Cultural relic digital point cloud knowledge base construction method based on style features

Abstract

The embodiment of the invention provides a cultural relic digital point cloud knowledge base construction method based on style characteristics, and belongs to the technical field of image processing, and the method specifically comprises the steps: analyzing the geometric composition of a target cultural relic, obtaining a three-dimensional basic shape, and generating and storing a cultural relic shape basic unit representing the characteristics of a cultural relic part through spatial distribution operation; segmenting the texture of the target cultural relic, and generating and storing a size-normalized cultural relic texture basic unit according to the integrity, glossiness, size and shape attributes of the texture; configuring multi-dimensional feature tags for the cultural relic shape basic units and the cultural relic texture basic units; and constructing a spatial association relationship feature library between the cultural relic shape basic units and the cultural relic texture basic units on the basis of the association of style features, the adaptability of shapes and textures and a mapping relationship learned through a deep learning model, thereby forming a cultural relic digital point cloud knowledge base. Through the scheme of the invention, the accuracy and adaptability of knowledge base construction are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to a method for constructing a digital point cloud knowledge base of cultural relics based on style features. BACKGROUND

[0002] At present, with the development and popularization of three-dimensional scanning technology, point clouds are gradually applied to the digital reconstruction of cultural relics due to their high acquisition accuracy and the advantage of not needing to contact objects. Meanwhile, deep learning methods are also widely used in the field of rapid reconstruction of ancient cultural relics. A reasonably designed and organized point cloud knowledge base is crucial for using deep learning to rapidly reconstruct ancient cultural relics.

[0003] Cultural relics of different periods have different style features due to differences in history, culture, and living environment. These features are not only reflected in the geometric shape of the cultural relics, but also in the surface texture information. However, existing methods have not addressed the understanding of the inherent style features of cultural relics, the correlation between style features and geometric shape and texture information, nor have they constructed a model object knowledge base targeting the style features of cultural relics.

[0004] Therefore, there is an urgent need for a method for constructing a digital point cloud knowledge base of cultural relics based on style features, which has efficient storage, fast retrieval, and multiple dimensions, and can provide a foundation for subsequent deep learning-based three-dimensional point cloud reconstruction of cultural relics. SUMMARY

[0005] Therefore, the embodiments of the present application provide a method for constructing a digital point cloud knowledge base of cultural relics based on style features, which at least partially solves some of the problems existing in the prior art.

[0006] The embodiments of the present application provide a method for constructing a digital point cloud knowledge base of cultural relics based on style features, which includes:

[0007] Step 1: Analyze the geometric composition of the target cultural relic to obtain a three-dimensional basic shape and generate and store a cultural relic shape basic unit representing the features of the cultural relic parts through spatial distribution operation;

[0008] Step 2: Segment the texture of the target cultural relic, generate and store size-normalized cultural relic texture basic units according to the integrity, glossiness, size, and shape attributes of the texture;

[0009] Step 3: Configure multi-dimensional feature labels for the cultural relic shape basic unit and the cultural relic texture basic unit, wherein the types of the feature labels include label type, digital type, character type, time type, and enumeration type;

[0010] Step 4, based on the relevance of style features, the adaptability of shape and texture, and the mapping relationship learned by the deep learning model, a spatial correlation relationship feature library between the basic shape unit of cultural relics and the basic texture unit of cultural relics is constructed, and accordingly a digital point cloud knowledge base of cultural relics is formed.

[0011] According to a specific implementation manner of the embodiment of the application, the spatial distribution operation includes Boolean operation, deformation operation and combination operation, wherein the Boolean operation includes set union, set difference or set intersection operation on the plurality of three-dimensional basic shapes, the deformation operation includes extrusion, scaling, moving or rotating operation on the three-dimensional basic shape or the intermediate shape after the Boolean operation, and the combination operation includes recombination of the basic shape unit of cultural relics after the Boolean operation and / or the deformation operation.

[0012] According to a specific implementation manner of the embodiment of the application, the size normalization includes:

[0013] The length and width dimensions of the segmented texture are uniformly adjusted to 10 pixels or an integer multiple of 10 pixels, and pixel completion is performed on the texture with insufficient size.

[0014] According to a specific implementation manner of the embodiment of the application, the data structure of the feature label contains a feature type field, a basic unit field, a display unit field, a display value field, a first numerical field for storing a numerical starting value or a specific value based on the basic unit, and a second numerical field for storing an ending value of the numerical type, and when feature retrieval is performed, the values of the first numerical field and the second numerical field are directly compared without unit conversion.

[0015] According to a specific implementation manner of the embodiment of the application, when the feature label is of a numerical type, storage and retrieval of single values or range values are supported, and when the feature label is of a time type, storage and retrieval of specific time points or time ranges are supported.

[0016] According to a specific implementation manner of the embodiment of the application, the feature label dimension configured for the basic shape unit of cultural relics includes at least one of an era feature, a regional feature, a cultural feature, a morphological feature, a process feature and a material feature.

[0017] The feature label dimension configured for the basic texture unit of cultural relics includes at least one of an era feature, a regional feature, a cultural feature, a morphological feature, a use feature, a glossiness feature, a shape feature and a size feature.

[0018] According to a specific implementation manner of the embodiment of the application, the manner of establishing the spatial correlation relationship includes at least one of the following:

[0019] The association is based on the common era characteristics, regional characteristics and / or cultural characteristics between the basic units of the artifact shape and the basic units of the artifact texture;

[0020] Based on the shape and / or use of the basic unit of the artifact shape, bind and adapt the basic unit of the artifact texture;

[0021] By using a deep learning model, the association mapping relationship between the basic units of the shape of cultural relics and the basic units of the texture of cultural relics is trained and established, so as to automatically update and expand the knowledge base.

[0022] According to one specific implementation of the present invention, the three-dimensional basic shapes include cubes, spheres, cylinders, cones, tori, and planes.

[0023] The scheme for constructing a digital point cloud knowledge base for cultural relics based on style features in this embodiment of the invention includes: Step 1, analyzing the geometric composition of the target cultural relic, obtaining a three-dimensional basic shape, and generating and storing basic units of cultural relic shape representing the features of cultural relic components through spatial distribution calculation; Step 2, segmenting the texture of the target cultural relic, and generating and storing basic units of cultural relic texture with normalized dimensions based on the integrity, gloss, size, and shape attributes of the texture; Step 3, configuring multi-dimensional feature labels for the basic units of cultural relic shape and the basic units of cultural relic texture, wherein the types of feature labels include tag type, number type, character type, time type, and enumeration type; Step 4, constructing a feature library of spatial association relationships between basic units of cultural relic shape and basic units of cultural relic texture based on the correlation of style features, the adaptability of shape and texture, and the mapping relationship learned through a deep learning model, thereby forming a digital point cloud knowledge base for cultural relics.

[0024] The beneficial effects of this invention are as follows: By using the solution of this invention, various cultural relic shapes are abstracted using three-dimensional basic shapes and basic cultural relic shape units, and various features of the basic cultural relic shape units and basic cultural relic texture units are expressed through feature tags. This allows for a comprehensive expression of the basic cultural relic shape units and basic cultural relic texture units, and these feature tags can be dynamically managed. By constructing the association relationship between the basic cultural relic shape units and basic cultural relic texture units, the correlation between the basic cultural relic shape units and basic cultural relic texture units is established, thereby establishing a multi-dimensional knowledge base model that can be applied to the training of deep learning models for cultural relic reconstruction, thus improving the accuracy and adaptability of knowledge base construction. Attached Figure Description

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings described below only illustrate some of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0026] Figure 1 A flowchart of a method for constructing a digital relic point cloud knowledge base based on style features provided by the embodiments of the present application is shown in the figure.

[0027] Figure 2 A schematic diagram of the constitutive relationship between three-dimensional basic shapes, basic units of relic shapes and relics provided by the embodiments of the present application is shown in the figure.

[0028] Figure 3 A schematic diagram of a real relic provided by the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0029] The embodiments of the present application will be described in detail below with reference to the drawings.

[0030] The embodiments of the present application will be described in detail below with reference to the drawings.

[0031] It should be noted that the various aspects of the embodiments described below are within the scope of the appended claims. It should be apparent that the aspects described herein can be embodied in a wide variety of forms and that any specific structure and / or function described herein is merely illustrative. Based on the teachings provided herein one skilled in the art should be able to contemplate these and similar aspects of the present application. For example, acts recited as being performed in a certain order can be performed in a different order. Additionally, acts recited as being performed at the same time can be performed at different times. Further, any of the aspects described herein can be implemented using any of a variety of technologies.

[0032] It should be noted that the drawings provided in the following embodiments only illustrate the basic concept of the present application in a schematic manner, and only show the components related to the present application in the drawings, not drawn according to the number, shape and size of the components in actual implementation. The actual implementation of each component type, number and proportion can be a random change, and the component layout type can be more complex.

[0033] In addition, in the following description, specific details are provided in order to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the aspects described can be practiced without these specific details.

[0034] The embodiment of the present application provides a kind of based on style feature's cultural relic digitization point cloud knowledge base construction method, the method can be applied to the point cloud knowledge base construction process of three-dimensional scanning scene.

[0035] Referring to Figure 1 A kind of based on style feature's cultural relic digitization point cloud knowledge base construction method provided by the embodiment of the present application is shown in the flow chart of Figure. As Figure 1 The method mainly includes the following steps:

[0036] Step 1, analyze the geometric structure of the target cultural relic, obtain three-dimensional basic shape and generate and store cultural relic shape basic unit representing cultural relic component characteristics by space distribution operation;

[0037] In specific implementation, the three-dimensional basic shape suitable for the shape of cultural relic object can be extracted by analyzing the characteristics of the shape of cultural relic object, mainly including cube, sphere, cylinder, cone, ring and plane, as shown in Table 1.

[0038] Table 1

[0039]

[0040] In order to facilitate the data source for future deep learning training, we further form the combination of basic shape units with cultural relic characteristics, i.e. cultural relic shape basic unit, by Boolean, deformation and combination space distribution operation on these three-dimensional basic shapes.

[0041] The main methods of three-dimensional basic shape space distribution operation include Boolean operation, deformation operation and combination operation:

[0042] Boolean operation: Boolean operation operates multiple three-dimensional basic shapes through set operation, mainly including set operation, difference set operation and intersection operation, wherein the set operation is to combine two or more three-dimensional basic shapes into one cultural relic shape basic unit. For example, multiple cubes are combined into a complex base. The difference set operation will "dig" a part from another three-dimensional basic shape with one three-dimensional basic shape, for example, a small ball is used to dig a pit on a large ball to make an eye model. The intersection operation only retains the overlapping part of two three-dimensional basic shapes.

[0043] Deformation operation: After Boolean operation, it is often necessary to continue to adjust the cultural relic shape basic unit from point, line and surface, which is deformation operation. Main deformation operations include extrusion, scaling, moving and rotating, wherein the extrusion is to select a surface and pull it out or push it in to create a new geometric body, and the scaling, moving and rotating are operations for point, line and surface to change the shape. For example, the top point of a sphere is scaled to become a water cup cover.

[0044] Combination operation: The combination operation is to combine the cultural relic shape basic units after Boolean operation and deformation operation again to quickly build more vivid and complex cultural relic shape basic units, which is an important means to quickly expand the cultural relic shape basic unit library.

[0045] The constitutive relationship among the three-dimensional basic shape, the cultural relic shape basic unit and the cultural relic is shown in Figure 2

[0046] Step 2, the texture of the target cultural relic is segmented, and the size normalized cultural relic texture basic unit is generated and stored according to the integrity, glossiness, size and shape attributes of the texture;

[0047] In specific implementation, the texture can be segmented, and the segmentation is performed according to the integrity, glossiness, size and shape of the texture. In order to quickly adapt to the cultural relic shape basic unit, the length and width sizes of the texture are both set to 10px or an integer multiple of 10px, and the insufficient part needs to be pixel completed. For example, if a texture is segmented to be 8px, the texture is pixel completed by 2px to make the length and width 10px, and if a texture is segmented to be 12px, the texture is further pixel completed to meet the integer multiple of 10px, and the completion is 20px.

[0048] Step 3, multi-dimensional feature labels are configured for the cultural relic shape basic unit and the cultural relic texture basic unit, wherein the types of the feature labels include label type, digital type, character type, time type and enumeration type;

[0049] ​In specific implementation, considering the need to facilitate the rapid retrieval of cultural relic shape basic units and cultural relic texture basic units from the style feature library, and in order to express the features of cultural relic shape basic units and cultural relic texture basic units in a complete and concise manner, the feature tag library can also be quickly applied to deep learning training, the present application establishes a style feature tag library from multiple dimensions.

[0050] Hereinafter, the cultural relic shape basic unit and the cultural relic texture basic unit are referred to as entities.

[0051] In the present application, the feature tags are classified into five types according to type, namely, tag type, numerical type, character type, time type and enumeration type.

[0052] The tag type indicates that the feature is a text tag, which can be directly marked on the entity without setting a specific value, and has a marking property, such as the tags: porcelain and Tang Dynasty. When the tags "porcelain" and "Tang Dynasty" are marked on the cultural relic shape basic unit, it can be expressed that the cultural relic shape basic unit is a "porcelain" of "Tang Dynasty", and the data structure is shown in Table 2.

[0053] Table 2

[0054]

[0055] The numerical type indicates that the feature can be represented by numbers, including integers, floating points and all types of comparable sizes, which can be specific numbers or a range of numbers, such as weight, volume, etc. At the same time, a basic unit can be set for the numerical type feature, and other units can be inferred from the conversion relationship between the basic unit and the specific storage value, such as weight, which is a numerical type feature, and the basic unit is set as kilogram, and other units including gram, ton, milligram, etc. The conversion relationship is:

[0056] 1 kilogram = 1000 grams

[0057] 1 kilogram = 0.001 tons

[0058] 1 kilogram = 1000 1000 milligrams

[0059] Through the conversion design of the basic unit and other units, the same feature can be expressed more readably by combining "number + unit" to express the features of cultural relics, such as the weight feature, which can be expressed by the unit "ton" for the bronze chariot, and the bronze chariot weighs 1 ton. For gold jewelry, it can be expressed by the unit "gram", such as the weight of a gold bracelet, which is 30 grams.

[0060] Meanwhile, the digital type data type designed by the application supports the storage of range values, such as the belonging age of cultural relics. This feature supports the storage of specific numbers and range numbers when storing, such as the belonging age of a certain porcelain being 1300 years, and the belonging age of another porcelain being a rough range. At this time, the range number should be used to express, such as (1300, 1400) years, wherein the first number is the start value of the range, and the second number is the end value of the range. This setting is for future convenient retrieval.

[0061] The digital type feature data structure is shown in Table 3.

[0062] Table 3

[0063]

[0064] The character type represents that the feature can set a specific character for an entity, such as "name: Tang Sancai", wherein "name" is the feature name of the character type, and "Tang Sancai" is the value set when the feature mark is on a certain entity. The data structure is shown in Table 4.

[0065] Table 4

[0066]

[0067] The time type represents that the feature has a time attribute, such as the unearthed date, the storage date, etc. The time type also supports time range, such as the attribute with the start date and the end date of the validity period, such as the validity period (2024-10-5, 2025-10-5). The first time is the start value of the range, and the second time is the end value of the range. This setting is for future convenient retrieval. The time type feature supports the date type and the time type, wherein the time type has date, time, minute and second time information; and the date type only has date without time, minute and second information. The data structure is shown in Table 5.

[0068] Table 5

[0069]

[0070] The enumeration type represents that the range of the feature value is limited, and the feature value should be selected in the specified range when setting. It is usually used to represent a group of related options, states or marks, such as the type of porcelain and the firing method. The data structure is shown in Table 6.

[0071] Table 6

[0072]

[0073] Based on the above five kinds of feature types, various feature labels can be created, such as the unearthed age, the type of cultural relics, the name, the value, etc. The features can also be dynamically created according to actual needs.

[0074] The feature type main data structure is as follows:

[0075] type: the type of the feature, including label type, number type, character type, time type and enumeration type, represented by 0, 1, 2, 3 and 4 respectively;

[0076] b_unit: the basic unit of the feature, such as kilogram;

[0077] s_unit: the display unit of the feature attribute with readability, such as the weight display unit “ton” of the stone statue;

[0078] s_value: the display value of the feature with readability, such as the weight value 1 of the stone statue, representing 1 ton, for the label type feature, this place stores the specific label, for the character type feature, this place stores the value of the character type feature, such as the value “porcelain” of the name feature, the unearthed date of the time type feature, this place stores the date, the enumeration type feature, this place stores the specific enumeration value;

[0079] start_val_num: stores the starting value or specific value of the number type feature, such as the specific value or starting value of the artifact belonging to the era;

[0080] end_val_num: stores the end value of the number type, such as the end value of the artifact belonging to the era;

[0081] start_val_str: stores the starting value or specific value other than the number type;

[0082] end_val_str: stores the end value other than the number type;

[0083] shape: associates the basic unit of the artifact shape;

[0084] texture: associates the basic unit of the artifact texture.

[0085] The s_value feature attribute is the value displayed on the system end. When the same feature attribute has different units, the value of s_value represents different feature values. For example, when the feature attribute weight s_value = 5, the unit can be kilograms and tons, and the numerical difference is huge, which brings great challenges to the retrieval of feature attributes; when retrieving cultural relics with a weight of 4 to 500 kilograms, the s_value in the database needs to be converted to data values in kilograms, and then compared, which greatly affects the retrieval efficiency. In order to solve the problem of unit conversion during retrieval, start_val_num and end_val_num store data values based on the basic unit. If the feature attribute is a single value, the data value is only saved in start_val_num, and if the feature attribute is a range value, the start value of the data value is saved in start_val_num, and the end value is saved in end_val_num. In this way, during retrieval, only start_val_num and end_val_num are directly compared, without the need for unit conversion, and indexes are established for start_val_num and end_val_num, further speeding up the retrieval speed. For non-numeric feature values, their comparison data based on the basic unit are saved in start_val_str and end_val_str, and fuzzy queries are performed on strings, and the data structure is shown in Table 7.

[0086] Table 7

[0087]

[0088] Secondly, in order to facilitate deep learning training, the application classifies the shape basic unit feature style of cultural relics from the dimension of feature style category, mainly including time feature, regional feature, cultural feature, morphological feature, process feature and material feature. Each feature style includes many specific feature items, as shown in Table 8.

[0089] Table 8

[0090]

[0091] At the same time, the feature style of the texture basic unit of cultural relics is classified from the dimension of feature style category, mainly including time feature, regional feature, cultural feature, morphological feature, use feature, glossiness feature and size feature. Each feature style includes many specific feature items, as shown in Table 9. It can be seen that the feature style classification of the shape basic unit of cultural relics and the texture basic unit of cultural relics has common places and differences.

[0092] Table 9

[0093]

[0094] The feature library built based on the above method provides training data for deep learning models, and provides a dataset foundation for improving the recognition and reconstruction of cultural relics.

[0095] Step 4: Based on the correlation of style features, the adaptability of shape and texture, and the mapping relationship learned through deep learning models, construct a feature library of spatial correlation between basic units of cultural relic shape and basic units of cultural relic texture, thereby forming a digital point cloud knowledge base for cultural relics.

[0096] In practice, the final step is to construct a feature library of spatial relationships between basic shape units and basic texture units of cultural relics:

[0097] Establish one-to-one, one-to-many, and many-to-many spatial relationships between basic units of cultural relic shape and basic units of cultural relic texture, construct style expression models of cultural relics from different periods, and form a complex style feature library with a multi-dimensional data structure.

[0098] This invention establishes the spatial correlation between the basic units of artifact shape and the basic units of artifact texture through three methods:

[0099] Relationships are established based on the stylistic relevance of the basic shape and texture units of cultural relics. Specifically, spatial relationships are established between basic shape and texture units that share the same stylistic characteristics, including those related to the era, region, and culture. For example, basic shape and texture units of cultural relics with characteristics of the Tang Dynasty can be spatially related, but basic shape units of cultural relics from the Shang and Zhou Dynasties and basic texture units of cultural relics from the Qing and Ming Dynasties generally do not.

[0100] Based on the basic characteristics of the cultural relic's shape and purpose, basic texture units are bound to it. For example, the following basic texture units are bound to the basic shape unit of a porcelain artifact: glossiness is smooth, size is basic size, and shape is landscape.

[0101] Based on Method 1 and Method 2, a model is established by training the spatial relationship between the basic units of the shape and the basic units of the texture of cultural relics through deep learning, so that the style feature knowledge base can be automatically upgraded and expanded.

[0102] The method for constructing a digital relic point cloud knowledge base based on style features provided by the embodiment abstracts various relic shapes by using three-dimensional basic shapes and relic basic shape units, expresses various features of the relic basic shape units and the relic texture basic units through feature labels, can express the relic basic shape units and the relic texture basic units in all directions, and the feature labels can be dynamically managed. By constructing the association relationship of the relic basic shape units and the relic texture basic units, the correlation of the relic basic shape units and the relic texture basic units is established, thereby a multidimensional knowledge base model is established, which can be applied to training of a relic reconstruction type deep learning model, and the precision and adaptability of constructing the knowledge base are improved.

[0103] The method of the present application will be further described below in combination with a specific embodiment. The present application uses a bronze chariot of the Qin Dynasty as shown in FIG. 1 as an optimal example: Figure 3

[0104] The first step is to construct basic shapes, including three-dimensional basic shapes and relic basic shape units.

[0105] Extract three-dimensional basic shapes: typical three-dimensional basic shapes in the bronze chariot include a circular ring (a chariot wheel), a cube (a chariot compartment, a window), a sphere (a horse's eye), a cylinder (a horse leg, an axle, etc.), and a cone (a chariot cover).

[0106] The three-dimensional basic shape space distribution operation constructs a relic basic shape unit that is more suitable for the relic. Two cubes of different sizes are used to perform set difference operation to construct a chariot compartment relic basic shape unit. The shape of the cone is deformed to construct a chariot cover, a horse leg, and a horse eye relic basic shape unit. The above relic basic shape units are divided and saved according to a size of 10px;

[0107] The second step is to analyze the texture basic feature attributes of the bronze chariot to construct a relic texture basic unit. According to the glossiness, size, shape, and position information of the bronze chariot texture, the bronze chariot texture is divided according to a size of 10px from the positions of the wheel, the compartment, the cover, the leg, the head, the body, and the hoof in turn, and is stored in the database.

[0108] The third step is to construct a feature label library of the relic basic shape units and the relic texture basic units that can be applied to deep learning training:

[0109] Among them, the era feature, the regional feature, and the cultural feature are common features of the relic basic shape units and the relic texture basic units.

[0110] Era feature: the Qin Dynasty, the age

[0111] Regional feature: Xianyang

[0112] ​​Cultural characteristics: Qin culture

[0113] Features unique to the basic shape unit of cultural relics:

[0114] Morphological features: round, cart, horse, cart cover, axle, cone, cylinder, cube

[0115] Process features: firing method

[0116] Material characteristics: bronze

[0117] Features unique to the basic texture unit of cultural relics:

[0118] Morphological features: cart, horse, rope

[0119] Use features: main body, background, interior, exterior, high light area, dark area

[0120] Glossiness features: smooth, round, rough

[0121] Shape features: mottling, uniform color

[0122] Size features: basic size (10 10px)

[0123] The binding relationship between the feature label and the basic shape unit and the basic texture unit of cultural relics is constructed, such as the basic shape unit of the wheel of cultural relics binding the feature label: Qin Dynasty, Xianyang, Qin culture, round, firing method, bronze, and the basic texture unit of the wheel part of cultural relics binding the feature label: Qin Dynasty, Xianyang, Qin culture, cart, main body, rough, mottling, basic size, and the feature labels of other basic shape units and basic texture units of cultural relics are sequentially bound.

[0124] Finally, the spatial correlation relationship feature library of the basic shape unit and the basic texture unit of cultural relics is constructed: the finally constructed basic shape unit and the basic texture unit of cultural relics segmented from the bronze chariot and horse have the same era characteristics, regional characteristics and cultural characteristics, so the correlation relationship can be established. The basic shape unit of the wheel of cultural relics and the basic texture unit of the wheel of cultural relics establish a binding relationship, and at the same time, the basic shape unit of the wheel of cultural relics and the basic texture unit of the horse hoof have similarity, so the correlation relationship can also be established.

[0125] A deep learning model can also be constructed to establish the spatial correlation relationship between the basic shape unit and the basic texture unit of cultural relics.

[0126] At this point, the digital point cloud knowledge base of the bronze chariot and horse is constructed.

[0127] It should be understood that various parts of the present application can be realized by hardware, software, firmware or a combination thereof.

[0128] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for constructing a digital point cloud knowledge base for cultural relics based on style features, characterized in that, include: Step 1: Analyze the geometric composition of the target cultural relic, obtain the three-dimensional basic shape, and generate and store the basic units of the cultural relic shape representing the characteristics of the cultural relic components through spatial distribution calculation. Step 2: Segment the texture of the target cultural relic, and generate and store the basic texture units of the cultural relic with normalized size based on the integrity, gloss, size and shape attributes of the texture. Step 3: Configure multi-dimensional feature labels for the basic units of cultural relic shape and the basic units of cultural relic texture. The types of feature labels include tag type, number type, character type, time type and enumeration type. Step 4: Based on the correlation of style features, the adaptability of shape and texture, and the mapping relationship learned through deep learning models, construct a feature library of spatial correlation between basic units of cultural relic shape and basic units of cultural relic texture, thereby forming a digital point cloud knowledge base for cultural relics.

2. The method according to claim 1, characterized in that, The spatial distribution operation includes Boolean operation, deformation operation and combination operation. The Boolean operation includes performing union, difference or intersection operations on multiple three-dimensional basic shapes. The deformation operation includes performing extrusion, scaling, movement or rotation operations on three-dimensional basic shapes or intermediate shapes after Boolean operation. The combination operation includes recombining the basic units of the cultural relic shape obtained after the Boolean operation and / or deformation operation.

3. The method according to claim 1, characterized in that, The size normalization includes: The length and width of the segmented textures are uniformly adjusted to 10 pixels or integer multiples of 10 pixels, and pixels are padded to complete the textures that are not large enough.

4. The method according to claim 1, characterized in that, The data structure of the feature tag includes a feature type field, a basic unit field, a display unit field, a display value field, a first numerical field for storing a numerical start value or a specific value based on the basic unit, and a second numerical field for storing a numerical end value. When performing feature retrieval, the values ​​of the first numerical field and the second numerical field are directly compared without the need for unit conversion.

5. The method according to claim 5, characterized in that, When the feature label is numeric, it supports the storage and retrieval of single values ​​or range values; when the feature label is time-based, it supports the storage and retrieval of specific time points or time ranges.

6. The method according to claim 1, characterized in that, The feature label dimensions configured for the basic unit of the cultural relic shape include at least one of the following: period features, regional features, cultural features, morphological features, craft features, and material features; The feature label dimensions configured for the basic texture unit of the cultural relic include at least one of the following: period features, regional features, cultural features, morphological features, usage features, gloss features, shape features, and size features.

7. The method according to claim 1, characterized in that, The methods for establishing the spatial association include at least one of the following: The association is based on the common era characteristics, regional characteristics and / or cultural characteristics between the basic units of the artifact shape and the basic units of the artifact texture; Based on the shape and / or use of the basic unit of the artifact shape, bind and adapt the basic unit of the artifact texture; By using a deep learning model, the association mapping relationship between the basic units of the shape of cultural relics and the basic units of the texture of cultural relics is trained and established, so as to automatically update and expand the knowledge base.

8. The method according to any one of claims 1 to 7, characterized in that, The basic three-dimensional shapes include cubes, spheres, cylinders, cones, tori, and planes.

Images