Target for three-dimensional laser scanning with night vision function and method of using the same

By integrating an infrared camera and an enhanced filter into a night vision module, and combining an unsupervised variational autoencoder and algorithm to plan the scanning path of the infrared camera, the accuracy and efficiency issues of 3D laser scanning in nighttime or low-light environments are solved, achieving high-quality 3D scanning data acquisition and target recognition.

CN118623768BActive Publication Date: 2024-11-08SHENZHEN AIHUA RECONNAISSANCE ENG CO LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411088308.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-11-08
Estimated Expiration
2044-08-09

AI Technical Summary

Technical Problem

Existing 3D laser scanning targets suffer from reduced scanning accuracy, slower speed, and difficulty in target object identification and positioning at night or in low-light environments.

Method used

A night vision module consisting of an infrared camera, an infrared enhancement filter, and an image processor is used. Combined with an unsupervised variational autoencoder and specific algorithms, the scanning path of the infrared camera is planned, and the visible light and infrared light coding sub-modules are used to identify and distinguish targets under different lighting conditions.

Benefits of technology

It improves the accuracy and efficiency of 3D laser scanning at night or in low-light environments, ensures stable scanning support for targets under various lighting conditions, and enhances the intelligence and automation level of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118623768B_ABST
    Figure CN118623768B_ABST
Patent Text Reader

Abstract

The invention discloses a target for three-dimensional laser scanning with night vision function and a method for using the target, comprising: a target, comprising a reflector, used for providing scanning points during three-dimensional laser scanning; a night vision module, comprising an infrared camera, an infrared enhancement filter and an image processor; a connecting mechanism, used for connecting the night vision module and the target, so that the infrared camera is aimed at the reflective surface of the target; a control module, connected to the three-dimensional laser scanner and the night vision module; a data processing module, used for obtaining a target feature representation and a nighttime environment feature representation from image data and scanning data; inputting the target feature representation into a fusion device composed of AISOA and SLSGAA algorithms to obtain a sensitivity value of the infrared camera; using the sensitivity value and the nighttime environment feature representation as input parameters of an ISPPA scanning path planning algorithm to obtain an optimal infrared camera scanning path, and generating a three-dimensional scanning result containing nighttime environment information.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional laser scanning, and in particular to a target for three-dimensional laser scanning with a night vision function and a method for using the target. Background Art

[0002] 3D laser scanning targets with night vision function can still assist 3D laser scanners in high-precision measurements at night or in low-light environments. By placing the target in the measurement area, the scanner can lock the target to determine its coordinate system and facing direction, thereby generating a high-precision 3D model. It is widely used in the fields of architecture, industry, medicine, etc. For example, in the field of architecture, it can be used for building measurement and modeling tasks to capture accurate information about buildings for better protection and restoration. In the industrial field, it can be used for high-precision measurement and modeling of objects of different shapes and sizes.

[0003] Existing targets with night vision functions include reflector targets and laser projection targets. Among them, reflector targets use high-precision reflectors to reflect lasers, and have the characteristics of high precision and being unaffected by light. They are widely used in the fields of cultural relics protection and construction, and can provide stable reflection signals under various lighting conditions. Laser projection targets achieve positioning and reference by projecting specified light spots, and are simple and convenient to use. They are suitable for scenes that require quick setup and movement, and can be quickly deployed and emit obvious light signals. However, although reflector targets have high precision, they are not suitable for dynamic scenes. Although laser projection targets are simple and convenient, they are easily affected by light interference. The above targets face many challenges at night or in low-light environments, such as reduced scanning accuracy, reduced scanning speed, and difficulty in identifying and locating target objects.

[0004] Therefore, how to enhance the working ability of 3D laser scanning at night or in low-light environments and improve scanning accuracy and efficiency has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] In view of the above problems, the present invention is proposed to provide a three-dimensional laser scanning target with night vision function and a method of using the target and the ... to enhance the working ability of the three-dimensional laser scanning at night or in low light environment.

[0006] According to one aspect of the present invention, there is provided a target for three-dimensional laser scanning with night vision function, comprising:

[0007] A target, including a reflector, is used to provide a scanning point during a three-dimensional laser scanning process;

[0008] A night vision module, comprising an infrared camera, an infrared enhancement filter and an image processor; wherein the infrared enhancement filter is used to enhance the sensitivity of the infrared camera to infrared light, and the image processor is used to convert the image captured by the camera into a processable electronic signal;

[0009] A connecting mechanism, used to connect the night vision module and the target so that the infrared camera is aimed at the reflective surface of the target;

[0010] A control module is connected to the 3D laser scanner and the night vision module, and is used to receive image data transmitted by the night vision module, and assist the 3D laser scanner in performing night scanning based on the image data, and control the opening and closing of the night vision module and the adjustment of the infrared light source;

[0011] A data processing module is connected to the control module and is used to receive and process image data and scan data obtained from the night vision module and the three-dimensional laser scanner; obtain target feature representation and night environment feature representation from the image data and scan data according to an unsupervised variational autoencoder; input the target feature representation into a fusion device composed of AISOA and SLSGAA algorithms to obtain a sensitivity value of the infrared camera; use the sensitivity value and the night environment feature representation as input parameters of the ISPPA scanning path planning algorithm to obtain an optimal infrared camera scanning path and generate a three-dimensional scanning result containing night environment information;

[0012] The coding module includes a visible light coding submodule and an infrared light coding submodule, wherein the visible light coding submodule is used to number the targets in a normal light environment, and the infrared light coding submodule is used to emit a recognizable coding signal under infrared light irradiation to identify and distinguish the targets at night or in a low light environment.

[0013] In an optional manner, the unsupervised variational autoencoder includes an input layer, a convolutional autoencoder layer, a sparse coding layer, a variational inference network layer, a decoding layer, and an output layer;

[0014] Among them, the sparse coding layer introduces KL divergence constraints; the variational inference network layer includes a latent variable mean layer, a latent variable logarithmic variance layer and a heavy parameter adoption layer; the decoding layer includes a deconvolution layer and a convolution layer.

[0015] In an optional manner, the adoption formula of the heavy parameter adoption layer is:

[0016]

[0017] in, The latent variable mean layer is mapped to the mean vector of the latent variable; The log-variance vector of the latent variable is mapped to the log-variance layer of the latent variable; is the noise sampled from a standard normal distribution.

[0018] In an optional manner, inputting the target feature representation into a fusion device composed of AISOA and SLSGAA algorithms to obtain the sensitivity value of the infrared camera further includes:

[0019] Input the target feature representation into the AISOA algorithm for feature transformation to obtain the first feature vector;

[0020] The first eigenvector is input into the SLSGAA algorithm for fusion to generate a second eigenvector;

[0021] The second eigenvector is mapped by a mapping function to obtain the sensitivity value of the infrared camera.

[0022] In an optional manner, the method of using the sensitivity value and the nighttime environment characteristic representation as input parameters of an ISPPA scanning path planning algorithm to obtain an optimal infrared camera scanning path and generate a three-dimensional scanning result containing nighttime environment information further includes:

[0023] The sensitivity value and the night environment characteristic representation are read, wherein the night environment characteristic representation includes temperature distribution, brightness, location and size information of the target object.

[0024] Divide the scanning area into multiple sub-areas and evaluate the brightness and target object density of each sub-area;

[0025] Adjust the focal length and exposure time of the camera according to the brightness and sensitivity values ​​of each sub-area to adapt to the environmental conditions of different sub-areas;

[0026] Identify the target objects in the scanning area and prioritize them according to their size;

[0027] The scanning path is planned according to the A× heuristic search algorithm, and the infrared camera scanning path that maximizes the target detection rate is obtained to guide the scanning operation of the infrared camera.

[0028] In an optional manner, the loss function of the unsupervised variational autoencoder is:

[0029]

[0030] Among them, N is the number of samples, M is the number of categories, is the true probability that the i-th sample belongs to category j, is the predicted probability that the i-th sample belongs to category j, J is the dimension of the latent space, is the mean of the latent variable, is the variance of the latent variable.

[0031] In an optional manner, the infrared light encoding submodule includes:

[0032] A coding signal generator, connected to the infrared light source, for generating a recognizable infrared coding signal;

[0033] The modulator is connected between the coding signal generator and the infrared light source, and is used to modulate the coding signal onto the infrared light source to control the infrared light source to flash according to the coding signal; wherein the infrared coding signal includes a plurality of pulse sequences, each pulse sequence having a preset pulse width, pulse interval and pulse frequency to distinguish different targets.

[0034] In an optional manner, the night vision module further includes an infrared light source regulator;

[0035] The infrared light source regulator is connected to the control module and is used to dynamically adjust the luminous intensity, luminous angle and flashing frequency of the infrared light source according to the real-time image data and environmental information of the infrared camera to optimize the imaging effect of the infrared camera at night or in low light environments.

[0036] In an optional manner, the control module further includes an automatic calibration module for automatically adjusting the relative position between the infrared camera and the target according to the image data of the night vision module to ensure that the infrared camera is always aligned with the reflective surface of the target.

[0037] According to another aspect of the present invention, there is provided a method for using a target for three-dimensional laser scanning with a night vision function, comprising:

[0038] The night vision module is activated through the control module to capture an infrared image including the reflective surface of the target;

[0039] The image data captured by the infrared camera is enhanced by an infrared enhancement filter and then transmitted to the image processor for processing, and the scanning data is obtained from the 3D laser scanner;

[0040] Use unsupervised variational autoencoders to process image data and scan data to extract target feature representation and night environment feature representation;

[0041] The target feature representation is converted into the sensitivity value of the infrared camera through the fusion of AISOA and SLSGAA algorithms, and the optimal infrared camera scanning path is planned in combination with the A× heuristic search algorithm;

[0042] According to the planned scanning path, the infrared camera is guided to scan; the infrared coding signal for numbering and identifying the target is used to identify and distinguish different targets; and the three-dimensional scanning result containing night environment information is output.

[0043] The solution provided by the present invention includes a target, including a reflector, which is used to provide a scanning point during the three-dimensional laser scanning process; a night vision module, including an infrared camera, an infrared enhancement filter and an image processor; wherein the infrared enhancement filter is used to enhance the sensitivity of the infrared camera to infrared light, and the image processor is used to convert the image captured by the camera into a processable electronic signal; a connecting mechanism is used to connect the night vision module and the target so that the infrared camera is aimed at the reflective surface of the target; a control module is connected to the three-dimensional laser scanner and the night vision module, and is used to receive the image data transmitted by the night vision module, and assist the three-dimensional laser scanner to perform night scanning based on the image data, and at the same time control the opening and closing of the night vision module and the adjustment of the infrared light source; a data processing module is connected to the control module, and is used to receive and process the data from the night vision module and the three-dimensional laser scanner The obtained image data and scan data; the target feature representation and the night environment feature representation are obtained from the image data and the scan data according to the unsupervised variational autoencoder; the target feature representation is input into the fusion device composed of the AISOA and SLSGAA algorithms to obtain the sensitivity value of the infrared camera; the sensitivity value and the night environment feature representation are used as input parameters of the ISPPA scanning path planning algorithm to obtain the best infrared camera scanning path, and generate a three-dimensional scanning result containing night environment information; the encoding module includes a visible light encoding submodule and an infrared light encoding submodule, wherein the visible light encoding submodule is used to number the target under a conventional light environment, and the infrared light encoding submodule is used to emit a recognizable encoding signal under infrared light irradiation to identify and distinguish the target at night or in a low light environment. The present invention integrates an infrared camera and an infrared enhancement filter, so that the target can capture a clear infrared image at night or in a low light environment, thereby ensuring that the three-dimensional laser scanner can accurately obtain the position information of the scanning point, improving the accuracy of night scanning, and obtaining high-quality three-dimensional scanning data even in insufficient light. The control module automatically adjusts the brightness and angle of the infrared light source, so that the target can provide stable scanning support under various lighting conditions, enhancing its environmental adaptability. The data processing module uses an unsupervised variational autoencoder and algorithm combination (AISOA and SLSGAA), which can quickly and accurately extract target feature representation and night environment feature representation from image data and scanning data. The ISPPA scanning path planning algorithm plans the best infrared camera scanning path, which improves scanning efficiency and reduces unnecessary scanning time and resource consumption. The encoding module combines visible light encoding submodules and infrared light encoding submodules to achieve dual encoding recognition in conventional light environments and infrared light environments, which not only improves the recognition accuracy and reliability of the target, but also enables the target to be accurately identified and located at night or in low light environments. In addition, the intelligence and automation level of the entire scanning system are improved.

[0044] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0046] Figure 1 A schematic diagram of a frame of a target for three-dimensional laser scanning with night vision function according to an embodiment of the present invention is shown;

[0047] Figure 2 A schematic diagram of the use process of a target for three-dimensional laser scanning with night vision function according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0048] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0049] Figure 1 FIG. 1 is a schematic diagram showing a frame of a target for three-dimensional laser scanning with night vision function according to an embodiment of the present invention. Specifically, Figure 1 As shown, including:

[0050] The target, including the reflector, is used to provide scanning points during the 3D laser scanning process. The target is placed in the area to be scanned, and the reflector of the target can accurately reflect the laser beam, providing accurate scanning points for the 3D laser scanner.

[0051] The night vision module includes an infrared camera, an infrared enhancement filter and an image processor; the infrared enhancement filter is used to enhance the infrared camera's sensitivity to infrared light, and the image processor is used to convert the image captured by the camera into a processable electronic signal. The infrared camera uses infrared light or other spectral bands that are invisible to the human eye for night vision, and achieves visible light shooting in the dark. Combined with the infrared enhancement filter, the infrared camera's sensitivity to infrared light is significantly enhanced, and it can capture clear images under extremely low light conditions. The image processor converts the image captured by the camera into a processable electronic signal in real time, and combines with the control system to realize intelligent recognition and analysis functions, thereby improving the automation and intelligence level of the system.

[0052] The connecting mechanism is used to connect the night vision module and the target so that the infrared camera is aimed at the reflective surface of the target.

[0053] The control module is connected to the 3D laser scanner and the night vision module, and is used to receive the image data transmitted by the night vision module, and assist the 3D laser scanner in performing night scanning based on the image data, while controlling the opening and closing of the night vision module and the adjustment of the infrared light source.

[0054] The data processing module is connected to the control module and is used to receive and process the image data and scanning data obtained from the night vision module and the three-dimensional laser scanner; obtain the target feature representation and the night environment feature representation from the image data and the scanning data according to the unsupervised variational autoencoder; input the target feature representation into the fusion device composed of the AISOA and SLSGAA algorithms to obtain the sensitivity value of the infrared camera; use the sensitivity value and the night environment feature representation as input parameters of the ISPPA scanning path planning algorithm to obtain the best infrared camera scanning path and generate a three-dimensional scanning result containing night environment information. In this embodiment, the unsupervised variational autoencoder (VAE) is used to extract the target feature representation and the night environment feature representation from the image data and the scanning data, which can capture the intrinsic structure of the data and effectively extract the key features. The extracted target feature representation is input into the fusion device composed of the AISOA (adaptive immune genetic algorithm) and the SLSGAA (adaptive learning genetic algorithm) algorithm to obtain the sensitivity value of the infrared camera, which can quickly and accurately calculate the sensitivity value. The sensitivity value and night environment characteristics are used as input parameters of ISPPA (Intelligent Scanning Path Planning Algorithm) to obtain the best infrared camera scanning path, which can adaptively adjust the scanning path according to the environment and target characteristics to improve scanning efficiency and accuracy. A three-dimensional scanning result containing night environment information is generated to provide data support for the perception and analysis of the night environment.

[0055] The coding module includes a visible light coding submodule and an infrared light coding submodule, wherein the visible light coding submodule is used to number the targets in a normal light environment, and the infrared light coding submodule is used to emit a recognizable coding signal under infrared light irradiation to identify and distinguish the targets at night or in a low light environment.

[0056] In an optional manner, the unsupervised variational autoencoder includes an input layer, a convolutional autoencoder layer, a sparse coding layer, a variational inference network layer, a decoding layer, and an output layer;

[0057] Among them, the sparse coding layer introduces KL divergence constraints; the variational inference network layer includes a latent variable mean layer, a latent variable logarithmic variance layer and a heavy parameter adoption layer; the decoding layer includes a deconvolution layer and a convolution layer.

[0058] In this embodiment, the unsupervised variational autoencoder (VAE) not only encodes and decodes the data, but also generates new data samples through latent variables. It has a strong generation ability, so that similar data points are also close in the latent space. The KL divergence constraint introduced by the sparse coding layer prevents the distribution of the latent variables from being too different from the prior distribution. The variational inference network layer can handle the uncertainty of the latent variables by introducing the mean and logarithmic variance of the latent variables and sampling using the reparameterization technique. The convolution layer and the deconvolution layer can effectively capture the spatial structure of the data. For example, the original data is received as input. The input data is encoded and features are extracted through the convolution operation. On the basis of encoding, the KL divergence constraint is introduced to make the distribution of the latent variables close to the prior distribution (such as the standard normal distribution). The latent variables are decoded by the deconvolution layer and the convolution layer to generate reconstructed data.

[0059] In an optional manner, the adoption formula of the heavy parameter adoption layer is:

[0060]

[0061] in, The latent variable mean layer is mapped to the mean vector of the latent variable; The log-variance vector of the latent variable is mapped to the log-variance layer of the latent variable; is the noise sampled from a standard normal distribution.

[0062] In this embodiment, by looking at the mean μ and logarithmic variance logσ² of the latent variables, the distribution of the target data can be analyzed. Similar targets have similar representations in the latent space. The input data is first encoded by the convolutional autoencoder layer, and then the mean μ and logarithmic variance logσ² of the latent variables are calculated by the sparse coding layer. The latent variable z is generated using the reparameterization technique. Finally, z is decoded into reconstructed data by the decoding layer.

[0063] In an optional manner, inputting the target feature representation into a fusion device composed of AISOA and SLSGAA algorithms to obtain the sensitivity value of the infrared camera further includes:

[0064] Input the target feature representation into the AISOA algorithm for feature transformation to obtain the first feature vector;

[0065] The first eigenvector is input into the SLSGAA algorithm for fusion to generate a second eigenvector;

[0066] The second eigenvector is mapped by a mapping function to obtain the sensitivity value of the infrared camera.

[0067] In this embodiment, the feature transformation capability of the AISOA algorithm and the fusion capability of the SLSGAA algorithm are combined to more effectively extract key information from the target feature representation and generate a feature vector that is highly correlated with the sensitivity of the infrared camera. By mapping the fused feature vector through a mapping function, the sensitivity value of the infrared camera can be accurately predicted to adjust the camera parameters.

[0068] In an optional manner, the method of using the sensitivity value and the nighttime environment characteristic representation as input parameters of an ISPPA scanning path planning algorithm to obtain an optimal infrared camera scanning path and generate a three-dimensional scanning result containing nighttime environment information further includes:

[0069] The sensitivity value and the night environment characteristic representation are read, wherein the night environment characteristic representation includes temperature distribution, brightness, location and size information of the target object.

[0070] Divide the scanning area into multiple sub-areas and evaluate the brightness and target object density of each sub-area;

[0071] Adjust the focal length and exposure time of the camera according to the brightness and sensitivity values ​​of each sub-area to adapt to the environmental conditions of different sub-areas;

[0072] Identify the target objects in the scanning area and prioritize them according to their size;

[0073] The scanning path is planned according to the A× heuristic search algorithm, and the infrared camera scanning path that maximizes the target detection rate is obtained to guide the scanning operation of the infrared camera.

[0074] In this embodiment, the camera parameters can be adjusted to adapt to different environmental conditions through the temperature distribution, brightness of the night environment, and the location and size information of the target object, thereby improving the efficiency of scanning. The A× heuristic search algorithm is used to plan the scanning path and find the path that maximizes the target detection rate, thereby ensuring that important targets are scanned first and reducing unnecessary scanning time and resource consumption. By identifying the target objects in the scanning area and prioritizing them according to their size, it can be ensured that important targets are scanned first and the overall target detection rate is improved. The generated scanning results contain three-dimensional information of the night environment and can more comprehensively display the characteristics of the scanning area.

[0075] Specifically, read the sensitivity value of the infrared camera and the night environment feature representation, including temperature distribution, brightness, location and size information of the target object. Divide the scanning area into multiple sub-areas. Evaluate the brightness of each sub-area and the density of the target object. Adjust the focal length and exposure time of the camera according to the brightness and sensitivity value of each sub-area. For example, in a sub-area with lower brightness, the exposure time can be increased to obtain a clearer image; in a sub-area with higher brightness, the exposure time can be appropriately reduced to avoid overexposure. Identify the target objects in the scanning area and prioritize them according to their size to ensure that important targets are scanned first. Use the A× heuristic search algorithm to plan the scanning path. During the planning process, consider factors such as the brightness of the sub-area, the density of the target object, and the priority of the target object. Calculate the infrared camera scanning path that maximizes the target detection rate. Guide the scanning operation of the infrared camera according to the planned scanning path. After the scanning is completed, generate a three-dimensional scanning result containing night environment information.

[0076] In an optional manner, the loss function of the unsupervised variational autoencoder is:

[0077]

[0078] Among them, N is the number of samples, M is the number of categories, is the true probability that the i-th sample belongs to category j, is the predicted probability that the i-th sample belongs to category j, J is the dimension of the latent space, is the mean of the latent variable, is the variance of the latent variable.

[0079] In this embodiment, VAE can learn the latent structure of data without category labels and generate new samples similar to the training data. Since the latent space of VAE is continuous, it allows exploration of the data space through interpolation and smooth transformation. Using the above loss function to train the VAE model, the reconstruction loss can be balanced during the training process. and KL divergence After training, latent variables are sampled from the prior distribution and a decoding layer is used to generate new 3D laser scanning target samples.

[0080] In an optional manner, the infrared light encoding submodule includes:

[0081] A coding signal generator, connected to the infrared light source, for generating a recognizable infrared coding signal;

[0082] The modulator is connected between the coding signal generator and the infrared light source, and is used to modulate the coding signal onto the infrared light source to control the infrared light source to flash according to the coding signal; wherein the infrared coding signal includes a plurality of pulse sequences, each pulse sequence having a preset pulse width, pulse interval and pulse frequency to distinguish different targets.

[0083] By generating a pulse sequence with a specific pulse width, pulse interval and pulse frequency for each target, the infrared coding signal can ensure the unique identification of each target and improve the accuracy of target identification in complex environments. Due to the uniqueness of the infrared coding signal, the stability of the system can be ensured in situations such as light changes and interference from other infrared sources.

[0084] In an optional manner, the night vision module further includes an infrared light source regulator;

[0085] The infrared light source regulator is connected to the control module and is used to dynamically adjust the luminous intensity, luminous angle and flashing frequency of the infrared light source according to the real-time image data and environmental information of the infrared camera to optimize the imaging effect of the infrared camera at night or in low light environments.

[0086] The infrared light source regulator dynamically adjusts the luminous intensity, luminous angle and flashing frequency of the infrared light source according to the real-time image data and environmental information of the infrared camera, and can maintain the best imaging effect in different night or low-light environments. By precisely controlling the luminous parameters of the infrared light source, the infrared camera can capture clearer image information. Especially in low-light environments, the enhanced infrared light source can significantly improve the brightness and contrast of the image, thereby helping to identify the target.

[0087] In an optional manner, the control module further includes an automatic calibration module for automatically adjusting the relative position between the infrared camera and the target according to the image data of the night vision module to ensure that the infrared camera is always aligned with the reflective surface of the target.

[0088] The automatic calibration module can adjust the relative position between the infrared camera and the target in real time based on the image data of the night vision module, ensuring that the camera is always aimed at the reflective surface of the target, significantly improving the image quality, and thus enhancing the recognition accuracy of the target. The traditional calibration method requires manual intervention, while the automatic calibration module can automatically complete this process without additional manual operation, which not only improves the convenience of the system, but also reduces the difficulty of operation and the error rate.

[0089] The solution provided by the present invention includes a target, including a reflector, which is used to provide a scanning point during the three-dimensional laser scanning process; a night vision module, including an infrared camera, an infrared enhancement filter and an image processor; wherein the infrared enhancement filter is used to enhance the sensitivity of the infrared camera to infrared light, and the image processor is used to convert the image captured by the camera into a processable electronic signal; a connecting mechanism is used to connect the night vision module and the target so that the infrared camera is aimed at the reflective surface of the target; a control module is connected to the three-dimensional laser scanner and the night vision module, and is used to receive the image data transmitted by the night vision module, and assist the three-dimensional laser scanner to perform night scanning based on the image data, and at the same time control the opening and closing of the night vision module and the adjustment of the infrared light source; a data processing module is connected to the control module, and is used to receive and process the data from the night vision module and the three-dimensional laser scanner The obtained image data and scan data; the target feature representation and the night environment feature representation are obtained from the image data and the scan data according to the unsupervised variational autoencoder; the target feature representation is input into the fusion device composed of the AISOA and SLSGAA algorithms to obtain the sensitivity value of the infrared camera; the sensitivity value and the night environment feature representation are used as input parameters of the ISPPA scanning path planning algorithm to obtain the best infrared camera scanning path, and generate a three-dimensional scanning result containing night environment information; the encoding module includes a visible light encoding submodule and an infrared light encoding submodule, wherein the visible light encoding submodule is used to number the target under a conventional light environment, and the infrared light encoding submodule is used to emit a recognizable encoding signal under infrared light irradiation to identify and distinguish the target at night or in a low light environment. The present invention integrates an infrared camera and an infrared enhancement filter, so that the target can capture a clear infrared image at night or in a low light environment, thereby ensuring that the three-dimensional laser scanner can accurately obtain the position information of the scanning point, improving the accuracy of night scanning, and obtaining high-quality three-dimensional scanning data even in insufficient light. The control module automatically adjusts the brightness and angle of the infrared light source, so that the target can provide stable scanning support under various lighting conditions, enhancing its environmental adaptability. The data processing module uses an unsupervised variational autoencoder and algorithm combination (AISOA and SLSGAA), which can quickly and accurately extract target feature representation and night environment feature representation from image data and scanning data. The ISPPA scanning path planning algorithm plans the best infrared camera scanning path, which improves scanning efficiency and reduces unnecessary scanning time and resource consumption. The encoding module combines visible light encoding submodules and infrared light encoding submodules to achieve dual encoding recognition in conventional light environments and infrared light environments, which not only improves the recognition accuracy and reliability of the target, but also enables the target to be accurately identified and located at night or in low light environments. In addition, the intelligence and automation level of the entire scanning system are improved.

[0090] Figure 2The following is a schematic diagram showing the use process of a target for three-dimensional laser scanning with night vision function according to an embodiment of the present invention. Specifically, Figure 2 As shown, the following steps are included:

[0091] Step S201, starting the night vision module through the control module to capture an infrared image including the target reflective surface.

[0092] Step S202, the image data captured by the infrared camera is enhanced by an infrared enhancement filter, and then transmitted to an image processor for processing, and scanning data is obtained from a three-dimensional laser scanner.

[0093] Step S203, using an unsupervised variational autoencoder to process the image data and the scanned data, and extracting target feature representation and nighttime environment feature representation.

[0094] Step S204, through the fusion device composed of AISOA and SLSGAA algorithms, the target feature representation is converted into the sensitivity value of the infrared camera, and the optimal infrared camera scanning path is planned in combination with the A× heuristic search algorithm.

[0095] Step S205, instructing the infrared camera to scan according to the planned scanning path; identifying and distinguishing different targets by infrared coding signals for numbering the targets; and outputting a three-dimensional scanning result including night environment information.

[0096] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into multiple submodules or subunits or subcomponents. All features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature that provides the same, equivalent or similar purpose. In addition, those skilled in the art will appreciate that, although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and to form different embodiments. For example, in the following claims, any one of the claimed embodiments may be used in any combination. The present invention may be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim that lists several means, several of these means may be embodied by the same hardware item. The steps in the above embodiments should not be understood as limiting the execution order unless otherwise specified.

Claims

1. A three-dimensional laser scanning target with night vision function, characterized in that: include: A target, including a reflector, is used to provide a scanning point during a three-dimensional laser scanning process; A night vision module, comprising an infrared camera, an infrared enhancement filter and an image processor; wherein the infrared enhancement filter is used to enhance the sensitivity of the infrared camera to infrared light, and the image processor is used to convert the image captured by the camera into a processable electronic signal; A connecting mechanism, used to connect the night vision module and the target so that the infrared camera is aimed at the reflective surface of the target; A control module is connected to the 3D laser scanner and the night vision module, and is used to receive image data transmitted by the night vision module, and assist the 3D laser scanner in performing night scanning based on the image data, and control the opening and closing of the night vision module and the adjustment of the infrared light source; A data processing module is connected to the control module and is used to receive and process image data and scan data obtained from the night vision module and the three-dimensional laser scanner; obtain target feature representation and night environment feature representation from the image data and scan data according to an unsupervised variational autoencoder; input the target feature representation into a fusion device composed of AISOA and SLSGAA algorithms to obtain a sensitivity value of the infrared camera; use the sensitivity value and the night environment feature representation as input parameters of the ISPPA scanning path planning algorithm to obtain an optimal infrared camera scanning path and generate a three-dimensional scanning result containing night environment information; The coding module includes a visible light coding submodule and an infrared light coding submodule, wherein the visible light coding submodule is used to number the targets in a normal light environment, and the infrared light coding submodule is used to emit a recognizable coding signal under infrared light irradiation to identify and distinguish the targets at night or in a low light environment.

2. The three-dimensional laser scanning target with night vision function according to claim 1, characterized in that: The unsupervised variational autoencoder comprises an input layer, a convolutional autoencoder layer, a sparse coding layer, a variational inference network layer, a decoding layer and an output layer; Among them, the sparse coding layer introduces KL divergence constraints; the variational inference network layer includes a latent variable mean layer, a latent variable logarithmic variance layer and a heavy parameter adoption layer; the decoding layer includes a deconvolution layer and a convolution layer.

3. The three-dimensional laser scanning target with night vision function according to claim 2, characterized in that: The adoption formula of the heavy parameter adoption layer is: in, The latent variable mean layer is mapped to the mean vector of the latent variable; The log-variance vector of the latent variable is mapped to the log-variance layer of the latent variable; is the noise sampled from a standard normal distribution.

4. The three-dimensional laser scanning target with night vision function according to claim 1, characterized in that: The step of inputting the target feature representation into a fusion device composed of AISOA and SLSGAA algorithms to obtain the sensitivity value of the infrared camera further includes: Input the target feature representation into the AISOA algorithm for feature transformation to obtain the first feature vector; The first eigenvector is input into the SLSGAA algorithm for fusion to generate a second eigenvector; The second eigenvector is mapped by a mapping function to obtain the sensitivity value of the infrared camera.

5. The three-dimensional laser scanning target with night vision function according to claim 1, characterized in that: The method of using the sensitivity value and the nighttime environment characteristic representation as input parameters of the ISPPA scanning path planning algorithm to obtain the best infrared camera scanning path and generate a three-dimensional scanning result containing nighttime environment information further includes: Reading sensitivity values ​​and nighttime environment characteristics, wherein the nighttime environment characteristics include temperature distribution, brightness, location and size information of the target object; Divide the scanning area into multiple sub-areas and evaluate the brightness and target object density of each sub-area; Adjust the focal length and exposure time of the camera according to the brightness and sensitivity values ​​of each sub-area to adapt to the environmental conditions of different sub-areas; Identify the target objects in the scanning area and prioritize them according to their size; The scanning path is planned according to the A× heuristic search algorithm, and the infrared camera scanning path that maximizes the target detection rate is obtained to guide the scanning operation of the infrared camera.

6. The three-dimensional laser scanning target with night vision function according to claim 2, characterized in that: The loss function of the unsupervised variational autoencoder is: Among them, N is the number of samples, M is the number of categories, is the true probability that the i-th sample belongs to category j, is the predicted probability that the i-th sample belongs to category j, J is the dimension of the latent space, is the mean of the latent variable, is the variance of the latent variable.

7. The three-dimensional laser scanning target with night vision function according to claim 1, characterized in that: The infrared light encoding submodule comprises: A coding signal generator, connected to the infrared light source, for generating a recognizable infrared coding signal; The modulator is connected between the coding signal generator and the infrared light source, and is used to modulate the coding signal onto the infrared light source to control the infrared light source to flash according to the coding signal; wherein the infrared coding signal includes a plurality of pulse sequences, each pulse sequence having a preset pulse width, pulse interval and pulse frequency to distinguish different targets.

8. The three-dimensional laser scanning target with night vision function according to claim 1, characterized in that: The night vision module also includes an infrared light source regulator; The infrared light source regulator is connected to the control module and is used to dynamically adjust the luminous intensity, luminous angle and flashing frequency of the infrared light source according to the real-time image data and environmental information of the infrared camera to optimize the imaging effect of the infrared camera at night or in low light environments.

9. The three-dimensional laser scanning target with night vision function according to claim 1, characterized in that: The control module also includes an automatic calibration module for automatically adjusting the relative position between the infrared camera and the target according to the image data of the night vision module to ensure that the infrared camera is always aligned with the reflective surface of the target.

10. A method for using a three-dimensional laser scanning target with night vision function according to any one of claims 1 to 9, characterized in that: include: The night vision module is activated through the control module to capture an infrared image including the reflective surface of the target; The image data captured by the infrared camera is enhanced by an infrared enhancement filter and then transmitted to the image processor for processing, and the scanning data is obtained from the 3D laser scanner; Use unsupervised variational autoencoders to process image data and scan data to extract target feature representation and night environment feature representation; The target feature representation is converted into the sensitivity value of the infrared camera through the fusion of AISOA and SLSGAA algorithms, and the optimal infrared camera scanning path is planned in combination with the A× heuristic search algorithm; Instruct the infrared camera to scan according to the planned scanning path; Infrared coding signals are used to identify and distinguish different targets by numbering the targets; and three-dimensional scanning results containing night environment information are output.

Citation Information

Cited By

  • Optical line scanning path generation method facing plane area

    CN121346648A