watermark carrier is independent. In summary, the local features are embedded by taking the spatial position and attribute information of the graphic primitive as watermark information, and watermark synchronization relation is established based on the coordinates of the graphic primitive. Watermark information detection is the inverse of watermark information embedding.

Several key techniques to which the proposed method of the present invention needs to be applied, including primitive perturbation, invisible characters, and a watermark synchronization mechanism based on a mapping mechanism, will be described below.

Primitive perturbation:

the primitive disturbance is a method for translating a movable primitive in a certain direction based on local characteristics of the spatial position of the primitive, the size of a disturbance range and the disturbance direction are controllable, and accurate change of coordinates can be realized. The BIM data coordinate precision reaches thirteen bits after decimal point, and the graphic primitive disturbance schematic diagram is shown in figure 1.

As shown in fig. 1, after a movable element such as a door, a window, a floor, etc. moves in a certain direction, watermark information can be embedded in the lower position of the movable element coordinate. The method has small influence on the graphic element and is not easy to cause model errors.

Different graphic elements in BIM data are positioned in different modes, and after model operations such as translation and the like, positioning points are possibly asynchronous with graphic element movement and have instability, so that coordinate points are used as watermark information carriers, and the graphic element space positioning modes are unified first. The coordinates of the center point of the bounding box are calculated according to the graphic primitive, the coordinate values of the center point of the bounding box change along with the change of the graphic primitive, and the coordinate values are mutually identicalFor stability, it can be considered as primitive spatial location coordinates. With the center point of the graphic elementFor example, the X coordinate of (c) is represented by formula (1).

（1）

After a unified graphic element space position positioning method is constructed, accurate embedding and detection of watermark information can be realized through graphic element disturbance. In summary, the spatial position of the movable primitive is selected as a local feature, and watermark information is embedded based on primitive disturbance.

Invisible characters:

the Unicode codes are internally provided with some format control special codes, the codes of invisible characters can express special information, the characters are embedded in the text under the conditions of being not easy to be perceived and not influencing reading, the Unicode codes can be used for watermark information steganography, the Unicode codes have concealment and content richness, the Unicode characters only occupy 2 bytes in size, and the increased data redundancy of the embedded invisible characters is small.

BIM data carries rich attribute information, invisible characters are embedded into the attributes of the non-movable primitives, the spatial topological relation of the primitives is not affected, the safety is high, and the number of watermark information carriers can be increased. In summary, attribute information of the non-movable primitives is selected as a local feature, and watermark information is embedded based on invisible characters.

Watermark synchronization mechanism based on mapping mechanism:

the synchronization mechanism is established by using the graphic element coordinates, the watermark bit mapping relation of each graphic element can be kept independent, the graphic element coordinate numerical range is expanded by using multiplication operation, the embedding position randomness of watermark information can be increased, and the information is more uniformly dispersed in BIM data. A schematic diagram of the watermark synchronization mechanism is shown in fig. 2.

Due to practical applicationThe coordinate precision of the BIM data in the middle is possibly reduced, and the mapping relation is constructed to select the decimal point of the primitive coordinate and then the high order is obtained. Watermark bits The calculation method is shown in the formula (2). （2）

The method enables the watermark mapping relation of each graphic element to be independent, and the deletion, movement, rotation and the like of the graphic element do not affect the mapping of other graphic elements to correct watermark bits, so that the method can effectively improve the robustness of the algorithm. In summary, the high order primitive coordinates are selected herein, and a watermark synchronization mechanism is constructed based on multiplication.

The following embodiments of the present invention will specifically describe a method for robust watermarking of BIM data based on primitive perturbation and invisible characters, which may include two steps of watermark information generation and watermark information embedding when implemented.

In the watermark information generation process, compared with nonsensical watermarks, the meaningful watermarks can more intuitively display experimental results, so that the meaningful watermarks are selected, the original watermarks are required to be preprocessed by considering that the meaningful watermarks have stronger correlation and are easy to break. Assuming a keyOriginal watermark sequence read by scanning +.>Defining scrambling function +.>As shown in formula (3).

（3）

L is a Logistic chaotic system, has no periodicity and is difficult to crack, and can be used for watermark decorrelation. When (when) The Logistic mapping is in a complete chaotic state, and the calculation formula is shown in formula (4).

（4）

In the middle of，/>And->By key->Setting. The original watermark image is shown in fig. 3 (a), scanned and read, modulated with a chaotic sequence, and finally the scrambled watermark sequence is obtained, as shown in fig. 3 (b).

The watermark information embedding flow is shown in fig. 4.

The watermark embedding step is as follows:

(1) Data preprocessing: firstly, considering addition and deletion of BIM data graphic elements, screening a main body and a structural element with relatively stable BIM data as a watermark information carrier. Calculating the primitive bounding box center point as in the above formula (1)Join set->Expressed as，/>，/>Is the total number of watermark carrier primitives. Secondly, considering the precision loss and the layering aggregation characteristic of the space position of the graphic primitive in the BIM using process, and selecting the high position after the decimal point of the XYZ coordinate, as in the above formula (2), synchronizing the watermark bit of the graphic primitive. Finally, the graphic element is subjected to mobile test, and the movable graphic element is added into the collection +.>Expressed as->Non-movable element center adding set +.>Expressed as->。

(2) Primitive perturbation: since most of the primitives of BIM data are limited by elevation, Z-axis direction disturbance may cause model generation errors, so that X-axis and Y-axis direction disturbance is performed on movable primitives. Traversing collections And embedding watermark information into the low positions of X coordinates and Y coordinates of the central point of the graphic primitive. Taking X coordinate as an example, the disturbance method is shown as a formula (5), when the calculated value of the coordinate setting position and the corresponding watermark information +.>And when the watermark information is not matched, the primitive is disturbed, so that the purpose of embedding the watermark information is realized.

（5）

In the middle ofRepresentation->Middle->Center point coordinates of each graphic element embedded with watermark information, < >>Embedding a position for watermark information->Representing the watermark value of the mapped bits.

(3) Invisible character embedding: the graphic element in BIM data carries multiple attribute information, and before Unicode invisible characters are embedded, the graphic element needs to be according to a secret keySetting invisible character code to be embedded +.>Attribute fields carrying watermark information with the primitives. When watermark is embedded, the set is traversed>Each non-movable picture element when the picture element corresponds to watermark bit information +.>1 and the primitive attribute field contains invisible characters +.>And (3) performing character embedding according to the formula (6), otherwise, performing no operation.

（6）

In some embodiments, the BIM data robust watermarking method based on primitive perturbation and invisible characters may further include watermark information detection, the watermark information detection flow is shown in FIG. 5.

The watermark information detection process is the inverse of watermark embedding. Firstly, preprocessing BIM data in the watermark information embedding process to obtain a movable element setAnd non-movable element set +.>Watermark information detection is then performed based on the BIM data local features, as follows:

(1) Movable element watermark detection: traversing collectionsPrimitive based on key->Setting watermark information carrying bit, extracting watermark information as shown in formula (7), and updating detected watermark value>And increases the number of times the watermark value is written at that location.

（7）

(2) Non-movable picture elementWatermark detection: traversing collectionsPrimitive based on key->Invisible character of the setting->Extracting watermark information from the primitive attribute field as in formula (8), and updating the detected watermark value>And increases the number of times the watermark value is written at that location.

（8）

(3) Obtaining a watermark: traversing all watermark bits according to a majority principle and updating watermark information as shown in formula (9) to obtain a watermark sequenceSubsequently +.>And (5) carrying out dimension lifting and reverse scrambling to obtain the BIM data watermark information image.

（9）

For the sequence->Position->Number of times of mapping into watermark->Indicate->Bit watermark information->. Since watermark bit watermark information is judged according to a majority principle, the change of the spatial position or attribute information of part of the graphic elements does not influence the correct extraction of the watermark information.

In order to evaluate the performance of the method provided by the invention, experiments and analyses are respectively carried out on imperceptibility, original data file size change and robustness. The experiment adopts mainstream BIM format RVT format data, and 3 RVT format models are selected respectively, wherein the model comprises a building model and a structural model, as shown in figure 6.

The basic information of the original BIM data is shown in table 1.

Table 1 basic information table of experimental data

Imperceptibility analysis:

(1) Qualitative analysis

In order to achieve the purpose of hiding the digital watermark information in BIM data, no obvious difference between the data before and after watermark embedding is required, namely, better imperceptibility is achieved. The detailed pair of BIM experimental data before and after watermark information embedding is shown in fig. 7.

From subjective visual observation in fig. 7, the buildings and structures before and after watermark embedding have no obvious difference, and the specific primitive attribute fields have no visual difference, so that the embedding of the digital watermark does not influence the visual quality of the original data, and the visual effect angle is considered to have good imperceptibility.

(2) Quantitative analysis

The watermark information imperceptibility is quantitatively analyzed by adopting Hausdorff (Hausdorff) distance, signal-to-noise ratio (SNR) and peak signal-to-noise ratio (PSNR). Hausdorff distance is a method for calculating the similarity between two point sets, and the smaller the value is, the higher the similarity between BIM data watermark embedding before and after, and the better the imperceptibility of the digital watermark is. The signal-to-noise ratio and the peak signal-to-noise ratio can be used as parameters for measuring the influence degree of the watermark-containing data on the original data, the original pixel space position coordinates and the coordinates after watermark information is embedded respectively consider normal signals and noise, the signal-to-noise ratio is the ratio of the normal signals to the background noise, and the peak signal-to-noise ratio is the ratio of the maximum signal intensity to the background noise. The larger the values of the signal-to-noise ratio and the peak signal-to-noise ratio, the smaller the influence of watermark information on the original data, and the better the imperceptibility.

The experimental results of Hausdorff (Hausdorff) distance, signal-to-noise ratio (SNR), peak signal-to-noise ratio (PSNR) are shown in Table 2.

TABLE 2 quantitative imperceptibility analysis results

According to GB/T14912-2017 1:500 1:1000 1:2000 field digital mapping protocol, 1: position point location error of ground object point plane under 500 scale + -0.30 m,1: position point location error of ground object point plane under 1000 scale + -0.60 m,1: as can be seen from Table 2, after watermark embedding, the maximum Hausdoff distance in all BIM data is 0.000224m, which is far lower than the position point location error of the ground object point plane, so that the disturbance range of the algorithm graphic primitive is small. In addition, the SNR and PSNR of the experimental data are higher than 100dB, and the BIM data similarity before and after watermark information embedding can be considered to be extremely high. In conclusion, the watermark has small influence on the original BIM data and has excellent imperceptibility.

And (3) analyzing the size change of the file after watermark embedding:

the invisible characters are invisible from the visual angle, but because additional characters are written in BIM data attribute information, the watermark information is embedded to cause the expansion of the file size, the original data generates information redundancy, and the redundancy is too large to reduce the practicality of the model, so that the file size change needs to be compared. For quantitative evaluation, the ratio of the redundant size to the original data size was evaluated as a quantization index, and the results are shown in table 3.

TABLE 3 incremental redundancy analysis results

As can be seen from Table 3, the redundancy increment ratio of the building model office building to the gym is far lower than 1%, because the movable graphic elements of the model are high in proportion, the invisible characters are embedded little, the graphic elements of the structural model are limited by the upper and lower elevations and the adjacent graphic elements, the proportion of the invisible graphic elements is high, the invisible characters are embedded much, and the redundancy increment ratio is relatively high, namely 0.09%. In general, the increment ratio of the three BIM experimental data is smaller than 1%, so that the BIM data redundant information after watermark embedding is considered to be less increased, and the influence on the file size change is small.

Robustness analysis:

to verify robustness, watermark sequences will be detectedIs->In contrast, the normalized correlation coefficient (NC) is used as a quantization index for correctly extracting the watermark. When NC is 1, it means that the correct watermark is completely extracted, and when NC is 0, it means that the watermark is not extracted at all. And setting a threshold value of 0.75, and considering watermark extraction to be successful when the NC coefficient is larger than the threshold value.

Watermark detection was performed without attacking the BIM data, and the detection results are shown in table 4.

Table 4 original watermark information detection results

As shown in table 4, the NC coefficients of the three BIM data for original watermark information detection are all 1, and it is known that watermark information can be correctly embedded into a carrier after early stage primitive screening and embedding condition judgment, and can be completely extracted through a blind detection process.

Translation is a common operation of BIM data, and in actual use, data is often subject to two kinds of translation attacks: (1) In the BIM+GIS application scene, BIM data translate along the whole map along with the project base points, and the mode does not influence the project coordinate system, so that watermark extraction is not influenced; (2) The whole translation of the project model relative to the project base point may cause automatic deletion of partial primitives, etc., resulting in noise generation of watermark images. Therefore, to verify the robustness of the algorithm against the translational attack, the translational attack experiment is designed at intervals of 100 meters, and the experimental results are shown in table 5 and fig. 8.

Table 5 translational attack experiments

As can be seen from table 5 and fig. 8, after the three BIM data translate along the X-axis, Y-axis and Z-axis, the NC coefficients of the extracted watermark information are all higher than 0.95, but since each watermark bit corresponds to a plurality of watermark information carriers, watermark extraction is less affected, and the NC coefficients of the final watermark extraction are still far higher than the set threshold. In summary, the design approach herein is robust to BIM data X, Y, Y axis translation attacks.

In practical application operation of BIM data, deletion and addition of elements are common attack modes. The watermark information embedding process of each element is mutually independent, so that the deletion or addition of certain elements does not basically influence the detection of watermark information of other watermark carriers. In addition, watermark extraction means are based on voting mechanisms, and therefore, variations in a small fraction of the primitives may not affect the final extraction of the watermark sequence. However, the BIM data has an automatic update mechanism, and deletion and addition of elements may cause mobility or change of spatial positions of adjacent elements, ultimately resulting in erroneous extraction of watermark information. To verify the ability of the algorithm to resist element deletion attacks, experiments were performed herein at 10% intervals on the model, which underwent element deletion and addition, as shown in fig. 9. Table 6, fig. 10 (a), and fig. 10 (b) show experimental results at different pixel deletion rates.

TABLE 6 experimental results at different pixel erasure rates

As can be seen from the data analysis of table 6, element deletion and addition within 10% have no effect on watermark information extraction of the three models; watermark extraction of office buildings is not affected by 20% and 30% attack, watermark information of gymnasium and structural model has a small part of noise, but still can completely represent copyright information; the number of elements of the office building model is greater, so that watermark information can still be completely extracted when the office building model is subjected to 40% and 50% element deletion. The total number of elements of the gymnasium model and the structural model is less, the NC value is higher than 0.90 and lower than or equal to 0.98, and compared with watermark information of an office building, the noise is larger; after 60% of the elements are deleted, the NC values of the three BIM models are not 1, but the NC values of the structure model are higher than 0.90 under the attack of element addition. Overall analysis shows that as the element deletion and addition intensity increases, the quality of watermark information extracted from BIM data gradually decreases, but the NC value of the watermark is equal to or greater than 0.89, which is far higher than the watermark information display threshold of 0.75. In conclusion, the method and the device have strong robustness to the primitive attack, and can meet the daily use requirement.

BIM data primitives comprise point-line-plane information, BIM data in the same view can influence the display detail form of the point-line-plane of the primitives to display different geometric contents of the building structure. The detailed level of BIM data may vary for different application scenarios, possibly resulting in watermark information extraction errors. To verify the robustness of the algorithm to the detailed level change attack, three detailed level attack experiments of detailed, medium and rough level are performed on three BIM data, and the experimental results are shown in table 7 and fig. 11.

TABLE 7 detail level of variation experiment

From table 7, it can be seen that the watermark information can be completely extracted from the BIM data from detailed level to rough level, and that the watermark information extraction is not affected by different level setting conditions. The method is characterized in that the algorithm constructs a watermark bit mapping relation by using the central point of the primitive boundary box, watermark information is carried by using spatial position and attribute information, and the detail change only affects the primitive form details, so that the watermark bit of the primitive and the watermark carrying information are not affected. In summary, the method provided by the invention can resist detailed degree change attacks.

In practical application, in order to realize access and use of different technical platforms to BIM data, digital-analog separation is often used to analyze BIM data, and classification management is performed on geometric data and attribute data, at this time, watermark information carried by two local features of spatial position and attribute information is stored separately, which may affect watermark information extraction. Therefore, an experiment is required to be carried out on the robustness of the algorithm against the digital-analog separation attack. For this purpose, the RVT format BIM data is converted into FBX geometry data and JSON attribute data, and watermark information extraction is performed respectively, and experimental results are shown in table 8.

TABLE 8 digital-analog separation experiments

/>

As can be seen from table 8, the decomposition of the original RVT data into geometric and attribute data has no influence on the watermark extraction of the office building model, and has a slight influence on the watermark extraction of the gymnasium model and the structural model, but the NC coefficients of the watermark information NC coefficients of the two models are higher than 0.98 and far higher than the set threshold value of 0.75, so that the watermark image can be considered to be complete. In summary, the BIM data digital watermarking algorithm herein innovatively realizes robustness of digital-analog separation attack.

The experimental result shows that the watermark method based on graphic element disturbance and invisible characters provided by the invention has strong robustness, can realize a blind detection process, and effectively solves the difficulty of applying the existing three-dimensional robust watermark to BIM data. For a better understanding of the features of the proposed method, a further discussion is given below in terms of three aspects.

Analysis of watermark synchronization mechanism:

the watermark method establishes a watermark multi-pair synchronization mechanism based on element coordinates, so that each bit of watermark information is embedded into BIM data for multiple times. The redundancy enables watermark extraction to vote based on majority principle, when elements in BIM data are subject to deletion and addition attacks, the possibility of errors caused by element changes can be reduced, and the probability of extracting complete correct watermark information is greatly improved. This view was verified in element deletion and element addition experiments. The elements of BIM data are relatively aggregated at each layer, so the watermark synchronization mechanism described herein is optimized by the principle of multiplication to more uniformly embed watermark information into the data.

In addition, the central point of the surrounding frame is used as the coordinate of the primitive, and the calculation method is suitable for the primitives positioned according to the points, the lines and the planes, unifies the spatial positioning of BIM data and improves the stability of watermark bit mapping.

Analysis of applicability of the method

In the watermarking method proposed herein, the embedding of watermark information is based on mobility analysis, the movable element performs a small-scale disturbance, and the watermark information cannot be carried in this way, so the watermark information is embedded in the form of invisible characters. The method can be used for all elements in the model, namely main elements such as walls and the like, elements such as doors and windows and the like, so that the method can be applicable to BIM data of any type, such as building model office buildings and structural models. In summary, the method has an extremely broad range of applications. It should be noted that, since the coordinate bit mapping and watermark information embedding of the movable element involve coordinate value operations, when the BIM data rotates and the X, Y coordinates of all elements in the model change, the correct mapping of the watermark bits may be affected, especially the BIM data with a higher proportion of the movable elements may be affected. Thus, this approach has found widespread use in practice, but there are still some directions in the future to be investigated further, including enhancement of rotation robustness, reduction of redundancy, and combination with other watermarking algorithms.

Analysis of watermark capacity

Watermark capacity refers to the maximum amount of watermark that can be embedded in data. Herein, coordinates and attributes of elements are selected as embedding targets. The movable element can carry 2 bits of watermark information, and the immovable element can carry at least 1 bit of watermark information, so that the watermark capacity of the theoretically proposed method is larger than the total element number of BIM data. However, some unstable elements are not suitable as watermark carriers, and it is necessary to ensure that a plurality of elements correspond to one watermark bit, so the watermark capacity is generally smaller than the total number of elements. In this section, the relation between watermark length and robustness in the proposed watermarking algorithm is discussed. With the BIM data as the experimental data, which were subjected to the experiments as described above, model panning, level of detail change, graphic element and digital model separation attack experiments were performed with watermark lengths of 32, 64, 128, 256 and 512, and the experimental results are shown in table 9 and fig. 12 (a) -12 (f).

Table 9 results of attack experiments with different watermark lengths

In summary, for the requirements of BIM data copyright protection, the data characteristics of the BIM common format RVT are analyzed, and for the characteristics of small BIM data redundancy, strict topological relation of the graphic elements and the like, a digital watermark algorithm is provided based on the local characteristics of the model, the disturbance of movable graphic elements and the invisible character embedding of the immovable graphic elements are combined, the number of watermark carriers is expanded, and the multiple embedding of digital watermark information is realized. Experiments show that the algorithm has good imperceptibility, the increased redundancy of embedded invisible characters is small, the algorithm has stronger robustness to translation, primitive deletion and detail degree change attack, and can resist digital-analog separation attack. Meanwhile, the watermark detection process is blind detection, and the participation of the original BIM data is not needed, so that the watermark detection method has strong practicability. The invention can provide a feasible solution for BIM data copyright protection.

Furthermore, although exemplary embodiments have been described herein, the scope thereof includes any and all embodiments having equivalent elements, modifications, omissions, combinations (e.g., of the various embodiments across), adaptations or alterations as pertains to the present application. The elements in the claims are to be construed broadly based on the language employed in the claims and are not limited to examples described in the present specification or during the practice of the application, which examples are to be construed as non-exclusive. It is intended, therefore, that the specification and examples be considered as exemplary only, with a true scope and spirit being indicated by the following claims and their full scope of equivalents.

The above description is intended to be illustrative and not restrictive. For example, the above-described examples (or one or more aspects thereof) may be used in combination with each other. For example, other embodiments may be used by those of ordinary skill in the art upon reading the above description. In addition, in the above detailed description, various features may be grouped together to streamline the application. This is not to be interpreted as an intention that the features of the claimed application are essential to any of the claims. Rather, inventive subject matter may lie in less than all features of a particular inventive embodiment. Thus, the following claims are hereby incorporated into the detailed description as examples or embodiments, with each claim standing on its own as a separate embodiment, and it is contemplated that these embodiments may be combined with one another in various combinations or permutations. The scope of the application should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. A BIM data robust watermarking method, the method comprising:

watermark information generation:

obtaining a keyOriginal watermark sequence read by scanning +.>As shown in formula (3):

（3）

wherein the method comprises the steps ofL is Logistic chaotic system for watermark decorrelation, whenWhen the Logistic mapping is in a complete chaotic state, the calculation formula is shown as formula (4):

（4）

in the middle of，/>、/>And +.>Respectively represent the 1 st part of the chaotic system sequence,n、nValue of +1 bit, +_1>Representing control parameters->And->By key->Setting;

watermark embedding:

Traversing collectionsEmbedding watermark information into the low position of X coordinate and Y coordinate of the central point of the graphic primitive; taking X coordinate as an example, the disturbance method is shown as a formula (5), when the calculated value of the coordinate setting position and the corresponding watermark information +. >When the two primitives are not matched, the corresponding primitive is disturbed to realize watermark information embedding;

（5）

in the middle ofRepresentation->Middle->Center point coordinates of each graphic element embedded with watermark information, < >>Embedding a position for watermark information->Representation->Middle->Center point coordinates before watermark information is embedded into each graphic element, < >>A watermark value representing the mapped bits;

（6）

2. The method of claim 1, wherein the primitive bounding box center point is calculated by the formulaCoordinates of (c):

，

in the method, in the process of the invention,representing the center point of the bounding box of the graphic element>Is>Represents the bounding box abscissa maximum, +.>Representing the minimum value of the abscissa of the bounding box, +.>Representing the center point of the bounding box of the graphic element>Ordinate of>Representing the maximum value of the ordinate of the bounding box,/ >Representing the minimum value of the ordinate of the bounding box, +.>Representing the center point of the bounding box of the graphic element>Vertical coordinates of>Representing bounding box vertical coordinate maximum,/-)>Representing bounding box vertical coordinate minima.

3. The method of claim 1, wherein the watermark bits are calculated by the formula：

（2）

4. The method of claim 1, further comprising watermark information detection, the watermark information detection comprising:

BIM data is acquired, and is preprocessed to obtain a movable element setWith non-movable element sets；

（7）

traversing a set of non-movable primitivesPrimitive based on key->Invisible character of the setting->Extracting watermark information from the primitive attribute field as in formula (8), and updating the detected watermark value >And increasing the number of times of writing watermark values in the corresponding positions;

（8）

（9）

5. A BIM data robust watermarking method apparatus, the apparatus comprising:

（3）

（4）

a watermark embedding module configured to embed a watermark:

based on BIM data primitive addition and deletion, the main body and the structural diagram element with stable BIM data are screened as watermark information carriers, and the central point of a primitive bounding box is calculated Join set->Expressed as->，/>，/>The total number of the watermark carrier primitives; selecting the high position of the decimal point of the XYZ coordinates, synchronizing the watermark bit of the graphic primitive, performing movement test on the graphic primitive, and adding the movable graphic primitive into the set +.>Non-movable element center adding set +.>；

Traversing collectionsEmbedding watermark information into the low position of X coordinate and Y coordinate of the central point of the graphic primitive; taking X coordinate as an example, the disturbance method is shown as a formula (5), when the calculated value of the coordinate setting position and the corresponding watermark information +.>When the two primitives are not matched, the corresponding primitive is disturbed to realize watermark information embedding;

（5）

（6）

6. The apparatus of claim 5, wherein the watermark information generation module is further configured to calculate a primitive bounding box center point byCoordinates of (c):

，

7. The apparatus of claim 5, wherein the watermark information generation module is further configured to calculate watermark bits by the following formula：

（2）

8. The apparatus of claim 5, further comprising a watermark information detection module configured to:

（7）

traversing a set of non-movable primitivesPrimitive based on key->Invisible character of the setting->Extracting watermark information from the primitive attribute field as in formula (8), and updating the detected watermark value>And increase corresponding location writingThe number of watermark values;

（8）

（9）

9. A readable storage medium storing one or more programs executable by one or more processors to implement the method of any of claims 1-4.