Rapid construction method and system of infrared three-dimensional model
By dividing the perspective of point cloud data of infrared three-dimensional models and analyzing the degree of noise impact, the problem of difficult region filtering in the existing technology is solved, and more accurate noise reduction and three-dimensional model construction is achieved.
Patent Information
- Application Number
- CN202411569885.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-06
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-11-06
AI Technical Summary
The existing filtering algorithms are difficult to filter the point cloud data of infrared three-dimensional models in different regions, resulting in difficulty in reducing noise interference and affecting the accuracy of the three-dimensional model.
By dividing, clustering and segmenting point cloud data from different perspectives, the noise impact degree of each block is determined, and noise reduction is performed based on the noise impact degree, and infrared three-dimensional model is constructed.
The segment filtering of point cloud data is realized, noise interference is reduced, and the accuracy and construction quality of the three-dimensional model are improved.
Smart Images

Figure CN119091054A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image data processing, and in particular to a method and system for quickly constructing an infrared three-dimensional model in the technical field of image data processing. Background Art
[0002] Infrared 3D model usually refers to the 3D model generated by computer processing of data obtained by infrared technology. This technology is mainly used in thermal imaging and infrared photography, and can be used to analyze the thermal distribution, temperature change and other information of objects. By collecting infrared image data of the target or scene and preprocessing it, computer vision and image processing technology are used to infer the 3D structure and morphology of the object from the preprocessed infrared image.
[0003] In the related art, in order to construct an infrared three-dimensional model of the target ship, the target object is scanned by infrared light, during which the transmission and environmental influences will interfere with the collected point cloud data and generate noise. The current common method for noise filtering of point cloud data is to directly use the existing filtering algorithm to perform noise filtering on the entire point cloud data. However, due to the different reflection capabilities of objects to infrared light, the noise sensitivity of different parts is different. Therefore, when performing filtering operations, it is necessary to analyze different parts for regional filtering. When the existing filtering algorithm is directly used to perform noise filtering on point cloud data, it is difficult to achieve the purpose of regional filtering of different parts of the object. Summary of the invention
[0004] In order to solve the technical problem that it is difficult to implement regional filtering of different parts of an object when using existing filtering algorithms to perform noise filtering on point cloud data, the purpose of the present invention is to provide a method and system for quickly building an infrared three-dimensional model. The technical solutions adopted are as follows: In a first aspect, an embodiment of the present invention provides a method for quickly constructing an infrared three-dimensional model, the method comprising: Step 101, obtaining point cloud data of the target scene; Step 102, dividing the point cloud data into different viewing angles to obtain a plurality of blocks; Step 103, determining the noise impact degree of each block in the plurality of blocks; Step 104, based on the noise impact degree, performing noise reduction processing on the point cloud data to obtain noise-reduced data; Step 105: construct an infrared three-dimensional model of the target scene based on the noise-reduced data.
[0005] Preferably, the specific steps of step 101 include: Using an infrared distance sensor to collect data on the target scene to obtain multiple distance data; The plurality of distance data are preprocessed to obtain the point cloud data.
[0006] Preferably, the specific steps of step 102 include: Clustering the point cloud data from different perspectives to obtain multiple clusters; Different regions corresponding to the multiple clusters are segmented to obtain the multiple blocks.
[0007] Preferably, segmenting different regions corresponding to the multiple clusters to obtain the multiple blocks includes: For any cluster among the multiple clusters, determine a projection plane of the any cluster in a cluster space coordinate system; Determine a neighborhood of each data point among a plurality of data points in the projection plane; Based on the neighborhood of each data point, different areas corresponding to the multiple clusters are divided into the multiple blocks.
[0008] Preferably, based on the neighborhood of each data point, different areas corresponding to the multiple clusters are divided into the multiple blocks, including: Determining the smoothness of any data point based on the maximum point in the neighborhood; Determining the data density in the neighborhood of any one of the data points; Based on the smoothness and the data density, determining a loss value for performing region segmentation within any cluster; Based on the loss value, different areas corresponding to the multiple clusters are divided into the multiple blocks.
[0009] Preferably, based on the loss value, different regions corresponding to the multiple clusters are divided into the multiple blocks, including: Determine the minimum loss value among the corresponding loss values in each cluster; Different regions in the cluster corresponding to the minimum loss value are divided into the multiple blocks.
[0010] Preferably, the specific steps of step 103 include: In any one of the plurality of blocks, determining a minimum point; In any of the blocks, determine candidate minimum points adjacent to the minimum point, and maximum points outside a preset radius of the minimum point; Determine the gully coefficient of any of the blocks based on the minimum point, the candidate minimum point and the maximum point; Based on the gully coefficient of any of the sub-blocks, the noise influence degree of any of the sub-blocks is determined.
[0011] Preferably, determining the gully coefficient of any block based on the minimum point, the candidate minimum point and the maximum point includes: In any of the blocks, determining a first distance between a minimum point and a candidate minimum point, and a second distance between the minimum point and the maximum point; Determine the difference between the minimum point and the maximum point; Based on the difference, the first distance and the second distance, a gully coefficient of any one of the blocks is determined.
[0012] Preferably, determining the noise influence degree of any block based on the gully coefficient of any block includes: In any of the blocks, determining abnormal data points; The noise influence degree of any one of the blocks is determined based on the mean value of the abnormal data points and the gully coefficient.
[0013] Preferably, the specific steps of step 104 include: The noise influence degree is used as the scale factor of the Wiener filter to filter the point cloud data to obtain the noise-reduced data.
[0014] In a second aspect, a system for rapidly constructing an infrared three-dimensional model is provided, the system comprising: An acquisition module is used to acquire point cloud data of the target scene; A division module, used for dividing the point cloud data from different perspectives to obtain multiple blocks; A determination module, used to determine the noise impact degree of each block in the multiple blocks; A noise reduction module, used for performing noise reduction processing on the point cloud data based on the noise influence degree to obtain noise-reduced data; A construction module is used to construct an infrared three-dimensional model of the target scene based on the noise-reduced data.
[0015] According to a third aspect, a computer program product is provided, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the first aspect or any possible implementation of the first aspect.
[0016] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method in the first aspect or any possible implementation manner described in the first aspect.
[0017] Compared with the prior art, the rapid construction method of an infrared three-dimensional model of the present invention has the beneficial effect that: after obtaining the point cloud data of the infrared image of the target scene, the infrared image is divided into different perspectives through the point cloud data to obtain multiple blocks, so that the image data of different perspectives in the infrared image can be reflected through the multiple blocks. Afterwards, the noise influence degree of each block in the multiple blocks is determined, so that the noise situation in each block can be accurately characterized by the noise influence degree, which is convenient for the subsequent noise reduction processing of the point cloud data based on the noise influence degree to obtain the noise-reduced data; finally, the three-dimensional model of the target scene is constructed according to the noise-reduced data. In this way, when performing the filtering operation, by analyzing the noise influence degree of each block under different perspectives, it is convenient to perform point filtering, so as to reduce the interference and noise caused to the point cloud data; and by using the noise-reduced data, the three-dimensional model of the target scene can be constructed, which can make the constructed three-dimensional model more accurate. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0019] Figure 1 It is a schematic diagram of the implementation process of a method for quickly constructing an infrared three-dimensional model provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of an application scenario of a method for quickly constructing an infrared three-dimensional model provided by an embodiment of the present invention; Figure 3 This is another implementation flow diagram of a method for quickly constructing an infrared three-dimensional model provided by an embodiment of the present invention; Figure 4 This is another implementation flow diagram of a method for quickly constructing an infrared three-dimensional model provided by an embodiment of the present invention; Figure 5 It is a schematic diagram of the composition structure of a rapid construction system of an infrared three-dimensional model provided by an embodiment of the present invention; Figure 6 It is a structural schematic diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0020] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the method for quickly constructing an infrared three-dimensional model proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0021] The specific scheme of the method for quickly constructing an infrared three-dimensional model provided by the present invention is described in detail below with reference to the accompanying drawings.
[0022] See also Figure 1 , which shows a flowchart of a method for quickly building an infrared three-dimensional model provided by an embodiment of the present invention. The method can be implemented by the following steps: Step 101, obtaining point cloud data of the target scene.
[0023] Here, the infrared distance sensor is used to collect data of the target scene to obtain multiple distances, and then the multiple distances are pre-processed to obtain the point cloud data. For example, the infrared distance sensor is used to collect data of the target scene in multiple directions, and the multiple distances are normalized to obtain the point cloud data. In this way, by collecting infrared data of the target scene in multiple directions, the obtained point cloud data is richer.
[0024] In some possible implementations, the target scene is Figure 2 Taking the ship shown as an example, for a ship that needs to build a three-dimensional model, since the construction of the three-dimensional model involves the spatial position of various parts of the ship, it is necessary to collect data from all directions of the ship.
[0025] For the ship, select a suitable drone and install an infrared distance sensor suitable for the ship above the drone, start the drone and turn on the infrared distance sensor, scan from above the ship to obtain the distance data of the scene. Among them, the distance data includes: the distance information of multiple points in the ship. In this way, the distance data of the ship is obtained, and the distance data is normalized to obtain point cloud data. The point cloud is a data set composed of a large number of discrete points. The position of each point is determined by the measurement value of the ranging sensor, and the data of each point is the three-dimensional point cloud data.
[0026] Step 102, dividing the point cloud data into different viewing angles to obtain a plurality of blocks.
[0027] Here, the point cloud data is divided according to the multiple perspectives included in the target scene, thereby obtaining multiple blocks, wherein different blocks are used to represent different areas in the target scene.
[0028] In some possible implementations, for the point cloud data of the target scene obtained by the above operation, since the subsequent construction of the three-dimensional model is based on the point cloud data, but due to environmental factors and other reasons during infrared ranging, noise will appear in the point cloud data obtained, and the point cloud data obtained needs to be denoised. Therefore, it is necessary to project the discrete data points of the obtained point cloud data into different planes. The data points projected into different planes are analyzed to obtain different blocks therein; according to the noise sensitivity of the blocks, the degree of noise influence of each block is obtained, and noise reduction processing is performed on different blocks according to the degree of noise influence. The above step 102 can be performed by Figure 3 The steps shown achieve: Step 301 : clustering the point cloud data from different perspectives to obtain a plurality of clusters.
[0029] Here, taking the target scene as a ship as an example, for the point cloud data obtained by the above operation, since the noise analysis of the collected ship point cloud data needs to be performed, it is necessary to analyze the ship from different perspectives, so it is necessary to cluster the discrete three-dimensional data points in the point cloud data and each cluster can better reflect the situation of the scattered points under each perspective. Therefore, the data points in the cluster should be able to reflect the characteristics of each face of the ship, and the difference between the data after cluster projection should be small, otherwise it is a point cluster that does not belong to the face; therefore, the data point clusters of several faces of the ship are obtained through K-means clustering, that is, multiple clusters are obtained.
[0030] Step 302: segment different regions corresponding to the multiple clusters to obtain the multiple blocks.
[0031] Here, taking the target scene as a ship as an example, after obtaining the data point clusters of different faces of the ship, there are many different partitions in one face of the ship, such as the deck, cockpit, etc., so there are different partitions on different faces, and due to the different surface materials of the partitions and the different surface flatness and complexity under the perspective of the cluster. When performing infrared ranging, due to the difference between materials and flatness, there are differences in density and flatness between the acquired point cluster data, so that the division of different areas can be obtained, thereby obtaining multiple blocks.
[0032] In some possible implementations, the above step 302 may be implemented by the following steps 321 to 323: Step 321: for any cluster among the multiple clusters, determine a projection plane of the any cluster in the cluster space coordinate system.
[0033] For example, for any cluster, the current observation position is projected onto a projection plane perpendicular to the observation position, and the projection plane is the x and y axes in the cluster space coordinate system, thereby obtaining several new clusters of cluster points in different viewing angle reference systems.
[0034] Step 322, determining a neighborhood of each data point among the plurality of data points in the projection plane.
[0035] For example, for each data point in the projection plane, a surrounding area with a radius of 3 and the data point as the center is determined, thereby obtaining the neighborhood of each data point.
[0036] Step 323: based on the neighborhood of each data point, divide different areas corresponding to the multiple clusters into the multiple blocks.
[0037] Here, in the neighborhood of each data point, the data density of the neighborhood and the smoothness of each data point are calculated, so that the division of different areas in the clustering cluster is achieved by combining the smoothness and the data density.
[0038] In some possible implementations, the above step 323 may be implemented by the following process: First, based on the maximum point in the neighborhood, the smoothness of any data point is determined.
[0039] Here, for each cluster, the scattered points in it are projected onto the xy plane in its new coordinate system, and its z-axis data is recorded as the value of the scattered point. The density is reflected by the discreteness of the scattered points on the projection plane, and the scattered point value reflects its flatness. For each data point on the projection plane, the surrounding area with a radius of 3 is determined with it as the center, and the maximum value point (if the data point has a larger value in the circle with a radius of 1 as the center, it is recorded as a maximum value point) and the minimum value point are obtained, and the data points between the maximum values are obtained, so that the smoothness of each data point in each cluster can be obtained. , as shown in formula (1): (1); in, is any data point in the cluster. is any maximum point in the area around the data point in the cluster (i.e., the neighborhood), and J represents the number of maximum points. It is the nearest maximum point to any maximum point in the area around the data point in the cluster. It is the first The area around the data point The value of the maximum point, It is the first The area around the data point The value of the maximum point closest to the maximum point, It is the first The area around the data point The numerical mean of the numerical points between a maximum point and its nearest maximum point, and the numerical mean is not 0; It is the first The area around the data point The distance between a maximum point and its nearest maximum point, and the distance value is not 0; The natural constant An exponential function with base .
[0040] Next, the data density in the neighborhood of any data point is determined.
[0041] Here, the data density in the neighborhood of the data point is obtained by the number of data points in the area around the data point.
[0042] Thirdly, based on the smoothness and the data density, a loss value for performing region segmentation within any cluster is determined.
[0043] Here, different neighborhoods of different face clusters are segmented according to the smoothness and density of the data points, and the segmentation loss value within the cluster after segmentation needs to be minimized, where the segmentation loss value is E, as shown in formula (2): (2); in, is any neighborhood in the cluster, N represents the number of neighbors, is the loss value of intra-cluster segmentation, It is the first The smoothness and density variance of the data points in the neighborhood, It is the information entropy of the mean product of the smoothness and density of the internal data points in each neighborhood of the cluster, and the information entropy is not 0; It is an exponential function with the natural constant e as its base.
[0044] Finally, based on the loss value, the different regions corresponding to the multiple clusters are divided into the multiple blocks. In this way, by combining the smoothness and data density, the loss value of the region segmentation in each cluster can be accurately calculated, so that the different regions corresponding to the clusters can be accurately divided into multiple blocks.
[0045] In some possible implementations, the minimum loss value is determined from the corresponding loss values in each cluster; for example, after obtaining several new cluster point clusters with different perspective reference systems, the loss values are calculated, where the loss value is the average of the maximum difference in the z-axis values of each cluster cluster in the new coordinate system. According to the obtained loss value, the cluster with the smallest loss value is obtained, because the smaller the loss value, the smaller the difference in the z-axis values of the data in the cluster point cluster, which indicates that they are on the same surface. And the different areas in the cluster with the smallest loss value are divided into the multiple blocks. In this way, by calculating the loss corresponding to each cluster cluster, the different areas in the cluster with the smallest loss value are segmented, so that the point cloud data can be accurately divided into multiple blocks, so as to facilitate the subsequent filtering operation according to each block.
[0046] Step 103: Determine the noise impact degree of each block in the multiple blocks.
[0047] In some possible implementations, taking the target scene as a ship as an example, for the obtained several clusters and their new coordinate systems, since a cluster reflects the scattered points distributed on a surface of the ship, and since there are different blocks on different surfaces of the ship, such as the deck, hull, cockpit, cargo, etc.; therefore, for the clusters of each surface obtained, it is necessary to divide each block according to the differences in structural features within different blocks, and obtain the noise impact degree of each block according to the internal abnormal data points of each block and the sensitivity of the noise within each block. The above step 103 can be performed by Figure 4 The steps shown achieve: Step 401: determine a minimum point in any of the multiple blocks.
[0048] Here, for different blocks in the clusters obtained by the above operation, since different blocks belong to the same part of the ship; and since different parts of the ship surface have different measurement accuracies due to different absorption of infrared light reflection when performing infrared ranging, different parts have different sensitivity to noise. If abnormal data points appear in parts with high noise sensitivity, then the possibility that they belong to noise is high, and vice versa; therefore, it is necessary to quantify the noise sensitivity according to the characteristics of the internal data points of different parts; the gully part will refract infrared light, resulting in some deviations in this part, so its noise sensitivity is relatively low. For any block, obtain the values of each data point in the block, and take the value of the smallest data point as the minimum point in the block.
[0049] Step 402: In any of the blocks, determine candidate minimum points adjacent to the minimum point, and maximum points outside a preset radius of the minimum point.
[0050] Step 403: Determine the gully coefficient of any of the blocks based on the minimum point, the candidate minimum points and the maximum point.
[0051] In some possible implementations, in any of the blocks, first, a first distance between a minimum point and a candidate minimum point, and a second distance between the minimum point and the maximum point are determined; second, a difference between the minimum point and the maximum point is determined; and finally, a gully coefficient of any of the blocks is determined based on the difference, the first distance, and the second distance. In this way, by analyzing the difference between the maximum point and the minimum point, and the difference between the minimum point and the surrounding candidate minimum points, the gully coefficient of each block can be accurately analyzed.
[0052] Here, the gully coefficient is obtained according to the distribution of data points inside each block. Since the gully-shaped part will produce partial refraction during infrared light ranging, it is easier to have abnormal points; the data values on both sides of the gully part are higher and the values inside are lower, so for the data points inside the block, the minimum and maximum points are determined (the data point has the largest or smallest value in the circle with itself as the center and a radius of 1, which is recorded as the maximum point or the minimum point); the gully coefficient of the block is obtained, as shown in formula (3): (3); in, is any minimum point in the cluster partition, is the number of minimum points in the cluster partition, is the nearest minimum point around any minimum point in the cluster partition (i.e., candidate minimum point), It is the nearest maximum point outside the radius 1 of any minimum point in the cluster block. It is the first The gully coefficient of each block, It is the first In the block The distance between a minimum point and its nearest minimum point, It is the first In the block The mean of the data points between a minimum point and its nearest minimum point, It is the first In the block The value of the minimum point, It is the first In the block The value of the nearest minimum point is It is the first In the block The minimum point and the The distance between the nearest maximum points outside the minimum point radius 1, It is the first In the block The difference between a minimum point and the nearest maximum point outside the radius 1, and the difference is not 0; It is an exponential function with the natural constant e as its base.
[0053] The first distance between the minimum point in the classification block of the cluster and its nearest minimum point (i.e., the candidate minimum point) , the larger its value is, the smaller the first distance between the minimum points is, so it is more likely to appear gully-like; and the interpolation of the numerical mean of the minimum point and its nearest minimum point and the numerical mean of the data points between them is The smaller the value, the smaller the difference in the data points between the minimum points is, so it is easy to form a gully shape; and for the second distance between the extreme point in the cluster block and its nearest maximum point The smaller the value, the smaller the second distance between the minimum point and the maximum point in the image, so it is easier to have gullies; and for the numerical difference between the minimum point and the maximum point (i.e. the difference between the minimum point and the maximum point), the larger the value is, the deeper the gully is, and therefore the higher the gully coefficient is.
[0054] Step 404: Determine the noise influence degree of any block based on the gully coefficient of any block.
[0055] In the above steps 401 to 404, for any block, the maximum point and the minimum point in the block are combined to first calculate the gully coefficient of any block, and then the noise influence degree of the block is accurately calculated through the gully coefficient of the block.
[0056] In some possible implementations, abnormal data points are determined in any of the blocks; and the noise impact degree of any of the blocks is determined based on the mean of the abnormal data points and the gully coefficient. In this way, by analyzing the abnormal data points in each block and combining the mean of the abnormal data points with the gully coefficient of the block, the noise impact degree of the block can be quickly and accurately calculated.
[0057] Here, the noise influence degree of each block is obtained according to the obtained gully coefficient. First, the abnormal points of each block are obtained according to the LOF algorithm. The LOF mean value of the abnormal point is compared with the gully coefficient of the block to obtain the noise influence degree, as shown in formula (4): (4); in, It is the first The noise impact of each block, It is the first The LOF mean of the outliers in the blocks, is a hyperbolic function; express The absolute value of .
[0058] Step 104: Based on the noise impact degree, perform noise reduction processing on the point cloud data to obtain noise-reduced data.
[0059] Here, according to the noise influence degree of each block in the cluster obtained by the above operation, a Wiener filter with better ability to retain the original information is selected, and the noise influence degree is used as the scale factor of the Wiener filter; wherein the scale factor acts to adjust the filtering strength of the Wiener filter. Generally, the image signal-to-noise ratio is used to set the scale factor, but the image signal-to-noise ratio cannot be obtained without a clear reference image. Therefore, the information difference is used as the scale factor of the Wiener filter, so that the Wiener filter can filter out the noise in the blocks of each cluster with the highest efficiency and the best filtering effect. In this way, the noise influence degree is used as the scale factor of the Wiener filter, and the point cloud data is filtered to achieve noise reduction processing, thereby obtaining the filtered and noise-reduced data, that is, the filtered point cloud data.
[0060] Step 105: construct a three-dimensional model of the target scene based on the denoised data.
[0061] Here, the denoised data is converted into a voxel grid. A voxel is a cubic unit in a three-dimensional space. By filling the denoised data into the voxels, voxelized three-dimensional data can be created. The method based on the Poisson equation can reconstruct a smooth surface from disordered point cloud data; the texture information is mapped to the surface of the generated three-dimensional model to increase realism and details.
[0062] In an embodiment of the present invention, the infrared image is divided into different viewing angles through point cloud data to obtain multiple blocks, so that the image data of different viewing angles in the infrared image can be reflected through the multiple blocks. Afterwards, the noise influence degree of each block in the multiple blocks is determined, so that the noise situation in each block can be accurately characterized by the noise influence degree, which is convenient for the subsequent noise reduction processing of the point cloud data based on the noise influence degree to obtain the noise-reduced data; finally, the three-dimensional model of the target scene is constructed according to the noise-reduced data. In this way, when performing the filtering operation, by analyzing the noise influence degree of each block under different viewing angles, it is convenient to perform point filtering, thereby reducing the interference and noise caused to the point cloud data; and by using the noise-reduced data, the three-dimensional model of the target scene can be constructed, which can make the constructed three-dimensional model more accurate.
[0063] The embodiment of the present invention provides a system for quickly constructing an infrared three-dimensional model. Figure 5 As shown, the system 500 includes: An acquisition module 501 is used to acquire point cloud data of a target scene; A division module 502 is used to divide the point cloud data into different viewing angles to obtain a plurality of blocks; A determination module 503 is used to determine the noise impact degree of each block in the multiple blocks; A noise reduction module 504 is used to perform noise reduction processing on the point cloud data based on the noise impact degree to obtain noise-reduced data; The construction module 505 is used to construct an infrared three-dimensional model of the target scene based on the noise-reduced data.
[0064] In some possible implementations, the segmentation module 502 is further configured to cluster the point cloud data from different perspectives to obtain a plurality of clusters; and segment different regions corresponding to the plurality of clusters to obtain the plurality of blocks.
[0065] In some possible implementations, the partitioning module 502 is further used to determine, for any cluster among the multiple clusters, a projection plane of the cluster in the cluster space coordinate system; determine a neighborhood of each data point among multiple data points in the projection plane; and based on the neighborhood of each data point, divide the different areas corresponding to the multiple clusters into the multiple blocks.
[0066] In some possible implementations, the partitioning module 502 is also used to determine the degree of smoothness of any data point based on the maximum point in the neighborhood; determine the data density in the neighborhood of any data point; determine the loss value for performing region segmentation within any cluster based on the smoothness and the data density; and divide the different regions corresponding to the multiple clusters into the multiple blocks based on the loss value.
[0067] In some possible implementations, the division module 502 is further used to determine a minimum loss value among the corresponding loss values in each cluster; and divide different areas in the cluster corresponding to the minimum loss value into the multiple blocks.
[0068] In some possible implementations, the determination module 503 is further used to determine a minimum point in any of the multiple blocks; determine candidate minimum points adjacent to the minimum point, and a maximum point outside a preset radius of the minimum point in any of the blocks; determine a groove coefficient of any of the blocks based on the minimum point, the candidate minimum points, and the maximum point; and determine a degree of noise influence of any of the blocks based on the groove coefficient of any of the blocks.
[0069] In some possible implementations, the determination module 503 is also used to determine, in any of the blocks, a first distance between a minimum point and a candidate minimum point, and a second distance between the minimum point and the maximum point; determine a difference between the minimum point and the maximum point; and determine a gully coefficient of any of the blocks based on the difference, the first distance, and the second distance.
[0070] In some possible implementations, the determination module 503 is further configured to determine abnormal data points in any of the blocks; and determine the degree of noise influence of any of the blocks based on the mean value of the abnormal data points and the gully coefficient.
[0071] In some possible implementations, the acquisition module 501 is further configured to collect data on the target scene using an infrared distance sensor to obtain a plurality of distance data; and pre-process the plurality of distance data to obtain the point cloud data.
[0072] Optionally, the transmission medium can be a wired link (such as but not limited to coaxial cable, optical fiber and digital subscriber line (DSL)) or a wireless link (such as but not limited to wireless Fidelity (WIFI), Bluetooth and mobile device network).
[0073] It should be noted that: the device provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.
[0074] Figure 6 is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention. Figure 6As shown, the computer device 600 includes: a memory 601, a processor 602, and a computer program 603 stored in the memory 601 and running on the processor 602, wherein when the processor 602 executes the computer program 603, the computer device can execute any one of the infrared three-dimensional model rapid construction methods described above.
[0075] In addition, an embodiment of the present invention also protects a device, which may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to perform a method for quickly constructing an infrared three-dimensional model provided by an embodiment of the present invention.
[0076] This embodiment can divide the functional modules of the device according to the above method example. For example, it can correspond to each functional module, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic and is only a logical function division. There may be other division methods in actual implementation. It should be noted that all relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, which will not be repeated here.
[0077] It should be understood that the device provided in this embodiment is used to execute the above-mentioned method for quickly building an infrared three-dimensional model, so the same effect as the above-mentioned implementation method can be achieved. In the case of an integrated unit, the device may include a processing module and a storage module. Wherein, when the device is applied to a device, the processing module can be used to control and manage the actions of the device. The storage module can be used to support the device to execute mutual program codes, etc. Wherein, the processing module can be a processor or a controller, which can implement or execute various exemplary logical boxes, modules and circuits described in conjunction with the disclosure of the present invention. The processor can also be a combination that implements a computing function, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc., and the storage module can be a memory.
[0078] In addition, the device provided in the embodiment of the present invention may be a chip, a component or a module, and the chip may include a connected processor and a memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip can execute a method for quickly building an infrared three-dimensional model provided in the above embodiment. This embodiment also provides a computer-readable storage medium, in which a computer program code is stored, and when the computer program code is run on a computer, the computer executes the above-mentioned related method steps to implement a method for quickly building an infrared three-dimensional model provided in the above embodiment.
[0079] It should be noted that the sequence of the above-mentioned embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous. The various embodiments in this specification are described in a progressive manner, and the same and similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments. The above content is only a specific implementation method of the present invention, but the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered within the protection scope of the present invention.
Claims
1. A method for rapidly constructing an infrared three-dimensional model, characterized in that: The method comprises: Step 101, obtaining point cloud data of the target scene; Step 102, dividing the point cloud data into different viewing angles to obtain a plurality of blocks; Step 103, determining the noise impact degree of each block in the plurality of blocks; Step 104, based on the noise impact degree, performing noise reduction processing on the point cloud data to obtain noise-reduced data; Step 105: construct an infrared three-dimensional model of the target scene based on the noise-reduced data.
2. The method for rapidly constructing an infrared three-dimensional model according to claim 1, characterized in that: The step 102 includes: Clustering the point cloud data from different perspectives to obtain multiple clusters; Different regions corresponding to the multiple clusters are segmented to obtain the multiple blocks.
3. The method for rapidly constructing an infrared three-dimensional model according to claim 2, characterized in that: The segmenting of different regions corresponding to the multiple clusters to obtain the multiple blocks includes: For any cluster among the multiple clusters, determine a projection plane of any cluster in the cluster space coordinate system; Determine a neighborhood of each data point among a plurality of data points in the projection plane; Based on the neighborhood of each data point, different areas corresponding to the multiple clusters are divided into the multiple blocks.
4. The method for rapidly constructing an infrared three-dimensional model according to claim 3, characterized in that: The dividing different areas corresponding to the multiple clusters into the multiple blocks based on the neighborhood of each data point includes: Determine the degree of smoothness of any data point based on the maximum point in the neighborhood; Determine the data density in the neighborhood of any data point; Based on the smoothness and the data density, determining a loss value for performing region segmentation within any cluster; Based on the loss value, different areas corresponding to the multiple clusters are divided into multiple blocks.
5. The method for rapidly constructing an infrared three-dimensional model according to claim 4, characterized in that: The dividing the different regions corresponding to the multiple clusters into the multiple blocks based on the loss value includes: Determine the minimum loss value among the corresponding loss values in each cluster; Different regions in the cluster corresponding to the minimum loss value are divided into the multiple blocks.
6. The method for rapidly constructing an infrared three-dimensional model according to claim 1, characterized in that: The step 103 comprises: In any one of the plurality of blocks, determining a minimum point; In any block, determine candidate minimum points adjacent to the minimum point, and maximum points outside a preset radius of the minimum point; Determine the gully coefficient of any block based on the minimum point, the candidate minimum point and the maximum point; Based on the gully coefficient of any block, the noise influence degree of any block is determined.
7. The method for rapidly constructing an infrared three-dimensional model according to claim 6, characterized in that: The determining of the gully coefficient of any of the blocks based on the minimum point, the candidate minimum point and the maximum point comprises: In any of the blocks, determining a first distance between a minimum point and a candidate minimum point, and a second distance between the minimum point and the maximum point; Determine the difference between the minimum point and the maximum point; Based on the difference, the first distance and the second distance, a gully coefficient of any one of the blocks is determined.
8. The method for rapidly constructing an infrared three-dimensional model according to claim 6, characterized in that: The determining the noise influence degree of any block based on the gully coefficient of any block includes: In any of the blocks, determining abnormal data points; The noise influence degree of any one of the blocks is determined based on the mean value of the abnormal data points and the gully coefficient.
9. The method for rapidly constructing an infrared three-dimensional model according to claim 1, characterized in that: The step 101 comprises: Using an infrared distance sensor to collect data on the target scene to obtain a plurality of distance data; The plurality of distance data are preprocessed to obtain the point cloud data.
10. A rapid construction system for infrared three-dimensional models, characterized in that: The system comprises: An acquisition module is used to acquire point cloud data of the target scene; A division module, used for dividing the point cloud data from different perspectives to obtain multiple blocks; A determination module, used to determine the noise impact degree of each block in the multiple blocks; A noise reduction module, used for performing noise reduction processing on the point cloud data based on the noise influence degree to obtain noise-reduced data; A construction module is used to construct an infrared three-dimensional model of the target scene based on the noise-reduced data.
Citation Information
Patent Citations
Laser radar point cloud denoising method based on multi-scale noise
CN115222625A
Structural point cloud data multi-scale filtering method considering environmental dynamic influence
CN115601269A
Electric reactor surface defect detection method
CN116310360A
Denoising method and device for point cloud data, storage medium and electronic equipment
CN118674654A
Wireless monitoring method and device for intelligent building construction site
CN118865259A
Cited By
Building structure design three-dimensional model construction method based on spatial point cloud data
CN119903588A
Three-dimensional Model Construction Method for Building Structure Design Based on Spatial Point Cloud Data
CN119903588B
Component stability analysis method and system based on BIM (Building Information Modeling) technology
CN121072260A
