Method for determining the construction progress of a building

Abstract

The invention relates to a method for determining the construction progress of a building, wherein the method is carried out by a computer on the basis of measurement data optically captured from the building, the measurement data is prepared as a point cloud having a horizontal plane as a sheet plane and a vertical axis perpendicular to the sheet plane, and the method has at least the following steps in the stated order: a. creating at least one section on the horizontal or vertical plane from the point cloud; b. detecting point-free regions in each of the sections; and c. creating a border around each detected point-free region. By means of the method proposed here, the exact shape and position of a building structure can be determined using simple means with the aid of a computer on the basis of an optically created point cloud.

Description

[0001] 202413711

[0002] 1

[0003] Description

[0004] Method for determining the construction status of a building

[0005] The invention relates to a method for determining the construction status of a building.

[0006] For many applications, it is desirable to determine the exact construction status of a building. The actual construction status often differs considerably from the (latest) planning stage. Determining the construction status is particularly challenging for older buildings and those undergoing one or more retrofits, restorations, or conversions. Previous approaches, which involve converting hand-drawn architectural plans (especially floor plans) into computer-aided formats—a process that is also (at least partially) manual and usually carried out extensively by service providers—lead to significant uncertainty regarding the details of the respective construction status.

[0007] One solution involves using optical measuring instruments (such as laser scanners or visual camera recordings) to capture the current state of construction and then creating a three-dimensional simulation environment using computer-aided object recognition. From this, conventional representations for architects, civil engineers, and building physicists are generated, such as a floor plan. A common challenge is that the buildings are often occupied and / or contain objects on a floor that are not part of the building structure (such as columns, walls, and shafts). These building structures must be reliably identified. Furthermore, there are usually extensions (such as cladding), building services (such as ventilation ducts, sprinkler systems, and electrical wiring), and other obstructed objects (such as security systems and signage) that can obscure the view of the building structure.In addition to object detection, it is also necessary to determine how the building structure hidden by it might actually run.

[0008] Based on this, the present invention aims to overcome, at least partially, the disadvantages known from the prior art. The features of the invention are defined in the independent claims, for which advantageous embodiments are shown in the dependent claims. The features of the claims can be combined in any technically meaningful way, taking into account the explanations in the following description as well as features from the [202413711]

[0009] 2

[0010] Figures can be added which include supplementary embodiments of the invention.

[0011] The invention relates to a method for determining the construction status of a building, wherein the method is executed by a computer based on optically recorded measurement data of the building, wherein the measurement data are processed as a point cloud with a horizontal plane as the sheet plane and a height axis perpendicular to the sheet plane.

[0012] wherein the method comprises at least the following steps in the order mentioned: a. Creating at least one section in the horizontal or vertical plane from the point cloud;

[0013] b. Capturing point-free areas in each of the sections; and

[0014] c. Creating a border around each captured point-free area.

[0015] Unless explicitly stated otherwise, ordinal numbers used in the preceding and following descriptions serve solely for unambiguous differentiation and do not indicate any order or ranking of the components referred to. An ordinal number greater than one does not necessarily imply the presence of another such component.

[0016] It should first be noted that while the method proposed here can also be used for object recognition, it does not necessarily lead to more efficient use given the multitude of established software tools. It is therefore recommended to perform (e.g., conventional) object recognition in addition to this method, or to use a different method for other tasks. For example, glass panes are at least partially transparent to most optical measurement methods, so measurement results, i.e., measurement points, are generated inside an object (e.g., a motor vehicle). It is possible that such anomalies could be automatically detected with high reliability in an independent step or process.

[0017] For example, the present method reveals that these measurement points appear to float in space because they are surrounded by areas without points. However, it should be noted that the strength of the proposed method lies precisely in the fact that such objects, due to their lack of continuity along the vertical axis (i.e., they generally do not extend to the floor slab), are recognized as objects that therefore cannot be building structures. 202413711

[0018] 3

[0019] For the sake of completeness, it is important to first explain how a point cloud, or rather the optically recorded measurement data, is generated. Using an optical measuring device, such as a laser scanner or a camera, a large number of measurement points are determined, with such measurement data being optically captured from as many different locations as possible. For example, a laser measurement is recorded as a vehicle or person moves through the building in question. This reliably captures surface-mounted installations (such as heating pipes, electrical rails, or ventilation ducts) from the side and, if necessary, from the side towards a wall or support (i.e., the building structure being analyzed). This results in at least some measurement points between such a surface-mounted installation and the building structure to which it is attached.This is particularly relevant for commercial buildings, such as factory halls or warehouses. In such buildings, aesthetic measures, such as concealed (i.e., plastered) cladding of installations, are generally avoided for cost reasons. The method proposed here is therefore particularly suitable for this purpose. Alternatively or additionally, such concealed, surface-mounted installations are automatically identified (for example, due to deviations, a characteristic shape, and / or a lack of continuity between the floor and ceiling) and / or (manually or automatically) compared with existing plans and (preferably automatically) disregarded in step c. A more detailed explanation follows.

[0020] Finally, it should be noted that such point clouds represent a very large amount of data, which even today pushes commonly used computers (or their processors and / or main memory) to their limits when processing, especially when neural networks are used to process the point cloud (for example, for object recognition).

[0021] Prior to the procedure, a point cloud is generated and used, which is spatially oriented in a natural orientation (i.e., pointing downwards with the Earth's gravity). It is not strictly necessary for the point cloud to be displayed in this orientation on a graphical user interface (e.g., a screen) or even to be displayed to a user at all. Rather, it is assigned and aligned to a (preferably Cartesian) coordinate system. This information is preferably obtained as metadata supplied with the measurement data (e.g., from a gravity sensor of the optical measuring device). Alternatively or additionally, the orientation is determined using old maps. A geodetic 202413711 is also preferred.

[0022] 4

[0023] The orientation is set, preferably also using supplied meta-information (for example, from a GPS sensor and / or compass of the optical measuring device). In one embodiment, this step is performed as part of this method and / or as known from conventional methods, for example, for object detection in a building.

[0024] This method preferably reduces the information content of the point cloud or its individual measurement points to coordinate information (i.e., the coordinates of a measurement point in the aligned coordinate system), unless this has already been done beforehand or during the creation of the measurement data. Colors, shapes, times, and other information are not considered in this method or are only considered in a subsequent process. Alternatively or additionally, time-dependent measurement points (e.g., of a moving object) are removed, provided they can be identified with (at least sufficient) certainty given spatial overlap and knowledge of the relative time of the measurement. It should be noted, however, that this is also not necessary for the present method or, in most cases, has no influence on the result.

[0025] In step a., a section in the horizontal plane (hereinafter referred to as a horizontal section or section in the xy-plane) or a section in a vertical plane (hereinafter referred to as a vertical section) is created from this aligned point cloud. It should be noted that a horizontal section shows the Earth's gravitational field running perpendicularly, i.e., normal to the image plane in a graphical representation. In one embodiment, the horizontal section represents the entire point cloud viewed from the direction of the Earth's gravitational field, i.e., looking at the horizontal plane. In a graphical representation, this creates the impression of a more or less dense fog of measurement points in an area (containing measurement points) of the point cloud, with anomaly-like patches within this fog, which are point-free areas in this horizontal section.Such point-free areas could only arise if there is not a single measuring point along the vertical axis (i.e., the Earth's gravitational field).

[0026] Vertical sections can also be created along the x-axis or y-axis (i.e., in the horizontal plane), which then lie in the zy-plane or zx-plane, or in a plane angled to it, in which the z-axis (i.e., the Earth's gravitational field) runs. Such a vertical section can be used to identify and reconstruct horizontally running beams and pipes analogously to the method described above. 202413711

[0027] 5

[0028] In step b., these (usually multiple or large) dot-free areas are automatically detected. Numerous software methods are already known for this purpose, for example, those that can detect pixel densities. Here are five examples of software methods that can be used individually or in combination:

[0029] 1. Edge Detection, Functionality: This method identifies areas where pixel density changes abruptly by analyzing the differences in brightness of neighboring pixels. Well-known algorithms include the Sobel detector and the Canny edge detector.

[0030] 2. Histogram Analysis, Functionality: This method analyzes the distribution of brightness or color values ​​of the pixels in an image. Areas with different pixel densities are reflected in the distribution of pixel values, which are displayed in a histogram.

[0031] 3. Segmentation using clustering algorithms, how it works: Algorithms such as K-means or the mean-shift algorithm group pixels with similar pixel density. This method divides the image into areas with similar pixel density by grouping pixels with similar brightness or color values ​​into clusters.

[0032] 4. Wavelet Transformation, Functionality: This method decomposes the image into different frequency bands. Areas with different pixel densities are represented in different frequency bands in the wavelet transformation, which enables the detection and analysis of pixel density in different image areas.

[0033] Finally, in step c, a border is drawn around each such point-free area. This data can now be used without the underlying point cloud, for example, to create a floor plan for use by an architect or other trades. It should be noted that detailed sections (usually in a side view, i.e., along the vertical axis or the Earth's gravitational field in the image plane) and side views can also be generated from this data. In one embodiment, this data is supplemented with an existing plan or with other methods (e.g., object recognition), for example, to identify or spatially correctly reference windows and / or fixtures. In one embodiment, this is done using steps b and c.Created (prominent) corners or offsets in the surrounding border are used to improve spatially correctly referenced display. 202413711.

[0034] 6

[0035] In one embodiment, a minimum size of a point-free area is specified; for example, a minimum size has an area larger than 1 cm². 2 [one square centimeter] to 5 cm 2 In one embodiment, only those point-free areas with a surrounding border are provided, to which there is an approximate spatial relationship to an old plan state, wherein preferably a point-free area discovered in step b. and step c. that deviates from this plan state is displayed in the result (preferably marked as an anomaly, for example by color).

[0036] Often, a building structure (according to the plan) is created in a simple geometric shape or one that is already known (for example, from old plans). In one embodiment, a surrounding outline is always created using a simple geometric shape, such as a rectangle or oval, starting from an edge or the center of the pointless area. The geometric shape is extended until each edge intersects a point. Preferably, the orientation in the section is arbitrary, i.e., not along a predetermined grid (for example, based on an old plan) and / or a cardinal direction. In one embodiment, deviations of the edge profile from the simple geometric shape are additionally indicated in the result (preferably as an anomaly, for example, by color).

[0037] Numerous software methods are already known, for example, those that can detect and draw edges. Here are five examples of such software methods (partially overlapping with step b.), which can be used individually or in combination: 1. Sobel detector, Function: The Sobel detector uses filters to calculate the brightness differences in the horizontal and vertical directions. These differences are combined to detect and draw the edges in the image.

[0038] 2. Canny Edge Detector, Functionality: The Canny detector uses several steps to detect edges: image smoothing, gradient calculation, non-maxima suppression, and hysteresis thresholding. These steps help to detect and mark both precise and continuous edges.

[0039] 3. Prewitt Detector, Function: The Prewitt detector is similar to the Sobel detector, but uses simpler filters to calculate brightness differences. These differences are used to detect and mark the edges in the image.

[0040] 4. Laplacian Detector, Functioning: The Laplacian detector uses the second derivative to detect edges. This method is sensitive to noise, but can detect edges that might be missed by other methods.

[0041] 5. Roberts detector, how it works: The Roberts detector uses filters to... 202413711

[0042] 7

[0043] The process involves calculating brightness differences in a diagonal direction. These differences are used to detect and mark the edges in the image. This method is simple, but sensitive to noise.

[0044] The method proposed here allows for the quick and easy generation of a floor plan of the building to be surveyed, using a complex (and potentially unprocessed) point cloud of optically acquired measurement data, with very little computational effort. Objects that do not belong to the building structure are automatically removed because they are part of the point cloud (i.e., the area of ​​the fog). In some cases, it is useful or sufficient to roughly define the building structure because the plan itself is sufficiently accurate, but, for example, the geodetic orientation of the building is not known with certainty (e.g., the building was constructed slightly rotated compared to the plan).By reliably and sufficiently accurately recognizing the building structure and also by quickly and easily generating a file with a small storage size (e.g., easily compressible) and / or processing size (e.g., as a so-called vector file instead of a pixel file) (if necessary manually, preferably by means of a computer calculation) and aligning an old plan version (e.g., as a floor plan file) and this data (e.g., as a floor plan file, preferably in the same file format) with each other.

[0045] In an advantageous embodiment of the method, it is further proposed that, after step b. and before or in addition to step c., in an additional step d., a transition sharpness between a point-free area and an area with the point cloud is detected and compared with a predetermined limit value, and

[0046] when the predetermined limit is exceeded:

[0047] a.' Creating several horizontal sub-sections from the horizontal section, wherein in step c. an analogous additional step c.' is performed for each sub-section to create a border around a respective captured point-free area in the respective sub-section,

[0048] wherein preferably after step b. and before or in addition to step c.' in an additional step d.', a transition sharpness between a point-free area and an area with the point cloud is detected and compared with the predetermined limit value, and

[0049] where preferably if the predetermined limit is exceeded in one of the intermediate cuts before, during or after additional step c.':

[0050] a." Creating multiple horizontal intermediate sections from the intermediate section in question, and 202413711

[0051] 8

[0052] Repeat the process until the predetermined limit for transition sharpness in the subsequently generated intermediate cuts is undercut and / or a predetermined cutting height limit is reached.

[0053] It should be noted that the previously described method is already sufficiently accurate for many applications. However, in some applications, a few centimeters or even millimeters are crucial, for example, for the utilization of the available floor space. In such cases, simply determining the location and approximate orientation of a building structure is insufficient. If, for instance, a column was cast from several stacked blocks or in sections using formwork over the floor height, the resulting cross-section (horizontal in this example) will show unclean or blurred transitions. It is also not uncommon for columns to be modified over the course of a building's history, for example, by adding a bevel or chamfer over a section of the floor. This is done, for instance, to make it possible to accommodate a specific machine or storage rack within the building in a permissible or efficient manner.Blurred transitions also occur with inclined or curved support elements. For the latter, a section plane with a normal parallel (possibly not along a straight line) to the extent of the respective (inclined or curved) support element, specifically between its two ends, is advantageous. This building structure is preferably detectable automatically (possibly upon human intervention). This information can be obtained, for example, from an old plan or was added to the metadata during surveying (by a person). The procedure can then be carried out analogously for this part.

[0054] The limit value is, for example, already implemented in the software method used to detect point-free areas and / or to generate the desired boundary. In one embodiment, instead of a single edge, an edge array, preferably with a minimum and a maximum edge, is output, so that the uncertainty about the execution status is visualized and quickly grasped by a person in a graphical representation. In some applications, this potential deviation is irrelevant, while in others (e.g., concerning the same building) it is important. In such cases, a more detailed analysis of the execution status can be performed (e.g., triggered by human input), preferably as described above.

[0055] described.202413711

[0056] 9

[0057] It should be noted that, for example, a manufacturing machine has a height of, say, 2 m [two meters], and therefore a precise analysis of a building structure in its vicinity must be conducted exclusively within this height range. This is specified, for example, as a cutting height limit or as a specification for a (geometric) window to be considered. This reduces the computational effort to the relevant areas. Such a specification is also possible for an entire building or a floor of a building. However, it should be noted that, preferably in a first step, the entire floor height, or slightly more than the maximum height of an object, is always used as the height of the horizontal section to ensure that only building structures result in point-free areas.

[0058] In one embodiment, a horizontal cut is divided into intermediate cuts in a fixed sequence, for example, always halved, in order to generate multiple intermediate cuts. In another embodiment, additional information is used, such as a known formwork height or construction height of a block of several blocks or of (for example, brick) stones which, when stacked on top of each other, form a support.

[0059] It should be noted that a blurring of the transition between the point-free area and the point cloud area always occurs when the building structure is offset, inclined, or curved in different (imaginary) planes of the section. Another cause is a locally present object (i.e., in an imaginary plane), possibly partially translucent. However, an object generally differs from the previous case in that it is also limited in the extent of its respective (outer) edge, such as a surface-mounted (waste)water pipe with a significantly smaller cross-section than a building column. In the case of a continuous object (e.g., spanning multiple floors), such as a downpipe, a point-free area immediately follows a blurring of the transition (due to a small gap at not every height in the horizontal section).Due to its shape, it is easily identifiable or is accepted as an anomaly creating a surrounding border. Such an object, mounted on plaster, is then easily classifiable and / or easily distinguishable from a building structure due to its shape and size (both manually and / or automatically).

[0060] In an advantageous embodiment, this process of dividing the cut into several intermediate cuts is repeated until a predetermined cutting height limit, i.e., a sufficient resolution (for example, in heights of at least 5 cm [five centimeters]202413711) is reached.

[0061] 10

[0062] up to 20 cm), and / or a sufficiently low level of blur is achieved when transitioning from the point-free area to the point cloud area in the respective intermediate section.

[0063] In a further advantageous embodiment of the method, it is proposed that object recognition is also performed to automatically classify objects in the building in question.

[0064] with a preference given to removing the detected and classified objects.

[0065] As mentioned previously, the method proposed here is ideally suited for many applications when combined with object recognition. For example, when measuring an actively used hall, where many objects are therefore scattered across the floor, it is advantageous to automatically classify these objects and, if necessary, remove them for further use or place them on a layer so that they can be easily selected (preferably all at once) or selected and hidden.

[0066] However, in an application, objects such as surface-mounted installations may also be recognizable. Using the method proposed here, these objects are more easily distinguishable from the building structure, for example, due to their size (minimum size of a point-free area to represent a building structure) and / or shape (e.g., a wired cable) and / or a lack of continuity along the vertical axis (e.g., an electrical installation) on the considered floor of the building or within the horizontal section. The same applies accordingly to a vertical section with a suitable orientation.

[0067] In a further advantageous embodiment of the method, it is proposed that in step e. the border created in step c. is extruded over the entire height of the relevant cut or intermediate cut.

[0068] The aim of this method is to obtain a replica of the point-free area as a three-dimensional parametric geometry based on the detected boundaries.

[0069] Building structures, especially in production halls and warehouses, usually run vertically (i.e., along the vertical axis or the Earth's gravitational field) across the entire horizontal section. The information that a horizontal section is pointless means that it is very likely such a vertical building structure. The other issue is what constitutes an inner and an outer area of ​​the point cloud 202413711.

[0070] 11

[0071] This is not an issue in this context. The information provided by the generated outline is sufficient to use this building structure in a three-dimensional representation or rendering by extruding the outline. The same applies to a vertical section with a suitable orientation.

[0072] If an imprecision is detected at the transition and intermediate cuts have been made, the surrounding border is preferably extruded over the height of each intermediate cut. The extruded sections of this border are then preferably joined together, preferably to form a closed three-dimensional structure. The specific spatial connection is either known due to the position of the intermediate cuts, the imprecision, or other information (for example, the design of the structure in question), or can be approximated with sufficient accuracy for the respective application by means of a suitable assumption.

[0073] In a further advantageous embodiment of the method, it is proposed that the at least one horizontal section be a floor section, preferably directly adjacent, between the floor and the floor slab.

[0074] When a point cloud is created, or when the optically acquired measurement data is generated for this purpose, a natural boundary is formed between the ground and the floor slab. The information about where the ground and floor slab begin is already intrinsically contained in the measurement data and is usually also (or alternatively, in the case of, for example, a non-level floor slab) known from plans and / or separate measurements.

[0075] To completely rule out the possibility that an installation element which is laid alone on the respective floor, i.e., does not penetrate the floor and the floor slab, does not form a point-free area and could therefore be incorrectly identified as a building structure, a simple procedure is to lay the horizontal section (at least in the first approach) over the entire height of the point cloud.

[0076] It should be noted that building structures may also exist that do not extend over the entire height of the floor. However, these are rarely, or usually, constructed using drywall, and can therefore be removed at manageable costs. Alternatively or additionally, these fixtures are known and can be identified using object recognition software (202413711).

[0077] 12

[0078] Determinable and / or reliably detected when determining point-free areas in (at least) one intermediate section.

[0079] In a further advantageous embodiment of the method, it is proposed that a boundary to a floor or a floor slab is automatically detected by means of a lack of measurement data and / or a lack of areas with point clouds.

[0080] In one embodiment, a floor or floor slab is detected by the formation of an (approximately) exact plane to which, in most cases, the vertical axis (and usually the orientation of the Earth's gravitational field) forms a normal. It should be noted, however, that inclined and curved floor slabs (and sometimes also floors) occur and can therefore be reliably detected with the embodiment of the method proposed here. These limitations, in contrast to a lack of measurement data and / or a region of the point cloud, share the characteristic that a (mathematical or freeform) plane can be formed which exhibits only minimal deviations. Such deviations are measurement errors or measurement data processing errors and / or result from a (too) high measurement data resolution (for example, detecting hooks for suspended objects such as lamps or smoke detectors).For example, to avoid such interfering factors, the measurement resolution is reduced, a smoothing function is applied in advance, and / or objects are detected and removed (at least for determining the course of the floor or the floor slab) and replaced by a probable course (of the floor or the floor slab).

[0081] If it is known (for example from old plans) that a simple geometric shape must be present, separate measurements to determine the height are carried out at one or more suitable (individual) locations, or the height is determined from the point cloud at one or more (individual) locations, from which the (for example constant) height of the floor section is then determined.

[0082] In a further advantageous embodiment of the method, it is proposed that several projectile sections be aligned with each other by means of:

[0083] the border created in step c.; and / or

[0084] the outer walls of the floors. 202413711

[0085] 13

[0086] Many buildings, including production halls and warehouses, comprise multiple floors. For installations spanning multiple floors (such as downpipes), the spatial relationship between the individual floors is crucial. While a technically and economically used building is often constructed from a basic framework of continuous columns and crossbeams (i.e., a building skeleton that defines the building structure), there are buildings where the structure is not precisely aligned. In one design, columns (for example, blocks of columns) are only constructed after the floor slabs of the floor below have been installed. These columns may then exhibit a significant offset from one another.

[0087] For purely technical reasons (e.g., the fixed depth of a floor's exterior wall), combined with the aesthetic requirement that, viewed from the outside, the floor's exterior walls appear flush and aligned across all floors, and / or for ease of implementation (e.g., continuous sheathing), the floor's exterior walls are usually a reliable reference point for the relative orientation of the floors or the point clouds (or the identified point-free areas) to each other. The floor's exterior walls, which, as previously explained, can be determined from inside the building using optical surveying with reference to the floor and / or the floor slab, are usually intrinsically known and / or easily determinable from existing plans and / or other separate measurement methods (e.g., object recognition, such as windows).

[0088] If, for example, it is known from old plans that the supports of a building are part of a building skeleton and run (at least with sufficient accuracy) exactly one above the other, i.e., continuously, then the resulting (circumferential) outlines provide a reliable indication of how the individual floors are oriented relative to each other. In one embodiment, several such criteria are applied in order to determine the most probable (and preferably sufficiently accurate) relative orientation of the floors to each other.

[0089] In a further advantageous embodiment of the method, it is proposed that point-free areas made visible by step c. are referenced to an absolute measuring point.

[0090] An absolute measurement point is a measurement that is integrated into a universally valid coordinate system. One such universally valid coordinate system is 202413711.

[0091] 14

[0092] for example, the geographical location in relation to the prime meridian (e.g., GPS coordinates) or a local coordinate system (e.g., for a plot of land or a building complex).

[0093] For indoor measurements, it is often difficult to reference them to an absolute measuring point using simple means, because usually an open sky, an undisturbed magnetic field (e.g., due to steel reinforcement), and an unobstructed radio signal are not available. Therefore, it is advantageous to rely on the correct geometric referencing of the measuring points to each other and then align these measuring points to a reliably referenced (absolute) measuring point within the universally accepted coordinate system. This is useful, for example, for accurately determining the position of the sun and / or the orientation of several buildings relative to each other. In one embodiment, the floors of a building are referenced independently to an absolute measuring point (either exclusively or additionally to the methods mentioned above), and their relative orientation to each other is then determined.

[0094] The invention described above is explained in detail below against the relevant technical background with reference to the accompanying drawings, which show preferred embodiments. The invention is in no way limited by the purely schematic drawings, although it should be noted that the drawings are not dimensionally accurate and are not suitable for defining size relationships. It is illustrated in

[0095] Fig. 1: a section of a multi-story building in a spatial representation; Fig. 2: a section of a point cloud of one story of a building;

[0096] Fig. 3: a detail section from a point cloud of Fig. 2; and

[0097] Fig. 4: a detailed view of a support column of a floor of a building.

[0098] Figure 1 shows a section of a multi-story (e.g., three-story) building 1 in a spatial representation. Firstly, it is easy to understand that it is a virtually impossible task for a computer to extract the building structure from this point cloud 2 alone. Ultimately, however, even a human estimation is merely a rough approximation based on experience. A solution using object recognition, especially with the help of artificial intelligence (specifically: neural networks), is therefore not better or not reliably better. However, it is 202413711

[0099] 15

[0100] It is not excluded to use such instruments additionally for other aspects or for improvement.

[0101] Furthermore, it is shown here that a horizontal section 5 is formed as a floor section (long dimension line on the middle floor) immediately adjacent to the floor 12 and the floor slab 13, whereby the floor 12 and the floor slab 13, or rather the Earth's gravitational field (along the height axis 4 of the drawn coordinate system 16), define the orientation of the horizontal plane 3. For a potentially more detailed representation of the course of a building structure (i.e., here the column), the horizontal section 5 is to be divided into several (here, for the sake of simplicity, two) intermediate sections 9 and these are to be considered individually. The result of the procedure is that which is shown in simplified form in Fig. 4.

[0102] Figure 2 shows a section of a point cloud 2 from one floor of a building 1. This point cloud 2 is viewed from the direction of Earth's gravitational field, i.e., looking along the vertical axis 4 towards the horizontal plane 3 (see the inscribed coordinate system 16). Here, the area 8 of the point cloud 2 and several point-free areas 6 are already easily recognizable, making the building structure (at least roughly) discernible to the human eye. It is interesting to note that a point cloud 2 is also generated outside the building 1 at openings in the walls (e.g., windows). It should be noted that the point cloud 2 is derived using optically generated measurement data from inside the building. It is possible that further information (such as the absolute position) was generated, possibly using other measurement methods or (if sufficient for the respective application) using old plans.

[0103] Figure 3 shows a detailed section of a point cloud 2 from Figure 2, in which objects 10 (represented by two, namely a motor vehicle and a ventilation duct) are already recognizable, and, more importantly, point-free areas 6 are clearly identifiable as the building structure. This is achieved by creating a horizontal section 5 over a height 11 (compare Figure 1) that is greater (i.e., higher) than the maximum height of the objects 10 in the measured floor. For example, the horizontal section 5 shown is a floor section between the respective floor 12 and floor slab 13 (compare Figure 1). Furthermore, it can be seen that (at least in this detailed section) the outer wall 14 of the floor forms a clear edge to an area 6 without measurement points. An absolute measurement point 15 outside the area is shown here as a purely exemplary example.

[0104] 16

[0105] Building 1 is shown, to which point cloud 2 can easily be referenced (i.e., aligned), for example, via the course of the outer wall of floor 14. It should be noted that the course of the outer wall of floor 14 shown here only depicts its inner side. However, due to the known thickness of the outer wall of floor 14, such a spatial reference can be defined with sufficient certainty for most applications.

[0106] Figure 4 shows a detailed view from the point cloud 2 of Figure 3 of a column of a story of a building 1, where a (optionally simplified) circumferential border 7 is already placed in the column (dashed line, filled with hatching for clarity). Here, objects 10 (represented pars pro toto as three, namely a shaft, a pipe, and a formwork) are shown with dashed lines (without filled areas) (for example, purely for understanding purposes and not, or only optionally, implemented in this computer-aided procedure). These objects can be classified, for example, using another method (e.g., object recognition). It is clearly visible that these objects 10 do not form any point-free areas 6 and / or do not have any (expected or predetermined) geometric shape or size. Therefore, they can be easily distinguished from a building structure using the horizontal section 5 in a computer-aided manner.It is therefore possible to determine a building structure very precisely in terms of location and shape, even if there are many (optical) obstructions present.

[0107] The method proposed here allows for the precise, computer-aided determination of the shape and position of building structures based on an optically generated point cloud, using simple means. 202413711

[0108] 17

[0109] Reference symbol list

[0110] 1 building

[0111] 2 point cloud

[0112] 3 Horizontal plane

[0113] 4 Height axis

[0114] 5 horizontal section 6 point-free area

[0115] 7 Border

[0116] 8 Area of ​​the point cloud 9 Intersection

[0117] 10 classified object 11 height

[0118] 12 Floor

[0119] 13th floor slab

[0120] 14 Floor exterior wall 15 Absolute measuring point 16 Coordinate system

Claims

202413711 18 Patent claims 1. Procedure for determining the construction status of a building (1) wherein the procedure is carried out by a computer on the basis of optically recorded measurement data of the building (1), wherein the measurement data are processed as a point cloud (2) with a horizontal plane (3) as the sheet plane and a height axis (4) perpendicular to the sheet plane, the procedure includes at least the following steps in the order mentioned: a. Creating at least one section (5) in a horizontal or vertical plane from the point cloud (2); b. Capturing point-free areas (6) in each of the sections (5); and c. Creating a border (7) around each captured point-free area (6).

2. The method according to claim 1, wherein after step b. and before or in addition to step c. in an additional step d., a transition sharpness between a point-free area (6) and an area (8) with the point cloud (2) is recorded and compared with a predetermined limit value, and if the predetermined limit value is exceeded: a.' Creating multiple horizontal intermediate sections (9) from the horizontal section (5), wherein in step c. for each intermediate section (9) an analogous additional step c.' is performed to create a border (7) around a respective captured point-free area (6) in the respective intermediate section (9), wherein preferably after step b. and before or in addition to step c.' in an additional step d.', a transition sharpness between a point-free area (6) and an area (8) with the point cloud (2) is detected and compared with the predetermined limit value, and where preferably if the predetermined limit is exceeded in one of the intermediate cuts (9) before, during or after additional step c.': a." Creating multiple horizontal intermediate cuts (9) from the intermediate cut (9) in question, and Repeat the process until the predetermined limit for transition sharpness in the subsequently generated intermediate cuts (9) is undershot and / or a predetermined cutting height limit is reached. 202413711 19 3. Method according to claim 1 or claim 2, wherein Furthermore, object recognition is performed to automatically classify objects (10) in the building (1) in question, where the detected and classified objects (10) are preferably removed.

4. Method according to any one of the preceding claims, wherein in step e. the border (7) created in step c. is extruded over the entire height (11) of the relevant section (5) or intermediate section (9).

5. Method according to any one of the preceding claims, wherein the at least one horizontal section (5) as a floor section, preferably immediately adjacent, lies between floor (12) and floor slab (13).

6. Method according to claim 5, wherein by means of a lack of measurement data and / or a lack of areas (8) with point cloud (2) a boundary to a floor (12) or to a floor slab (13) is automatically detected.

7. Method according to claim 5 or claim 6, wherein Several floor sections are aligned with each other using: the border created in step c. (7); and / or the outer walls of the floors (14).

8. Method according to any one of the preceding claims, wherein The point-free areas (6) made visible by step c. are referenced to an absolute measuring point (15).

Images