A three-dimensional model construction method and system for energy-saving and environmentally friendly factory buildings

Abstract

The present invention relates to the field of image preprocessing technology, and in particular to a method and system for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building. First, a factory building wall area is divided into various grayscale similar areas based on threshold segmentation, and then, according to the more important wall features in modeling, an interference feature area that interferes with feature point detection is screened out, and further, according to the distribution of each initial feature point relative to the interference feature area, the influence degree of the feature point is determined; further, combined with the characteristic that the discrete initial clarity usually has a large detail requirement, the detail requirement degree is determined; thereby, a more accurate and adaptive correction scale space factor is determined in combination with the feature point influence degree, the detail requirement degree and the initial clarity, so that the accuracy of the modeling feature points determined according to the correction scale space factor is higher, and the effect of constructing the three-dimensional model of the energy-saving and environmentally friendly factory building according to the detected modeling feature points is better.

Description

Technical Field

[0001] The present invention relates to the technical field of image preprocessing, and in particular to a method and system for constructing a three-dimensional model of an energy-saving and environment-friendly factory building. Background Art

[0002] Factory buildings are buildings or places used for the production, manufacture, assembly, storage and maintenance of products in industrial production. They are the core part of industrialization and manufacturing, with different structures and characteristics in different industries' needs and production processes. With the advancement of technology and the improvement of environmental awareness, modern factory buildings pay more and more attention to energy efficiency and sustainability. The design of factory buildings is becoming more energy-saving and environmentally friendly, and gradually tends to be environmentally friendly. Energy-saving and environmentally friendly factory buildings refer to industrial buildings that adopt a series of energy-saving, emission-reduction and environmental protection measures during the design, construction and operation process to reduce energy consumption, reduce pollution and improve resource utilization efficiency. Through efficient insulation materials, reasonable architectural design, intelligent control systems and other technical means, they effectively reduce energy consumption, reduce production costs, and reduce the impact on the environment.

[0003] After the construction of energy-saving and environmentally friendly factories is completed, the three-dimensional modeling of the factories can be used to understand the spatial layout of the factories more easily and accurately, and the three-dimensional model of the factories can also be used to better plan and operate the entire factory area. When constructing the three-dimensional model of the factory, considering that the accurate selection of feature points can ensure the accuracy of the geometric and topological analysis of the three-dimensional model, and will directly affect the quality and reliability of the three-dimensional model, the modeling feature points are usually obtained through the scale-invariant feature transform (SIFT) algorithm based on the grayscale images of the factory surface taken from various shooting angles of the energy-saving and environmentally friendly factories, and then the three-dimensional model of the factory is constructed based on the obtained modeling feature points.

[0004] When the prior art obtains the modeling feature points of the grayscale image of the plant surface based on the SIFT algorithm, a fixed scale space factor is usually set a priori for feature point detection. If the scale space factor is set too small, more details will be retained, but the influence of noise will also increase; if the scale space factor is set too large, details may be lost, and the calculation complexity will be higher; so it is crucial to choose a suitable scale space factor. However, when modeling, it is necessary to consider that different areas of the plant components have different degrees of importance to the model construction. Setting the same size of scale space factor may result in the loss of important details and the retention of unimportant details, resulting in a greater impact of noise, making the accuracy of the obtained modeling feature points low. That is, according to the prior art, the accuracy of feature point detection by setting a fixed scale space factor a priori is low, making the effect of constructing a three-dimensional model of an energy-saving and environmentally friendly plant based on the detected modeling feature points poor. Summary of the invention

[0005] The present application provides a method and system for constructing a three-dimensional model for an energy-saving and environmentally friendly factory building. After determining the initial feature points based on the initial clarity confirmed only according to the grayscale change, the factory building wall area is first divided into various grayscale similar areas based on threshold segmentation; then, according to the more important wall features in the modeling, the interference feature areas that interfere with the feature point detection are screened out in the grayscale similar areas, and further according to the distribution of each initial feature point relative to the interference feature area, the influence degree of the feature points in the factory building wall area is determined; further, in combination with the characteristic that the discrete initial clarity usually has a large detail requirement, the detail requirement degree is determined; thereby, a more accurate and adaptive correction scale space factor is determined in combination with the feature point influence degree, the detail requirement degree and the initial clarity, so that the accuracy of the modeling feature points determined according to the correction scale space factor is higher, which solves the problem of low accuracy of feature point detection by setting a fixed scale space factor a priori in the prior art, so that the effect of constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to the detected modeling feature points is better.

[0006] The present application provides a method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building, the method comprising:

[0007] Obtain each plant wall area in the plant surface grayscale image of the energy-saving and environmentally friendly plant; determine the corresponding initial clarity according to the change of the grayscale value in each plant wall area; determine the initial feature points of each plant wall area according to the initial clarity combined with the scale-invariant feature algorithm;

[0008] In the grayscale image of the factory building surface, each factory building wall area is divided into at least two grayscale similar areas according to the similarity of the grayscale value distribution in each factory building wall area by using the threshold segmentation method; the wall credibility of each grayscale similar area is determined according to the wide distribution of each grayscale similar area and the consistency of the gradient direction of the edge pixel points; and the interference feature area is screened out according to the wall credibility;

[0009] In each plant wall area, the influence degree of the feature points in each interfering feature area is determined according to the number of initial feature points in each interfering feature area and the distribution proximity of all initial feature points relative to each interfering feature area; the detail requirement degree of each plant wall area is determined according to the numerical discreteness of the initial clarity of each plant wall area relative to the overall initial clarity of each plant wall area;

[0010] According to the influence degree of the feature points, the detail requirement degree and the initial clarity degree, the corrected scale space factor of each factory building wall area is determined; according to the corrected scale space factor combined with the scale invariant feature algorithm, all modeling feature points in the grayscale image of the factory building surface are determined; and according to the modeling feature points, a three-dimensional model of an energy-saving and environmentally friendly factory building is constructed.

[0011] Furthermore, the process of obtaining the initial clarity includes:

[0012] According to the gray value of each pixel in each factory wall area, the initial clarity of each factory wall area is calculated by the Brenner gradient function.

[0013] Furthermore, the process of obtaining the initial feature points includes:

[0014] The positive correlation mapping value of the initial clarity of each factory building wall area is used as the initial scale space factor corresponding to each factory building wall area;

[0015] All initial feature points in each factory building wall area are determined according to the initial scale space factor combined with a scale invariant feature algorithm.

[0016] Furthermore, the process of obtaining the grayscale similar area includes:

[0017] Each factory building wall area is segmented into at least two grayscale similar areas by using a multi-threshold Otsu threshold segmentation algorithm with a preset threshold number; the preset threshold number is a positive integer greater than or equal to 2.

[0018] Furthermore, the process of obtaining the wall credibility includes:

[0019] The difference between the area of ​​each factory building wall area and the area of ​​the minimum circumscribed rectangle of each corresponding gray-level similar area is taken as the regional distribution limitation of each gray-level similar area;

[0020] In each grayscale similar region, each edge pixel is combined with each other edge pixel into an edge pixel binary; all edge pixel binary tuples of each grayscale similar region are obtained;

[0021] The angle between the straight lines corresponding to the gradient directions of the two edge pixel points in each edge pixel binary is taken as the reference angle of each edge pixel binary; the normalized value of the mean of the reference angles of all edge pixel binary groups is taken as the gradient direction difference of each grayscale similar region;

[0022] The product of the regional distribution limitation and the gradient direction difference is subjected to negative correlation normalization processing to determine the wall credibility of each grayscale similar area.

[0023] Furthermore, the process of acquiring the interference feature area includes:

[0024] In each factory building wall area, the grayscale similar area corresponding to the wall credibility less than the preset credibility threshold is taken as the interference feature area.

[0025] Furthermore, the process of obtaining the influence degree of the feature points includes:

[0026] The accumulated value of the number of initial feature points in all the interfering feature areas in each factory building wall area is used as the number of interfering feature points in each factory building wall area; the ratio between the number of interfering feature points and the total number of initial feature points in each factory building wall area is used as the first influence degree of each factory building wall area;

[0027] In each factory building wall area, other areas outside the interference feature area are taken as wall feature areas; the boundary lines between all interference feature areas and the wall feature area are taken as reference intersection lines of each factory building wall area; the minimum distance between each initial feature point in other areas outside the interference feature area in each factory building wall area and the reference intersection line is taken as the interference area proximity of each initial feature point; wherein the interference area proximity of the initial feature point inside the interference feature area is set to 0; the negative correlation mapping value of the mean of the interference area proximity of all initial feature points in each factory building wall area is taken as the second influence degree of each factory building wall area;

[0028] The product of the second influence degree and the first influence degree is used as the influence degree of the feature point of each factory building wall area.

[0029] Furthermore, the process of obtaining the detail requirement level includes:

[0030] The average of the initial clarity of all factory building wall areas is taken as the overall clarity; the difference between the initial clarity of each factory building wall area and the overall clarity is normalized to determine the detail requirement of each factory building wall area.

[0031] Furthermore, the process of obtaining the modified scale space factor includes:

[0032] The product of the negative correlation mapping value of the detail requirement degree and the influence degree of the feature point is used as the adjustment weight; the positive correlation mapping value of the product of the adjustment weight and the initial clarity degree is used as the modified scale space factor of each factory wall area.

[0033] The present application also proposes a three-dimensional model construction system for energy-saving and environmentally friendly factory buildings, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of a method for constructing a three-dimensional model for an energy-saving and environmentally friendly factory building.

[0034] This application has the following beneficial effects:

[0035] The clarity of the image area affects the significance of local features. Therefore, the prior art usually determines the scale space factor based on the clarity, thereby selecting the appropriate scale space factor according to the image details, and further determining the required modeling feature points according to the scale space factor. Therefore, the corresponding initial clarity can be determined first according to the change of the grayscale value in each factory wall area. However, there may be some non-factory components in the factory wall area of ​​the present application, such as slogans or logo patterns on the wall doors and windows. If the feature point detection is performed directly according to the scale space factor determined by the initial clarity, the detected feature points will be affected by the areas corresponding to these non-factory components, that is, the interference feature areas. These interference feature areas will cause some unnecessary feature points to be generated and will affect the distribution of the feature points; therefore, it is necessary to obtain the initial feature points of each factory wall area, so as to quantify the influence of all interference feature areas on the feature point detection in combination with the distribution of the initial feature points, so as to make a more accurate correction to the scale space factor.

[0036] Before analyzing the influence of interference feature areas, we first need to screen out these interference feature areas. In contrast to interference feature areas, there are usually factory walls that contribute more to modeling. These factory walls are usually of the same shape and have concave and convex grooves interspersed throughout the wall. The concave and convex parts show different grayscale values ​​in the image, but their texture is relatively simple; while the interference feature areas corresponding to the patterns and slogans on the wall usually only occupy a small part of the wall and have a relatively complex texture; therefore, after screening out each grayscale similar area based on the threshold segmentation method, the grayscale similar areas corresponding to the concave and convex parts of the factory wall usually still maintain the characteristics of wide distribution and single texture, so the wall credibility is further determined based on the wide distribution of each grayscale similar area and the consistency of the gradient direction of the edge pixel points; thereby screening out interference feature areas with low wall credibility.

[0037] For each factory wall area, the more initial feature points contained in the interference feature area, and the closer the initial feature points are as a whole to the interference feature area, it means that the existence of the interference feature area has a greater impact on the selection of modeling feature points, that is, the greater the degree of influence of the feature points of the factory wall area; therefore, the present application determines the degree of influence of the feature points of each factory wall area according to the number of initial feature points in each interference feature area and the distribution proximity of all initial feature points relative to each interference feature area. In addition, in the same grayscale image of the factory surface, the initial clarity of each factory wall area should generally be similar. If the initial clarity of one or part of the factory wall area is relatively discrete, in order to protect the details, it is generally based on a higher degree of detail requirement; therefore, the degree of detail requirement of each factory wall area is determined according to the numerical discreteness of the initial clarity of each factory wall area relative to the overall initial clarity of each factory wall area.

[0038] For each factory wall area, when the influence of the feature points is greater, if the influence of the interference area is to be reduced, the number of detail features corresponding to the interference area needs to be reduced, so a larger spatial scale factor is required; and the greater the detail requirement, the more detail features need to be added, and the smaller the corresponding spatial scale factor requirement is; therefore, based on the influence of the feature points and the detail requirement, the optimized corrected scale space factor is comprehensively determined in combination with the initial clarity, so that each factory wall area can adaptively determine a more accurate corrected scale space factor; the accuracy of the modeling feature points determined according to the corrected scale space factor is higher, which solves the problem of low accuracy of feature point detection by setting a fixed scale space factor a priori in the prior art, so that the effect of constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to the detected modeling feature points is better. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 A flow chart of a method for constructing a three-dimensional model of an energy-saving and environment-friendly factory building provided by an embodiment of the present invention;

[0041] Figure 2 A schematic diagram of a grayscale image of the surface of an energy-saving and environment-friendly factory building provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a three-dimensional model construction method and system for energy-saving and environmentally friendly factory buildings proposed in accordance with the present invention, in combination with the accompanying drawings and preferred embodiments, its specific implementation method, structure, characteristics and effects are described as follows. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment, and the specific features, structures or characteristics in one or more embodiments may be combined in any suitable form. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0043] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0044] The following is a detailed description of a method and system for constructing a three-dimensional model of an energy-saving and environment-friendly factory building provided by the present invention in conjunction with the accompanying drawings.

[0045] The present application embodiment provides a method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building. Figure 1 , which shows a flow chart of a method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building provided by an embodiment of the present invention, the method comprising:

[0046] Step S101: Obtain each plant wall area in the grayscale image of the plant surface of the energy-saving and environmentally friendly plant; determine the corresponding initial clarity based on the change of the grayscale value in each plant wall area; determine the initial feature points of each plant wall area based on the initial clarity combined with the scale-invariant feature algorithm.

[0047] The clarity of the image area affects the significance of local features. Therefore, the prior art usually determines the scale space factor based on the clarity, thereby selecting the appropriate scale space factor according to the image details, and further determining the required modeling feature points according to the scale space factor; therefore, the corresponding initial clarity can be first determined according to the change in the grayscale value in each factory wall area. However, the factory wall area of ​​the present application may have some non-factory components such as slogans or logo patterns on the wall doors and windows. If the feature point detection is performed directly according to the scale space factor determined by the initial clarity, the detected feature points will be affected by the areas corresponding to these non-factory components, that is, the interference feature areas. These interference feature areas will cause some unnecessary feature points to be generated and will affect the distribution of the feature points; therefore, it is necessary to obtain the initial feature points of each factory wall area, so as to quantify the influence of all interference feature areas on the feature point detection in combination with the distribution of the initial feature points, so as to make a more accurate correction to the scale space factor.

[0048] In a specific implementation of the embodiment of the present invention, a drone equipped with a high-precision industrial camera is used to shoot the outer surface of the energy-saving and environmentally friendly factory building including the factory building wall from all directions, thereby obtaining a grayscale image of the factory building surface corresponding to each shooting; and all grayscale images of the factory building surface can completely contain the entire energy-saving and environmentally friendly factory building. Figure 2 , which shows a schematic diagram of a grayscale image of the surface of an energy-saving and environment-friendly factory building provided by an embodiment of the present invention.

[0049] Considering that the grayscale image of the factory surface contains background areas and areas corresponding to the factory wall parts, and the background areas will affect the subsequent analysis, it is necessary to screen out the areas corresponding to the factory wall parts. In a specific implementation method of an embodiment of the present invention, each factory surface grayscale image is input into a trained convolutional neural network, and each factory wall area in each factory surface grayscale image is output, wherein each factory wall area corresponds to a wall, and the loss function adopts a cross entropy loss function. It should be noted that convolutional neural networks are a technical means well known to those skilled in the art, and will not be further limited or elaborated here.

[0050] Preferably, in a specific implementation of the embodiment of the present invention, the process of acquiring the initial clarity includes:

[0051] According to the grayscale value of each pixel in each factory wall area, the initial clarity of each factory wall area is calculated by the Brenner gradient function. The Brenner gradient function evaluates the clarity of the image by calculating the grayscale difference between adjacent pixels in the image; it should be noted that the Brenner gradient function is a technical means well known to those skilled in the art and will not be further defined or elaborated herein.

[0052] Preferably, in a specific implementation of the embodiment of the present invention, the process of acquiring the initial feature points includes:

[0053] The positive correlation mapping value of the initial clarity of each factory wall area is used as the initial scale space factor corresponding to each factory wall area; all initial feature points in each factory wall area are determined according to the initial scale space factor combined with the scale invariant feature algorithm. It should be noted that the scale-invariant feature transform (SIFT) is a technical means well known to those skilled in the art and will not be further defined or elaborated here. In a specific implementation of an embodiment of the present invention, the process of obtaining the initial scale space factor is expressed by the formula: ;in, For the The initial scale space factor of the factory wall area; For the The initial clarity of the plant wall area; is a linear normalization function, and the implementer can adjust the normalization method according to the specific implementation environment. When determining the scale space factor based on the clarity, if the image is relatively blurred, in order to retain more image details, a smaller scale space factor should be used; on the contrary, if the image clarity is high, in order to reduce the impact of noise, a larger scale space factor can be appropriately given to extract more critical detail features; and according to existing empirical values, the scale space factor in the SIFT algorithm is usually between 1 and 2. Therefore, combined with the initial clarity of each plant wall area, the value of the initial scale space factor of each plant wall area is limited to 1 to 2 through the positive correlation mapping method.

[0054] Step S102: In the grayscale image of the factory surface, each factory wall area is divided into at least two grayscale similar areas according to the similarity of the grayscale value distribution in each factory wall area through the threshold segmentation method; the wall credibility of each grayscale similar area is determined according to the wide distribution of the area of ​​each grayscale similar area and the consistency of the gradient direction of the edge pixel points; and the interference feature area is screened out according to the wall credibility.

[0055] Before analyzing the influence of interference feature areas, we first need to screen out these interference feature areas. In contrast to interference feature areas, there are usually factory walls that contribute more to modeling. These factory walls are usually of the same shape and have concave and convex grooves interspersed throughout the wall. The concave and convex parts show different grayscale values ​​in the image, but their texture is relatively simple; while the interference feature areas corresponding to the patterns and slogans on the wall usually only occupy a small part of the wall and have a relatively complex texture; therefore, after screening out each grayscale similar area based on the threshold segmentation method, the grayscale similar areas corresponding to the concave and convex parts of the factory wall usually still maintain the characteristics of wide distribution and single texture, so the wall credibility is further determined based on the wide distribution of each grayscale similar area and the consistency of the gradient direction of the edge pixel points; thereby screening out interference feature areas with low wall credibility.

[0056] Preferably, in some possible implementations of the embodiments of the present invention, the process of acquiring grayscale similar regions includes:

[0057] Each factory wall area is divided into at least two grayscale similar areas by a multi-threshold Otsu threshold segmentation algorithm with a preset threshold number; the preset threshold number is a positive integer greater than or equal to 2. In a specific implementation of an embodiment of the present invention, the preset threshold number is set to 2, that is, the segmentation threshold of the multi-threshold Otsu threshold segmentation algorithm in the embodiment of the present invention is set to 2, and each factory wall area is divided into three grayscale similar areas, specifically: the smaller threshold of the two segmentation thresholds is used as the first threshold, and the larger threshold is used as the second threshold; the area corresponding to all pixels in each factory wall area whose grayscale value is greater than the second threshold is used as the first grayscale similar area; the area corresponding to all pixels in each factory wall area whose grayscale value is less than the first threshold is used as the second grayscale similar area; the area composed of other pixels outside the first grayscale similar area and the second grayscale similar area is used as the third grayscale similar area. It should be noted that the size of the preset threshold number can be adjusted according to the specific implementation environment, and the multi-threshold Otsu threshold segmentation algorithm is a technical means well known to those skilled in the art, and will not be further defined or elaborated here. In another specific implementation of the embodiment of the present invention, two thresholds may be directly set a priori to perform threshold segmentation, which will not be further elaborated here.

[0058] Preferably, in some possible implementations of the embodiments of the present invention, the process of acquiring the wall credibility includes:

[0059] The difference between the area of ​​each factory wall area and the area of ​​the minimum circumscribed rectangle of each corresponding grayscale similar area is taken as the regional distribution limitation of each grayscale similar area. The factory wall is usually of the same shape and has concave and convex grooves interspersed throughout the wall, and the concave and convex parts show different grayscale values ​​in the image, but their texture is relatively simple. The interference feature areas corresponding to the patterns and slogans on the wall usually only occupy a small part of the wall. Therefore, the grayscale similar areas corresponding to the concave parts and the grayscale similar areas corresponding to the convex parts are widely distributed in the factory wall area, and the areas of the corresponding minimum circumscribed rectangles are close to the area of ​​the factory wall area, and the area of ​​the minimum circumscribed rectangle corresponding to the interference feature area is usually much smaller than the area of ​​the factory wall area. Therefore, the smaller the regional distribution limitation, the more likely the corresponding grayscale similar area is to belong to the wall, that is, the higher the corresponding wall credibility.

[0060] In each grayscale similar region, each edge pixel is combined with each other edge pixel into an edge pixel binary; all edge pixel binary tuples of each grayscale similar region are obtained; the angle between the straight lines corresponding to the gradient directions of the two edge pixel points in each edge pixel binary tuple is used as the reference angle of each edge pixel binary tuple; the normalized value of the mean of the reference angles of all edge pixel binary tuples is used as the gradient direction difference of each grayscale similar region. Since the textures corresponding to the concave and convex parts are relatively simple and both appear as rectangles with longer long sides, the straight lines corresponding to the gradient directions between the edge pixel points belonging to the long sides in the grayscale similar region are usually parallel or close to parallel, and since the edge pixel points belonging to the long sides account for the majority, the overall reference angle of the edge pixel binary corresponding to the wall is usually small; while the texture of the interference feature region is more complex, and the corresponding edge direction has more complex changes, so the corresponding reference angle is usually larger overall; so for each grayscale similar region, the smaller the corresponding gradient direction difference is, the more likely it is to belong to the wall, that is, the higher the corresponding wall credibility is.

[0061] Since the smaller the regional distribution limitation is and the smaller the gradient direction difference is, the higher the wall credibility of the corresponding grayscale similar area is, the product of the regional distribution limitation and the gradient direction difference is further subjected to negative correlation normalization to determine the wall credibility of each grayscale similar area. In a specific implementation of the embodiment of the present invention, the wall credibility acquisition process is expressed by the formula: ;in, For the In the wall area of ​​the factory building The wall credibility of the gray-scale similar area; For the The area of ​​the factory wall area; For the In the wall area of ​​the factory building The area of ​​the smallest circumscribed rectangle of grayscale similar regions; is the absolute value symbol; For the In the wall area of ​​the factory building The number of edge pixel pairs in grayscale-similar areas; For the In the wall area of ​​the factory building The first grayscale similar area The reference angle of the edge pixel pair; For the In the wall area of ​​the factory building The regional distribution limitation of similar grayscale areas; For the In the wall area of ​​the factory building Differences in gradient directions between regions with similar grayscale values; It is a linear normalization function. The implementer can adjust the normalization method according to the specific implementation environment. Through normalization, the value of the wall credibility is limited to 0 to 1, which is convenient for the subsequent screening of interference feature areas.

[0062] Preferably, in some possible implementations of the embodiments of the present invention, the process of acquiring the interference feature area includes:

[0063] The greater the wall credibility, the greater the possibility of belonging to the factory wall, and the smaller the possibility of the corresponding interference feature area corresponding to the pattern and slogan on the wall; therefore, in each factory wall area, the grayscale similar area corresponding to the wall credibility less than the preset credibility threshold is used as the interference feature area. In a specific implementation of the embodiment of the present invention, the preset credibility threshold is set to 0.6, which can be adjusted according to the specific implementation environment.

[0064] Step S103: In each factory wall area, the influence degree of the feature points of each factory wall area is determined according to the number of initial feature points in each interference feature area and the distribution proximity of all initial feature points relative to each interference feature area; the detail requirement degree of each factory wall area is determined according to the numerical discreteness of the initial clarity of each factory wall area relative to the overall initial clarity of each factory wall area.

[0065] For each factory wall area, the more initial feature points contained in the interference feature area, and the closer the initial feature points are as a whole to the interference feature area, it means that the existence of the interference feature area has a greater impact on the selection of modeling feature points, that is, the greater the degree of influence of the feature points of the factory wall area; therefore, the present application determines the degree of influence of the feature points of each factory wall area according to the number of initial feature points in each interference feature area and the distribution proximity of all initial feature points relative to each interference feature area. In addition, in the same grayscale image of the factory surface, the initial clarity of each factory wall area should generally be similar. If the initial clarity of one or part of the factory wall area is relatively discrete, in order to protect the details, it is generally based on a higher degree of detail requirement; therefore, the degree of detail requirement of each factory wall area is determined according to the numerical discreteness of the initial clarity of each factory wall area relative to the overall initial clarity of each factory wall area.

[0066] Preferably, in some possible implementations of the embodiments of the present invention, the process of obtaining the influence degree of the feature points includes:

[0067] The cumulative value of the number of initial feature points in all interference feature areas in each factory wall area is taken as the number of interference feature points in each factory wall area; the ratio between the number of interference feature points and the total number of initial feature points in each factory wall area is taken as the first degree of influence of each factory wall area. The existence of interference feature areas will affect the extraction of feature points for factory modeling. The more initial feature points in the interference feature area, the more unnecessary feature points generated by the interference feature area, and the greater the impact on feature point detection; therefore, the number of interference feature points is further normalized by the total number of initial feature points in each factory wall area to determine the first degree of influence that characterizes the impact of feature point detection.

[0068] In each factory building wall area, other areas outside the interference feature area are taken as wall feature areas; the boundary lines between all interference feature areas and the wall feature areas are taken as reference intersection lines of each factory building wall area; the minimum distance between each initial feature point in other areas outside the interference feature area in each factory building wall area and the reference intersection line is taken as the interference area proximity of each initial feature point; wherein the interference area proximity of the initial feature points inside the interference feature area is set to 0; the negative correlation mapping value of the mean of the interference area proximity of all initial feature points in each factory building wall area is taken as the second influence degree of each factory building wall area. Feature points are points that can reflect the key information of the factory wall and should be distributed in the factory wall area. If the minimum distance between all initial feature points and the reference boundary line is smaller as a whole, it means that the overall initial feature points are more concentrated at the junction of the factory wall area and the interference feature area, that is, closer to the interference feature area, which means that the interference feature area has a greater impact on feature point detection; therefore, the mean of the interference area proximity of all initial feature points in each factory wall area is negatively correlated to obtain the second influence degree that characterizes the impact of feature point detection. It should be noted that the reason why the interference area proximity of the initial feature points inside the interference feature area is set to 0 is that because the corresponding initial feature points are already located in the interference feature area, using the minimum distance to characterize the interference area proximity does not conform to objective facts, so it is replaced by the preset value 0.

[0069] Since both the first influence degree and the second influence degree characterize the influence of the interference feature area on the feature point detection, the product of the second influence degree and the first influence degree is further used as the influence degree of the feature points of each factory wall area. In a specific implementation of the embodiment of the present invention, the process of obtaining the influence degree of the feature points is expressed by the formula: ;in, For the The influence degree of characteristic points in the wall area of ​​a factory building; For the The number of interference feature points in the wall area of ​​the plant, that is, The cumulative value of the number of initial feature points in all interference feature areas in the wall area of ​​a factory building; For the The total number of initial feature points in the wall area of ​​the factory building; For the The first degree of impact on the wall area of ​​a factory building; For the In the wall area of ​​the factory building The proximity of the interference areas of the initial feature points; For the The second impact level of the plant wall area; It is a linear normalization function. The implementer can adjust the normalization method according to the specific implementation environment.

[0070] Preferably, in some possible implementations of the embodiments of the present invention, the process of obtaining the detail requirement level includes:

[0071] The average of the initial clarity of all plant wall areas is taken as the overall clarity; the difference between the initial clarity of each plant wall area and the overall clarity is normalized to determine the detail requirement of each plant wall area. Since the more discrete the initial clarity of a plant wall area is relative to the initial clarity of other plant wall areas, the higher the corresponding detail requirement is, and the smaller the required spatial scale factor should be; therefore, after calculating the overall clarity that represents the overall numerical characteristics of the initial clarity of all plant wall areas, the greater the difference between the initial clarity of each plant wall area and the overall clarity, the more discrete the initial clarity of the corresponding plant wall area is, and the higher the corresponding detail requirement is.

[0072] In other possible implementation methods of the embodiments of the present invention, the initial clarity difference between each factory wall area and each other factory wall area can also be calculated, and the normalized value of the cumulative sum of all initial clarity differences corresponding to each factory wall area can be used as the detail requirement degree of each factory wall area.

[0073] In a specific implementation of the embodiment of the present invention, the process of obtaining the detail requirement degree is expressed by the formula: ;in, For the The degree of detail required for each plant wall area; For the The initial clarity of the plant wall area; It is the average of the initial clarity of all factory wall areas, that is, the overall clarity.

[0074] Step S104: Determine the corrected scale space factor of each factory building wall area according to the degree of influence of the feature points, the degree of detail required and the initial clarity; determine all modeling feature points in the grayscale image of the factory building surface according to the corrected scale space factor combined with the scale invariant feature algorithm; and construct a three-dimensional model of the energy-saving and environmentally friendly factory building according to the modeling feature points.

[0075] For each factory wall area, when the influence of the feature points is greater, if the influence of the interference area is to be reduced, the number of detail features corresponding to the interference area needs to be reduced, so a larger spatial scale factor is required; and the greater the detail requirement, the more detail features need to be added, and the smaller the corresponding spatial scale factor requirement is; therefore, based on the influence of the feature points and the detail requirement, the optimized corrected scale space factor is comprehensively determined in combination with the initial clarity, so that each factory wall area can adaptively determine a more accurate corrected scale space factor; the modeling feature points determined according to the corrected scale space factor are more accurate, which solves the problem of low accuracy of feature point detection by setting a fixed scale space factor a priori in the prior art, so that the effect of constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to the detected modeling feature points is better.

[0076] Preferably, the process of obtaining the modified scale space factor includes:

[0077] Since the smaller the detail requirement is and the greater the influence of the feature points is, the larger the required corrected scale space factor is, the product of the negative correlation mapping value of the detail requirement and the influence of the feature points is used as the adjustment weight, so that when the adjustment weight is larger, the corresponding corrected scale space factor is larger; further, the initial clarity obtained initially is adjusted by adjusting the weight, so as to obtain the corrected scale space factor based on the comprehensive clarity, so that the obtained corrected scale space factor is more accurate; finally, the positive correlation mapping value of the product between the adjustment weight and the initial clarity is used as the corrected scale space factor of each factory wall area.

[0078] In a specific implementation of the embodiment of the present invention, the process of obtaining the modified scale space factor is expressed by the formula: ;in, For the The modified scale space factor of the plant wall area; For the The degree of detail required for each plant wall area; For the The influence degree of characteristic points in the wall area of ​​a factory building; For the The initial clarity of the plant wall area; For the Adjustment weight of each factory wall area; It is a linear normalization function. The implementer can adjust the normalization method according to the specific implementation environment. is an exponential function with a natural constant as the base. Implementers can adopt other negative correlation mapping methods according to the specific implementation environment, such as , and the reciprocal, where is the hyperbolic tangent function, which will not be further described here. Since the scale space factor is usually between 1 and 2, Perform positive correlation mapping.

[0079] After obtaining the corrected scale-invariant factor of each factory building wall area in the grayscale image of each factory building surface, all modeling feature points in the grayscale image of each factory building surface are finally determined based on the corrected scale space factor combined with the scale-invariant feature algorithm; and a three-dimensional model of the energy-saving and environmentally friendly factory building is constructed based on the modeling feature points. In a specific implementation of an embodiment of the present invention, three-dimensional reconstruction is performed using multi-view geometry technology based on all the obtained modeling feature points to determine the three-dimensional model of the energy-saving and environmentally friendly factory building; it should be noted that the multi-view geometry technology is a technical means well known to those skilled in the art, and is not further defined or elaborated herein.

[0080] To summarize, after determining the initial feature points based on the initial clarity confirmed only according to the grayscale change, the present application first divides the factory wall area into various grayscale similar areas based on threshold segmentation; then, according to the more important wall features in the modeling, the interference feature areas that interfere with the feature point detection are screened out in the grayscale similar areas, and further according to the distribution of each initial feature point relative to the interference feature area, the degree of influence of the feature points in the factory wall area is determined; further combined with the characteristics that the discrete initial clarity usually has a large demand for details, the degree of detail demand is determined; thereby combining the degree of influence of the feature points, the degree of detail demand and the initial clarity to determine the corrected scale space factor more accurately and adaptively, so that the modeling feature points determined according to the corrected scale space factor are more accurate, which solves the problem of low accuracy of feature point detection in the prior art by setting a fixed scale space factor a priori, so that the effect of constructing an energy-saving and environmentally friendly factory three-dimensional model based on the detected modeling feature points is better.

[0081] The present application also proposes a three-dimensional model construction system for energy-saving and environmentally friendly factory buildings, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of a method for constructing a three-dimensional model for an energy-saving and environmentally friendly factory building.

[0082] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0083] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building, characterized in that: The method comprises: Obtain each plant wall area in the plant surface grayscale image of the energy-saving and environmentally friendly plant; determine the corresponding initial clarity according to the change of the grayscale value in each plant wall area; determine the initial feature points of each plant wall area according to the initial clarity combined with the scale-invariant feature algorithm; In the grayscale image of the factory building surface, each factory building wall area is divided into at least two grayscale similar areas according to the similarity of the grayscale value distribution in each factory building wall area by using the threshold segmentation method; the wall credibility of each grayscale similar area is determined according to the wide distribution of each grayscale similar area and the consistency of the gradient direction of the edge pixel points; and the interference feature area is screened out according to the wall credibility; In each plant wall area, the influence degree of the feature points in each interfering feature area is determined according to the number of initial feature points in each interfering feature area and the distribution proximity of all initial feature points relative to each interfering feature area; the detail requirement degree of each plant wall area is determined according to the numerical discreteness of the initial clarity of each plant wall area relative to the overall initial clarity of each plant wall area; According to the influence degree of the feature points, the detail requirement degree and the initial clarity degree, a modified scale space factor of each plant wall area is determined; all modeling feature points in the plant surface grayscale image are determined according to the modified scale space factor combined with a scale-invariant feature algorithm; and a three-dimensional model of an energy-saving and environmentally friendly plant is constructed according to the modeling feature points; The process of obtaining the influence degree of feature points includes: The accumulated value of the number of initial feature points in all the interfering feature areas in each factory building wall area is used as the number of interfering feature points in each factory building wall area; the ratio between the number of interfering feature points and the total number of initial feature points in each factory building wall area is used as the first influence degree of each factory building wall area; In each factory building wall area, other areas outside the interference feature area are taken as wall feature areas; the boundary lines between all interference feature areas and the wall feature area are taken as reference intersection lines of each factory building wall area; the minimum distance between each initial feature point in other areas outside the interference feature area in each factory building wall area and the reference intersection line is taken as the interference area proximity of each initial feature point; wherein the interference area proximity of the initial feature point inside the interference feature area is set to 0; the negative correlation mapping value of the mean of the interference area proximity of all initial feature points in each factory building wall area is taken as the second influence degree of each factory building wall area; The product of the second influence degree and the first influence degree is used as the influence degree of the feature point of each factory building wall area; The process of obtaining the required level of detail includes: The average of the initial clarity of all factory building wall areas is taken as the overall clarity; the difference between the initial clarity of each factory building wall area and the overall clarity is normalized to determine the detail requirement of each factory building wall area.

2. A method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to claim 1, characterized in that: The process of obtaining the initial clarity includes: According to the gray value of each pixel in each factory wall area, the initial clarity of each factory wall area is calculated by the Brenner gradient function.

3. The method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to claim 1, characterized in that: The process of obtaining the initial feature points includes: The positive correlation mapping value of the initial clarity of each factory building wall area is used as the initial scale space factor corresponding to each factory building wall area; All initial feature points in each factory building wall area are determined according to the initial scale space factor combined with a scale invariant feature algorithm.

4. The method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to claim 1, characterized in that: The process of obtaining the grayscale similar region includes: Each factory building wall area is segmented into at least two grayscale similar areas by using a multi-threshold Otsu threshold segmentation algorithm with a preset threshold number; the preset threshold number is a positive integer greater than or equal to 2.

5. The method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to claim 1, characterized in that: The process of obtaining the wall credibility includes: The difference between the area of ​​each factory building wall area and the area of ​​the minimum circumscribed rectangle of each corresponding gray-level similar area is taken as the regional distribution limitation of each gray-level similar area; In each grayscale similar region, each edge pixel is combined with each other edge pixel into an edge pixel binary; all edge pixel binary tuples of each grayscale similar region are obtained; The angle between the straight lines corresponding to the gradient directions of the two edge pixel points in each edge pixel binary is taken as the reference angle of each edge pixel binary; the normalized value of the mean of the reference angles of all edge pixel binary groups is taken as the gradient direction difference of each grayscale similar region; The product of the regional distribution limitation and the gradient direction difference is subjected to negative correlation normalization processing to determine the wall credibility of each grayscale similar area.

6. The method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to claim 1, characterized in that: The process of obtaining the interference feature area includes: In each factory building wall area, the grayscale similar area corresponding to the wall credibility less than the preset credibility threshold is taken as the interference feature area.

7. The method for constructing a three-dimensional model of an energy-saving and environmentally friendly factory building according to claim 1, characterized in that: The process of obtaining the modified scale space factor includes: The product of the negative correlation mapping value of the detail requirement degree and the influence degree of the feature point is used as the adjustment weight; the positive correlation mapping value of the product of the adjustment weight and the initial clarity degree is used as the modified scale space factor of each factory wall area.

8. A three-dimensional model building system for energy-saving and environmentally friendly factory buildings, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

Images