Automatic Determination Method for Color Harmony of Building Facades and Roofs Based on Point Cloud Data

Abstract

This invention discloses an automatic method for determining the color harmony of building facades and roofs based on point cloud data. The method includes the following specific steps: field data acquisition → photo preprocessing → point cloud data preprocessing → LAS building classification → building coverage area extraction → building color extraction and conversion → building color harmony determination, thus completing the automatic determination of the color harmony of building facades and roofs. This invention not only delineates color reference areas and color coordination areas, proposing different building color control methods for buildings within different ranges, and quickly identifies a list of buildings with color disharmony within a region, but also quickly identifies a list of buildings with color disharmony based on the harmony differences between each individual building and its surrounding neighboring buildings.

Description

Technical Field

[0001] This invention relates to the field of color research in urban areas, and more specifically to an automatic method for determining the color harmony of building facades and roofs based on point cloud data. Background Technology

[0002] Urban color is a systematic phenomenon. A complete urban color planning and design should analyze and plan all urban color components in a unified manner, determine the city's main color system or auxiliary color system, then determine the permanent and inherent base colors of various buildings and other objects, and then determine the mobile colors, including urban advertisements and public transportation, as well as the temporary colors, including street decorations and window decorations, etc., providing an important reference for urban landscape planning, construction and management.

[0003] An assessment of the color scheme of existing urban blocks typically includes:

[0004] 1. Number the buildings along the street in the city block; obtain street view images and base colors of the building walls;

[0005] 2. Construct a real-life image of the building's facade along the street from the street view images;

[0006] 3. Extract the color information from the actual street-facing facade of the building, and draw a color harmony analysis table based on the color information;

[0007] 4. Based on the judgment results, rectify the buildings in the urban blocks that have disharmonious colors.

[0008] The existing technology has the following shortcomings in the process of determining color harmony:

[0009] First, the scope of architectural color research is insufficient;

[0010] Due to the limitations of existing technologies in color recognition, research on urban colors is mainly limited to the street level with street view images. There is a lack of research on the colors of large urban areas from a macro perspective, as well as a lack of detailed research on the colors of individual buildings, their roofs, and walls from a micro perspective.

[0011] Secondly, the accuracy of architectural color extraction methods is insufficient;

[0012] Existing technologies have limitations in extracting building colors. Street view images can only identify the facade colors of buildings on both sides of the street and can only be applied to main and secondary roads with relatively wide road widths. They cannot obtain the color of each building, nor can they extract the color of the roof and walls of individual buildings. The completeness and accuracy of data collection are insufficient.

[0013] Third: The assessment of architectural color harmony is incomplete;

[0014] (1) Existing technology can only determine the degree of harmony of colors of adjacent buildings, and lacks an effective means to analyze the overall color harmony of buildings in a large scene.

[0015] (2) Current technical methods can only find discordant building units by checking each adjacent building one by one, and lack a method to quickly extract discordant buildings in one go. Summary of the Invention

[0016] This invention discloses an automatic method for determining the color harmony of building facades and roofs based on point cloud data. Its main purpose is to overcome the above-mentioned shortcomings and deficiencies of the existing technology.

[0017] The technical solution adopted in this invention is as follows:

[0018] An automatic method for determining the color harmony of building facades and roofs based on point cloud data, the method comprising the following specific steps:

[0019] Step 1: Field data collection, using drone photography to acquire photo data;

[0020] Step 2: Photo preprocessing. The original scene photos collected in Step 1 are processed through white balance correction to obtain images that are closer to those under natural light.

[0021] Step 3: Point cloud data preprocessing to form a point cloud database that can directly extract the vector contours of all buildings within the complete study area and identify their roofs and walls;

[0022] Step 4: Classify LAS buildings. After turning off ground points and noise in the LAS dataset, use the aggressive classification method in the LAS building classification tool, select the lower part of the roof to classify, and assign the classification code of the lower part of the roof to 32. In the calculation result, the points with the classification code of 6 are determined to be building roof points, and the points with the classification code of 32 are building wall points, thus completing the building classification.

[0023] Step 5: Extract building coverage areas. Extract the 2D building coverage areas from the LAS points of the buildings obtained in Step 4.

[0024] Step Six: Building Color Extraction and Conversion. For each LAS point, associate it with the building ID, building part classification, and conversion from RGB color space to Munsell color space to obtain the dominant roof color and dominant wall color of the building.

[0025] Step 7: Architectural color harmony assessment. First, assess the color harmony of the area, then assess the color harmony of the individual building with its neighboring buildings. Determine whether the tested building is harmonious with its surroundings, thus completing the automatic assessment of the color harmony of the building facade and roof.

[0026] Furthermore, the white balance correction process in step two includes the following specific methods:

[0027] (1) Use a fixed value. For an 8-bit image (0-255), take 128 as the gray value.

[0028] (2) Calculate the gain coefficient, and calculate the average value R of the three channels respectively. aver G aver B aver ,but

[0029]

[0030]

[0031] Among them, R k G k B k For the new pixel value, R i G i B i The original pixel values ​​are used to construct a new image.

[0032] Furthermore, the point cloud data preprocessing in step three includes the following specific steps:

[0033] (1) Generate LAS point cloud data based on photogrammetry principles;

[0034] (2) Create a LAS dataset. Using ArcGIS Pro 2.9 software, create a new LAS dataset in the folder and add multiple LAS datasets to the newly created LAS dataset.

[0035] (3) Classify LAS ground points. Use the LAS ground point classification tool in ArcGIS Pro 2.9 to classify the unassigned LAS point cloud into two categories: ground points and unassigned points.

[0036] (4) Generate DEM: By filtering out the ground points in the LAS point cloud, use the LAS dataset to raster tool to generate a digital elevation model (DEM);

[0037] (5) Classify LAS noise by using the relative height method from the ground in the LAS noise classification tool. The reference plane is the DEM generated in the previous step. When distinguishing abnormal high points, the absolute height method in the LAS noise classification tool is used. The maximum height is the highest point of the tallest building.

[0038] Furthermore, the specific steps for extracting the building coverage area in step five are as follows:

[0039] (1) Generate building grid and extract building coverage area. First, the building points are converted into grid data to obtain the location of the LAS building points.

[0040] (2) Generate building feature classes. To convert building rasters to polygons, you need to use the raster to polygon tool. The building cells in the raster will become polygons in the new layer.

[0041] (3) Clean the building surface, clean the holes in the building, and clean the boundaries of irregular shapes.

[0042] Furthermore, the specific content of cleaning the building surface includes the following three aspects:

[0043] (1) Some faces are too small to correspond to the actual buildings: Combine the images to determine the minimum building area and filter the faces that are larger than the minimum building area.

[0044] (2) Most faces contain holes: Use the face removal tool to remove them;

[0045] (3) Irregular building outlines: Use the Regularize Building Coverage tool to clean up the jagged boundaries of the extracted building surfaces and make them smoother.

[0046] Furthermore, the specific method for extracting and converting architectural colors in step six is ​​as follows:

[0047] (1) Extract the RGB values ​​of the LAS points of the roof and walls of each building unit: associate the classification of the building roof and walls in step four with the coverage area of ​​each building unit extracted in step five to obtain the building ID and building part classification (roof, wall) where the LAS point is located.

[0048] (2) Conversion of RGB color space to Munsell color system;

[0049] (3) Extract the dominant color of the roof and the dominant color of the wall of each building; through the above steps (1) and (2), each point cloud data has a building ID, building part classification (roof or wall), and Munsell HVC color value. Statistically analyze the distribution of HVC values ​​of the roof and wall of each building, and extract the top 5 color values ​​by number of points as the dominant color of the roof and the dominant color of the wall of that building.

[0050] Furthermore, the specific method for determining the color harmony of the area in step seven is as follows:

[0051] (1) The ideal color area within the area is manually delineated, that is, the area that can be referenced in the future renovation and construction process, and is defined as the "color benchmark area". Other areas are defined as "color coordination areas".

[0052] (2) Based on the Munsell color system, the hue (S), value (V), and chroma (C) of all buildings in the "color reference area" are statistically analyzed. After judging and eliminating extreme, erroneous, or unreasonable values ​​that are obviously inconsistent with the current situation, three relatively concentrated value distribution ranges are obtained. This result can be regarded as the "color reference range" for judging the color harmony of buildings in the area.

[0053] (3) Use the “color reference interval” to detect each building in the “color reference area”, extract the building that is not in the interval, and include it in the “color inharmonious” building list;

[0054] (4) Expand the brightness range of each hue in the "color reference range" by 1 and the chroma range by 2 to obtain the "color coordination range" for judging the color harmony of buildings in the "color coordination area".

[0055] (5) Use the “color coordination interval” to detect each building in the “color coordination area”, extract the building that is not in the interval, and add it to the “color disharmony” building list;

[0056] (6) Compile a list of buildings with disharmonious colors in the “Color Baseline Area” and “Color Coordination Area”, and complete the work of determining the degree of color harmony in the entire area and identifying problematic buildings.

[0057] Furthermore, the specific method for determining the color harmony between a single building and its adjacent buildings in step seven is as follows:

[0058] (1) Based on determining the hue, brightness and chroma of Munsell colors for each building wall and roof, take a single building as the main body and make it form n adjacent building combinations with all adjacent buildings. The harmony is detected by comparing colors in pairs.

[0059] (2) During the testing process, if the hue difference of all adjacent building combinations is within 8 (i.e., within 72° on the color wheel), the lightness difference is within 7, and the chroma difference is within 8, then the color of the tested building is considered to be in harmony with the surroundings, and the building is included in the list of color-harmonious buildings.

[0060] (3) If any one or more of the following three situations occur in adjacent building combinations: hue difference of 8 or more (i.e., a difference of 72° or more on the color wheel), lightness difference of 7 or more, or chroma difference of 8 or more, then it is necessary to continue to determine the proportion of the number of disharmonious combinations to the total number of combinations; if the proportion of the number of disharmonious combinations to the total number of combinations is less than 50%, then the tested building is determined to be harmonious with the surrounding colors and is included in the list of harmonious buildings; if the proportion of the number of disharmonious combinations to the total number of combinations is greater than or equal to 50%, then the tested building is determined to be disharmonious with the surrounding colors and is included in the list of disharmonious buildings.

[0061] As can be seen from the above description of the present invention, compared with the prior art, the advantages of the present invention are:

[0062] 1. Expanding the research scope of architectural color: Through four steps—field drone photography, aerial photo white balance processing, oblique photography, and the establishment of LAS point cloud models—a point cloud database is created that can directly extract the vector outlines of all buildings within the complete research area and identify their roofs and walls. Compared to previous techniques, this breakthrough extends the research scope from both sides of streets to the spatial level of entire areas, making color harmony determination more complete and comprehensive.

[0063] 2. Automatic identification of building roofs and walls: Through laser point cloud data, it can not only automatically generate building outlines and help identify the colors of individual buildings, but also automatically identify the roofs and walls of buildings based on algorithms. This enables precise control of building colors in urban design and provides technical support for the implementation of urban design color guidelines in urban renewal, the protection of historical buildings, and architectural design.

[0064] 3. Automatic extraction of building colors: Previous methods involved manual extraction of building colors, followed by conversion to RGB color system values, which was labor-intensive and had low accuracy. This invention utilizes LAS point cloud model dataset technology to automatically extract all building color data and convert it to the Munsell color system across the entire research area in a single step, significantly reducing labor costs and improving the scientific accuracy of color data extraction.

[0065] 4. Scientifically determine the color harmony of a region and identify problematic buildings, providing a method for determining the color harmony of buildings within a region and identifying problematic buildings. The research scope is divided into a "color baseline zone" and a "color coordination zone." Based on current data, calculations are performed to determine the hue, brightness, and chroma ranges of the "color baseline zone," while simultaneously identifying incompatible buildings within the "baseline zone." The numerical ranges of the "color baseline zone" are then expanded to a certain extent to generate the hue, brightness, and chroma ranges of the "color coordination zone," while simultaneously identifying incompatible buildings within the "coordination zone," thus completing the color harmony determination at the region level.

[0066] 5. Scientifically determine the harmony between adjacent individual buildings and provide a method for judging the harmony between adjacent buildings. Using a point cloud database, based on determining the hue, value, and chroma values ​​of each building's walls and roof using Munsell color theory, adjacent buildings are compared pairwise. A judgment criterion is proposed: "In the Munsell color system, a difference of 8 or more in hue (i.e., a difference of 72° or more on the color wheel), a difference of 7 in value, or a difference of 8 in chroma is considered a color incompatibility."

[0067] 6. Provide data support for urban renewal work in the area. In the analysis of the current problems in urban renewal work, this technical method scientifically analyzes and determines the buildings with color problems in the area from the perspective of urban color, providing a reference for the area's renewal work. In the planning and construction phase of urban renewal work, this technical method provides the color selection range for future new buildings in the area, providing technical support for the area's renewal and reconstruction.

[0068] 7. Automatic Algorithm Recognition and Judgment: Compared with existing technologies, this technology can quickly acquire the colors of individual building roofs and walls for street view color recognition, and can automatically and accurately identify them through algorithms. It requires minimal personnel, needing only 1 to 2 professional designers to extract the colors of building roofs and walls, thereby automatically judging the color harmony of buildings. The harmony judgment is more objective, not only analyzing the harmony of buildings within the area, but also judging the harmony level of a building by comparing its colors with adjacent buildings. The judgment method is more universal and unaffected by subjective factors. Attached Figure Description

[0069] Figure 1 This is a flowchart illustrating the present invention.

[0070] Figure 2 This is a schematic diagram of the point cloud data preprocessing process of the present invention.

[0071] Figure 3 This is a schematic diagram of the process for extracting the building coverage area according to the present invention.

[0072] Figure 4 This is a flowchart illustrating the process of determining the color harmony of a region in this invention.

[0073] Figure 5 This is a flowchart illustrating the process of determining the color harmony between a single building and its surrounding adjacent buildings according to the present invention. Detailed Implementation

[0074] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0075] An automatic method for determining the color harmony of building facades and roofs based on point cloud data, the method comprising the following specific steps:

[0076] Step 1: Field data collection, using drone photography to acquire photo data;

[0077] Before setting off for surveying, the specific weather conditions are checked, including cloud height, sunlight, and visibility, to avoid any unexpected events. Before the actual aerial surveying, the flight angle is adjusted, the flight path is set, and pre-flight tests are conducted. During flight, altitude, speed, and flight trajectory are monitored, and the number of aerial photographs taken is checked.

[0078] Step 2: Photo preprocessing. The original scene photos collected in Step 1 are processed through white balance correction to obtain images that are closer to those under natural light.

[0079] White balance is a crucial concept in television video recording. It's an indicator describing the accuracy of white generated by mixing the red, green, and blue primary colors in a display. It addresses a range of issues related to color reproduction and tone processing. Since this invention requires identifying the RGB colors of buildings, and RGB images are significantly affected by lighting conditions, white balance correction is necessary to make yellowish, bluish, and reddish images more closely resemble images taken under natural light. Common white balance algorithms for addressing lighting issues include mean white balance, perfect reflection algorithm, grayscale world algorithm, and dynamic thresholding algorithm.

[0080] This invention utilizes the Gray World Assumption (GW) algorithm. The GW algorithm is based on the Gray World Assumption, which states that for an image with numerous color variations, the average values ​​of the R, G, and B components tend to converge to the same gray value, Gray. Physically, the Gray World algorithm assumes that the average reflectance of light by objects in the natural world is a constant value overall, approximating "gray." The color balance algorithm forces this assumption onto the image to be processed, eliminating the influence of ambient light and obtaining the original scene image.

[0081] Algorithms typically use two methods to determine the Gray value:

[0082] (1) Use a fixed value. For an 8-bit image (0-255), take 128 as the gray value.

[0083] (2) Calculate the gain coefficient, and calculate the average value R of the three channels respectively. aver G aver B aver ,but

[0084]

[0085]

[0086] Among them, R k G k B k For the new pixel value, R iG i B i The original pixel values ​​are used to construct a new image.

[0087] Step 3: Point cloud data preprocessing to form a point cloud database that can directly extract the vector contours of all buildings within the complete study area and identify their roofs and walls;

[0088] Step 4: Classify LAS buildings. After disabling ground points and noise in the LAS dataset, use the aggressive classification method in the LAS building classification tool. Select the lower part of the roof and assign it a classification code of 32. In the calculation result, points with a classification code of 6 are identified as building roof points, and points with a classification code of 32 are identified as building wall points, thus completing the building classification. In the classification scheme specified by LAS file version 1.4 in the ArcGIS platform, points with a class value of 6 are building points, class values ​​of 0-18 are the predefined class codes in ASPRS, and class values ​​of 32-255 can be user-defined (supports LAS version 1.4). After classifying and disabling ground points and noise, building points can be classified from unassigned points.

[0089] Step 5: Extract building coverage areas. Extract the 2D building coverage areas from the LAS points of the buildings obtained in Step 4.

[0090] Step Six: Building Color Extraction and Conversion. For each LAS point, associate it with the building ID, building part classification, and conversion from RGB color space to Munsell color space to obtain the dominant roof color and dominant wall color of the building.

[0091] Step 7: Architectural color harmony assessment. First, assess the color harmony of the area, then assess the color harmony of the individual building with its neighboring buildings. Determine whether the tested building is harmonious with its surroundings, thus completing the automatic assessment of the color harmony of the building facade and roof.

[0092] Furthermore, the point cloud data preprocessing in step three includes the following specific steps:

[0093] (1) Generate LAS point cloud data based on photogrammetry principles;

[0094] Traditional aerial photography can only capture images of the ground from a vertical angle, while UAV oblique photogrammetry uses multiple sensors mounted on a single platform to simultaneously acquire images from multiple angles, effectively compensating for the shortcomings of traditional aerial photography. UAV oblique photogrammetry technology can provide various output formats such as 3D point clouds, 3D models, orthophotos, and digital surface models. LAS format point cloud data is one type of 3D point cloud format generated by UAV oblique photogrammetry.

[0095] A point cloud is a collection of points obtained after acquiring the spatial coordinates of each sampled point on the surface of an object; it is a set of data points in space. Point clouds obtained based on laser measurement principles include 3D coordinates (XYZ) and laser reflection intensity. Point clouds obtained based on photogrammetry principles include 3D coordinates (XYZ) and color information (RGB). Point clouds obtained by combining laser measurement and photogrammetry principles include 3D coordinates (XYZ), laser reflection intensity, and color information (RGB). Point cloud attributes include spatial resolution, point accuracy, and surface normal vectors. There are many storage formats for point clouds, including *.pts, *.asc, *.dat, *.stl, *.imw, *.xyz, and *.las.

[0096] LAS point clouds are a standard LiDAR data format developed by the LiDAR Committee of the American Society for Photogrammetry and Remote Sensing (ASPRS). The data structure consists of three parts: a common header area, a variable-length record area, and a point set record area. First, the common header area records general information about the LAS file, such as the acquisition date, the generating software, and the number of point records. The next section, the variable-length record area, contains metadata such as coordinate system information. Finally, the core of the LAS file is the point set record area, which records the X, Y, and Z coordinates of the points and their attributes. These three modules are closely connected and interconnected.

[0097] Point cloud data generated by UAV oblique photography is a point cloud obtained based on photogrammetry principles. To save space, the actual coordinates of the points are stored as long integers (X, Y, Z), and the decimal places are stored in the scaling factor field of the common file header.

[0098] X coordinate =(X record ×X scale )+X offset

[0099] Y coordinate =(Y record ×Y scale )+Y offset

[0100] Z coordinate =(Z record ×Z scale )+Z offset

[0101] In addition to coordinate information, it also includes Classification, R, G, and B (color) information. The RGB values ​​need to be interpreted from the point data. These fields will be used for building component classification and color analysis in this invention.

[0102]

[0103] (2) Create a LAS dataset. Using ArcGIS Pro 2.9 software, create a new LAS dataset in the folder and add multiple LAS data to the newly created LAS dataset. Since the drone has a large collection range and the research area is large, the data will be distributed in multiple LAS point clouds. Therefore, it is necessary to collect multiple LAS point cloud data into a single LAS dataset first.

[0104] (3) Classify LAS ground points. Use the LAS ground point classification tool in ArcGIS Pro 2.9 to classify the unassigned LAS point cloud into two categories: ground points and unassigned. The LAS point cloud data generated by UAV oblique photography is classified as unassigned. Before classifying building points, it is necessary to distinguish the ground points in the point cloud.

[0105] (4) Generate DEM: By filtering out ground points in the LAS point cloud, a Digital Elevation Model (DEM) is generated using the LAS dataset to raster tool. A Digital Elevation Model (DEM) is a physical ground model that represents the undulation of ground elevation using an ordered array of numerical values. The DEM represents ground elevation as a continuous raster. While LAS ground points provide a wealth of information about ground elevation, there is space between unknown elevation points, requiring the calculation of elevations in these unknown areas based on the existing ground point values.

[0106] (5) Classify LAS noise points. Using the relative height method from the ground in the LAS noise classification tool, with the reference plane being the DEM generated in the previous step, the absolute height method in the LAS noise classification tool is used to distinguish abnormally high points. The maximum height is the highest point of the tallest building. After identifying the LAS ground points, the next step is to determine the noise points among the remaining unassigned points. Noise points correspond to points that are abnormally high or low, which may be the result of random errors in the LiDAR data. Generally, points below the elevation of low-lying areas such as the ground or rivers are considered abnormally low points, while points above the elevation of the highest point of a building are generally considered abnormally high points.

[0107] Furthermore, the specific steps for extracting the building coverage area in step five are as follows:

[0108] (1) To generate a building raster and extract the building coverage area, the building points must first be converted into raster data to obtain the location of the LAS building points. Based on the building points with classification code 6, this invention uses the LAS point statistics tool used as raster data to search for LAS building points for each cell in the raster. If a building point exists, the value of the cell is the building class code (6); if no building point exists, the cell remains empty (NoData).

[0109] (2) Generate building feature classes. To convert building rasters to polygons, you need to use the raster to polygon tool. The building cells in the raster will become polygons in the new layer.

[0110] (3) Clean the building surface, clean the holes in the building, and clean the boundaries of irregular shapes.

[0111] Furthermore, the specific content of cleaning the building surface includes the following three aspects:

[0112] (1) Some faces are too small to correspond to the actual buildings: Combine the images to determine the minimum building area and filter the faces that are larger than the minimum building area.

[0113] (2) Most faces contain holes: Use the face removal tool to remove them;

[0114] (3) Irregular building outlines: Use the regularized building coverage area tool to clean up the jagged edges of the extracted building surfaces, making them smoother. After cleaning the building surfaces, assign a unique identifier ID to each building surface to distinguish each individual building.

[0115] Furthermore, the specific method for extracting and converting architectural colors in step six is ​​as follows:

[0116] (1) Extract the RGB values ​​of the LAS points of the roof and walls of each building unit: associate the classification of the building roof and walls in step four with the coverage area of ​​each building unit extracted in step five to obtain the building ID and building part classification (roof, wall) where the LAS point is located.

[0117] (2) Conversion of RGB color space to Munsell color system;

[0118] The Munsell Color System, invented in 1898 by American artist A. Munsell, is currently the most widely used method internationally for classifying and labeling the colors of object surfaces. Art and printing color education in my country is also based on the Munsell system. It describes and labels colors using three parameters: hue, value, and chroma. The Munsell Color System is a three-dimensional, spherical color solid model. The central axis represents value, with white at the north pole and black at the south pole. The hue ring around the central axis consists of 10 basic hues arranged clockwise, including 5 primary colors and 5 intermediate colors. The horizontal direction perpendicular to the central value axis represents the chroma value; the chroma value on the central axis is 0, and the outermost layer represents the most saturated hue. For example, the Munsell color chart 5YR7 / 6 represents a medium-high value (7) and medium chroma (6) true red-yellow (5YR).

[0119] Any color in nature can be produced by mixing and matching the three primary colors: red, green, and blue. Geometrically, this can be represented by a spatial coordinate system consisting of three mutually perpendicular axes: R, G, and B, which is called the RGB tristimulus space or RGB color space.

[0120] Although the conversion between Munsell color space and RGB is a non-linear process, it is impossible to directly establish an accurate mathematical model between them. Considering that a certain amount of conversion accuracy can be sacrificed, this invention is based on the rgb2mnsl function in the munsell package of R language. After testing, the conversion accuracy is within a controllable range.

[0121] (3) Extract the dominant color of the roof and the dominant color of the wall of each building; through the above steps (1) and (2), each point cloud data has a building ID, building part classification (roof or wall), and Munsell HVC color value. Statistically analyze the distribution of HVC values ​​of the roof and wall of each building, and extract the top 5 color values ​​by number of points as the dominant color of the roof and the dominant color of the wall of that building.

[0122] Furthermore, the specific method for determining the color harmony of the area in step seven is as follows:

[0123] (1) The ideal color area within the area is manually delineated, that is, the area that can be referenced in the future renovation and construction process, and is defined as the "color benchmark area". Other areas are defined as "color coordination areas".

[0124] (2) Based on the Munsell color system, the hue (S), value (V), and chroma (C) of all buildings in the "color reference area" are statistically analyzed. After judging and eliminating extreme, erroneous, or unreasonable values ​​that are obviously inconsistent with the current situation, three relatively concentrated value distribution ranges are obtained. This result can be regarded as the "color reference range" for judging the color harmony of buildings in the area.

[0125] (3) Use the “color reference interval” to detect each building in the “color reference area”, extract the building that is not in the interval, and include it in the “color inharmonious” building list;

[0126] (4) Expand the brightness range of each hue in the "color reference range" by 1 and the chroma range by 2 to obtain the "color coordination range" for judging the color harmony of buildings in the "color coordination area".

[0127] (5) Use the “color coordination interval” to detect each building in the “color coordination area”, extract the building that is not in the interval, and add it to the “color disharmony” building list;

[0128] (6) Compile a list of buildings with disharmonious colors in the “Color Baseline Area” and “Color Coordination Area”, and complete the work of determining the degree of color harmony in the entire area and identifying problematic buildings.

[0129] Furthermore, the specific method for determining the color harmony between a single building and its adjacent buildings in step seven is as follows:

[0130] (1) Based on determining the hue, brightness and chroma of Munsell colors for each building wall and roof, take a single building as the main body and make it form n adjacent building combinations with all adjacent buildings. The harmony is detected by comparing colors in pairs.

[0131] (2) During the testing process, if the hue difference of all adjacent building combinations is within 8 (i.e., within 72° on the color wheel), the lightness difference is within 7, and the chroma difference is within 8, then the color of the tested building is considered to be in harmony with the surroundings, and the building is included in the list of color-harmonious buildings.

[0132] (3) If any one or more of the following three situations occur in adjacent building combinations: hue difference of 8 or more (i.e., a difference of 72° or more on the color wheel), lightness difference of 7 or more, or chroma difference of 8 or more, then it is necessary to continue to determine the proportion of the number of disharmonious combinations to the total number of combinations; if the proportion of the number of disharmonious combinations to the total number of combinations is less than 50%, then the tested building is determined to be harmonious with the surrounding colors and is included in the list of harmonious buildings; if the proportion of the number of disharmonious combinations to the total number of combinations is greater than or equal to 50%, then the tested building is determined to be disharmonious with the surrounding colors and is included in the list of disharmonious buildings.

[0133] As can be seen from the above description of the present invention, compared with the prior art, the advantages of the present invention are:

[0134] 1. Expanding the research scope of architectural color: Through four steps—field drone photography, aerial photo white balance processing, oblique photography, and the establishment of LAS point cloud models—a point cloud database is created that can directly extract the vector outlines of all buildings within the complete research area and identify their roofs and walls. Compared to previous techniques, this breakthrough extends the research scope from both sides of streets to the spatial level of entire areas, making color harmony determination more complete and comprehensive.

[0135] 2. Automatic identification of building roofs and walls: Through laser point cloud data, it can not only automatically generate building outlines and help identify the colors of individual buildings, but also automatically identify the roofs and walls of buildings based on algorithms. This enables precise control of building colors in urban design and provides technical support for the implementation of urban design color guidelines in urban renewal, the protection of historical buildings, and architectural design.

[0136] 3. Automatic extraction of building colors: Previous methods involved manual extraction of building colors, followed by conversion to RGB color system values, which was labor-intensive and had low accuracy. This invention utilizes LAS point cloud model dataset technology to automatically extract all building color data and convert it to the Munsell color system across the entire research area in a single step, significantly reducing labor costs and improving the scientific accuracy of color data extraction.

[0137] 4. Scientifically determine the color harmony of a region and identify problematic buildings, providing a method for determining the color harmony of buildings within a region and identifying problematic buildings. The research scope is divided into a "color baseline zone" and a "color coordination zone." Based on current data, calculations are performed to determine the hue, brightness, and chroma ranges of the "color baseline zone," while simultaneously identifying incompatible buildings within the "baseline zone." The numerical ranges of the "color baseline zone" are then expanded to a certain extent to generate the hue, brightness, and chroma ranges of the "color coordination zone," while simultaneously identifying incompatible buildings within the "coordination zone," thus completing the color harmony determination at the region level.

[0138] 5. Scientifically determine the harmony between adjacent individual buildings and provide a method for judging the harmony between adjacent buildings. Using a point cloud database, based on determining the hue, value, and chroma values ​​of each building's walls and roof using Munsell color theory, adjacent buildings are compared pairwise. A judgment criterion is proposed: "In the Munsell color system, a difference of 8 or more in hue (i.e., a difference of 72° or more on the color wheel), a difference of 7 in value, or a difference of 8 in chroma is considered a color incompatibility."

[0139] 6. Provide data support for urban renewal work in the area. In the analysis of the current problems in urban renewal work, this technical method scientifically analyzes and determines the buildings with color problems in the area from the perspective of urban color, providing a reference for the area's renewal work. In the planning and construction phase of urban renewal work, this technical method provides the color selection range for future new buildings in the area, providing technical support for the area's renewal and reconstruction.

[0140] 7. Automatic Algorithm Recognition and Judgment: Compared with existing technologies, this technology can quickly acquire the colors of individual building roofs and walls for street view color recognition, and can automatically and accurately identify them through algorithms. It requires minimal personnel, needing only 1 to 2 professional designers to extract the colors of building roofs and walls, thereby automatically judging the color harmony of buildings. The harmony judgment is more objective, not only analyzing the harmony of buildings within the area, but also judging the harmony level of a building by comparing its colors with adjacent buildings. The judgment method is more universal and unaffected by subjective factors.

[0141] This invention innovatively utilizes LAS point cloud data from oblique UAV photography. The method boasts advantages such as fast point cloud data acquisition, broad research scope, and complete and accurate research objects. It can acquire relatively complete basic information about building rooftops that cannot be covered by street view images and perspectives invisible to humans. The LAS point cloud data is evenly and densely distributed on building surfaces, resulting in a large and objective sample size for building recognition and color extraction.

[0142] Traditional methods for building color analysis rely on street view images to identify building colors at the block scale. However, these methods are limited by the timeliness of street view image capture and cannot accurately identify the color of each individual building, limiting their application to macro- and meso-level urban color analysis. This invention addresses both of these issues. It innovatively utilizes LAS point cloud data generated by UAV oblique photography, analyzing building colors based on current conditions, ensuring high timeliness. Point cloud data categorized as buildings is extracted based on their spatial distribution characteristics with other features. This data is then converted to raster data and further transformed into building vector outlines to distinguish the point cloud data for each building. Furthermore, the algorithm differentiates building walls and roofs based on their location within the building. This invention achieves automatic and accurate identification of analysis objects, expanding the analysis scope from the block scale to different parts of each individual building.

[0143] Traditional color extraction techniques are inefficient and inaccurate. Manually extracting building roof and wall colors using color pickers in software like Photoshop is inefficient and labor-intensive. Deep learning algorithms for extracting streetscape wall colors only identify the overall color of buildings at the block scale, failing to accurately identify the colors of individual parts of each building. This technology, using LAS point cloud data classification algorithms, can quickly and accurately identify different parts of buildings, with rich and uniform sampling points, unaffected by subjective color selection, resulting in a more objective assessment of building harmony.

[0144] Based on the technological advantages of this invention, the dominant colors of the roof and walls of individual buildings can be quickly obtained, which is a microscale that previous inventions could not accurately identify. It is not only applicable to the macro-level color control in urban design of a district, but also to the precise control of the color harmony of each individual building, so that urban design color control can be better implemented in actual building design and streetscape management.

[0145] This invention provides a more objective and comprehensive assessment of architectural color harmony than existing technologies. It not only delineates color benchmark areas and color coordination areas, but also proposes different architectural color control methods for buildings within different ranges, quickly identifying a list of buildings with disharmonious colors in a region. Furthermore, it quickly identifies a list of buildings with disharmonious colors based on the harmony differences between each individual building and its surrounding neighboring buildings.

[0146] The above are merely specific embodiments of the present invention, but the design concept of the present invention is not limited thereto. Any non-substantial improvements made to the present invention using this concept should be considered as infringing on the protection scope of the present invention.

Claims

1. An automatic method for determining the color harmony of building facades and roofs based on point cloud data, characterized in that: The determination method includes the following specific steps: Step 1: Field data collection, using drone photography to acquire photo data; Step 2: Photo preprocessing. The original scene photos collected in Step 1 are processed through white balance correction to obtain images that are closer to those under natural light. Step 3: Point cloud data preprocessing to form a point cloud database that can directly extract the vector contours of all buildings within the complete study area and identify their roofs and walls; Step 4: Classify LAS buildings. After turning off ground points and noise in the LAS dataset, use the aggressive classification method in the LAS building classification tool, select the lower part of the roof to classify, and assign the classification code of the lower part of the roof to 32. In the calculation result, the points with the classification code of 6 are determined to be building roof points, and the points with the classification code of 32 are building wall points, thus completing the building classification. Step 5: Extract building coverage areas. Extract the 2D building coverage areas from the LAS points of the buildings obtained in Step 4. Step Six: Architectural Color Extraction and Conversion. For each LAS point, associate it with the building ID, building part classification, and conversion from RGB color space to Munsell color space to obtain the dominant color of the building's roof and walls. Step 7: Architectural color harmony assessment. First, assess the color harmony of the area, then assess the color harmony between the individual building and its neighboring buildings. Determine whether the building's colors are harmonious with the surrounding environment, and complete the automatic assessment of the color harmony of the building facade and roof.

2. The method for automatically determining the color harmony of building facades and roofs based on point cloud data according to claim 1, characterized in that: The white balance correction process in step two includes the following specific methods: (1) Use a fixed value. For an 8-bit image, take 128 as the grayscale value. (2) Calculate the gain coefficient and calculate the average value of the three channels respectively. , , ,but , , , ,in, , , For the new pixel value, , , The original pixel values ​​are used to construct a new image.

3. The method for automatically determining the color harmony of building facades and roofs based on point cloud data according to claim 1, characterized in that: The point cloud data preprocessing in step three includes the following specific steps: (1) Generate LAS point cloud data based on photogrammetry principles; (2) Create a LAS dataset. Using ArcGIS Pro 2.9 software, create a new LAS dataset in the folder and add multiple LAS datasets to the newly created LAS dataset; (3) Classify LAS ground points. Use the LAS ground point classification tool in ArcGIS Pro 2.9 to classify the unassigned LAS point cloud into two categories: ground points and unassigned points. (4) Generate DEM: By filtering out the ground points in the LAS point cloud, use the LAS dataset to raster tool to generate a digital elevation model (DEM); (5) Classify LAS noise by using the relative height method of the LAS noise classification tool. The reference plane is the DEM generated in the previous step. When distinguishing abnormal high points, the absolute height method of the LAS noise classification tool is used. The maximum height is the highest point of the tallest building.

4. The method for automatically determining the color harmony of building facades and roofs based on point cloud data according to claim 1, characterized in that: The specific steps for extracting the building coverage area in step five are as follows: (1) Generate building grid and extract building coverage area. First, the building points are converted into grid data to obtain the location of the LAS building points; (2) Generate building feature classes. To convert building rasters to polygons, you need to use the raster to polygon tool. The building cells in the raster will become polygons in the new layer. (3) Clean the building surface, clean the holes in the building, and clean the irregularly shaped boundaries.

5. The method for automatically determining the color harmony of building facades and roofs based on point cloud data according to claim 4, characterized in that: The specific content of cleaning the building surface includes the following three aspects: (1) Some faces are too small to correspond to the actual buildings: Combine the images to determine the minimum building area and filter the faces that are larger than the minimum building area; (2) Most faces contain holes: use the face removal tool to remove them; (3) Irregular building outline: Use the Regularize Building Coverage tool to clean up the jagged boundaries of the extracted building surfaces to make them smoother.

6. The method for automatically determining the color harmony of building facades and roofs based on point cloud data according to claim 1, characterized in that: The specific methods for extracting and converting architectural colors in step six are as follows: (1) Extract the RGB values ​​of the LAS points of the roof and walls of each building unit: associate the classification of the building roof and walls in step four with the coverage area of ​​each building unit extracted in step five to obtain the building ID and building part classification of the LAS point. (2) Conversion from RGB color space to Munsell color system; (3) Extract the dominant color of the roof and the dominant color of the walls of each building; Through the above steps (1) and (2), each point cloud data has a building ID, building part classification, and Munsell HVC color value. The distribution of HVC values ​​of the roof and walls of each building is statistically analyzed, and the top 5 color values ​​are extracted as the dominant color of the roof and the dominant color of the walls of that building.

7. The method for automatically determining the color harmony of building facades and roofs based on point cloud data according to claim 1, characterized in that: The specific method for determining the color harmony of the area in step seven is as follows: (1) The ideal color area within the area is manually delineated, that is, the area that can be referenced in the future renovation and construction process, and is defined as the "color benchmark area". Other areas are defined as "color coordination areas". (2) Based on the Munsell color system, the hue S, lightness V and chroma C of all buildings in the "color reference area" are statistically analyzed. After judging and eliminating extreme, erroneous or unreasonable values ​​that are obviously inconsistent with the current situation, three relatively concentrated value distribution ranges are obtained. This result can be regarded as the "color reference range" for judging the color harmony of buildings in the area. (3) Use the "color reference interval" to detect each building in the "color reference area", extract the building that is not in the interval, and include it in the "color inharmonious" building list; (4) Expand the lightness range of each hue in the "color reference range" by 1 and the chroma range by 2 to obtain the "color coordination range" for judging the color harmony of buildings in the "color coordination area"; (5) Use the "color coordination interval" to detect each building in the "color coordination area", extract the building that is not in the interval, and include it in the "color disharmony" building list; (6) Compile a list of buildings with disharmonious colors in the "color benchmark area" and "color coordination area", and complete the work of judging the degree of color harmony in the entire area and identifying problematic buildings.

8. The method for automatically determining the color harmony of building facades and roofs based on point cloud data according to claim 1, characterized in that: The specific method for determining the color harmony between a single building and its adjacent buildings in step seven is as follows: (1) Based on the determination of the three values ​​of hue, brightness and chroma of Munsell color for each building wall and roof, take a single building as the main body and make it form n adjacent building combinations with all adjacent buildings. The harmony is detected by comparing the colors in pairs. (2) During the testing process, if the hue difference of all adjacent building combinations is within 8 (i.e., within 72° on the color wheel, within 7 in brightness, and within 8 in chroma), the building is considered to be in harmony with the surrounding environment and is included in the list of color-harmonious buildings. (3) If any one or more of the following three situations occur in adjacent building combinations: hue difference of 8 or more (i.e., a difference of 72° or more on the color wheel), lightness difference of 7 or more, or chroma difference of 8 or more, then it is necessary to continue to determine the proportion of the number of disharmonious combinations to the total number of combinations; if the proportion of the number of disharmonious combinations to the total number of combinations is less than 50%, then it is determined that the tested building is harmonious with the surrounding colors and the building is included in the list of harmonious buildings. If the number of discordant combinations accounts for more than or equal to 50% of the total number of combinations, the building being tested is determined to be incompatible with the surrounding colors, and the building is included in the list of buildings with incompatible colors.

Images