Intelligent three-dimensional visual perception method based on parking space scene

Abstract

The invention discloses an intelligent three-dimensional visual perception method based on a parking space scene, and the method comprises the following steps: S1, responding to the illumination change of a parking space environment, and automatically switching a passive or active visual mode; s2, in a passive vision mode, capturing a natural light image by using a binocular camera, and reconstructing a three-dimensional scene; s3, in the active visual mode, projecting a coding pattern, and reconstructing a three-dimensional scene through pattern change; s4, performing semantic segmentation on different objects in the passive visual mode or the three-dimensional parking space scene image in the passive visual mode, and endowing semantic tags; s5, mapping the target object in the two-dimensional image into the three-dimensional scene, and obtaining the position and posture of the target object; by automatically switching the passive vision mode and the active vision mode, the adaptability and the recognition precision of the parking space scene under different illumination conditions are remarkably improved; in a passive vision mode, the system utilizes natural light to capture images, a binocular camera system and a stereoscopic vision technology to accurately reconstruct a three-dimensional scene.

Description

Technical Field

[0001] The present invention relates to the field of visual perception technology, and in particular to an intelligent three-dimensional visual perception method based on parking space scenes. Background Art

[0002] With the acceleration of urbanization, parking issues are becoming an increasingly difficult problem in urban management. Traditional parking lot management relies on manual operations and simple monitoring systems, which are unable to cope with large-scale, dynamically changing parking demands. Especially under complex lighting conditions, such as at night or in cloudy weather, traditional visual perception systems often have difficulty accurately identifying parking space status and vehicle location, resulting in low parking lot operation efficiency and poor user experience.

[0003] In existing technologies, the visual perception of parking scenes mainly relies on passive vision systems, that is, using cameras to capture images under natural light to perform scene analysis; however, this method will significantly reduce image quality when lighting conditions are poor, such as at night or on cloudy days, resulting in reduced recognition accuracy. In addition, passive vision systems cannot work in completely dark environments, limiting their scope of application.

[0004] Therefore, it is necessary to provide an intelligent three-dimensional visual perception method based on parking space scenes to solve the above technical problems. Summary of the invention

[0005] The present invention overcomes the deficiencies of the prior art and provides an intelligent three-dimensional visual perception method based on parking space scenes.

[0006] To achieve the above object, the technical solution adopted by the present invention is: an intelligent three-dimensional visual perception method based on parking scene, comprising the following steps:

[0007] S1: automatically switching between passive vision mode and active vision mode in response to changes in ambient lighting in the parking space;

[0008] S2: In passive vision mode, a binocular camera is used to capture natural light images and reconstruct the 3D scene;

[0009] S3: In active vision mode, the coded pattern is projected and the 3D scene is reconstructed through pattern changes;

[0010] S4: performing semantic segmentation on different objects in the three-dimensional parking scene image in the passive vision mode or in the passive vision mode, and assigning semantic labels;

[0011] S5: Map the target object in the two-dimensional image into the three-dimensional scene to obtain its position and posture.

[0012] In a preferred embodiment of the present invention, in S2, a binocular camera system is used to capture binocular images under natural light, and the binocular images are registered to determine the correspondence between the binocular images. Stereo vision technology is used in combination with the registered binocular images to reconstruct a three-dimensional parking scene image in a passive vision mode.

[0013] In a preferred embodiment of the present invention, the binocular image registration determines the spatial correspondence between the binocular images by identifying and matching the same feature points in the binocular images, thereby providing accurate disparity information for three-dimensional reconstruction.

[0014] In a preferred embodiment of the present invention, the stereoscopic vision technology calculates the depth information of each point in the scene by analyzing the parallax in the image captured by the binocular camera, thereby reconstructing a three-dimensional parking scene map.

[0015] In a preferred embodiment of the present invention, in S3, a specifically coded optical pattern is projected onto the parking scene, the position and shape changes of the optical pattern in the parking scene are identified and analyzed, and a three-dimensional parking scene image in the active vision mode is reconstructed.

[0016] In a preferred embodiment of the present invention, the projecting of the specifically coded optical pattern is accomplished by using a projection device to project a light beam or pattern with specific geometric or coded information onto the parking space scene, so that the binocular camera system can recognize and utilize changes in the pattern to calculate the three-dimensional structure of the scene.

[0017] In a preferred embodiment of the present invention, in S4, the different objects include but are not limited to: vehicles, parking space lines, obstacles, ground markings, etc., and semantic segmentation can identify and distinguish different objects in a three-dimensional scene.

[0018] In a preferred embodiment of the present invention, in S5, the original two-dimensional image captured by the binocular camera system is used, and the target object is located in the original two-dimensional image. According to the registration relationship between the original two-dimensional image and the three-dimensional parking scene image, the target object is mapped into the three-dimensional parking scene image to obtain the position and posture information of the target object.

[0019] In a preferred embodiment of the present invention, the original two-dimensional image is the basis for achieving accurate three-dimensional scene reconstruction and target object positioning.

[0020] In a preferred embodiment of the present invention, an intelligent three-dimensional visual perception device based on a parking scene comprises:

[0021] Ambient light sensor: used to monitor the light intensity of the current environment in real time and control the switching of visual modes based on the detection results;

[0022] Binocular camera system: used to capture natural light images in passive vision mode and projected coded pattern images in active vision mode;

[0023] Pattern projection device: used to project a specific coded optical pattern onto the parking scene in active vision mode;

[0024] Image processing unit: used to process images captured by the binocular camera system, perform image registration, parallax analysis, depth information calculation, etc.;

[0025] Semantic segmentation module: used to identify and distinguish different objects in a three-dimensional scene (such as vehicles, parking lines, obstacles, ground markings, etc.).

[0026] The present invention solves the defects existing in the background technology and has the following beneficial effects:

[0027] (1) The present invention provides an intelligent three-dimensional visual perception method based on parking space scenes. By automatically switching between passive vision mode and active vision mode, the adaptability and recognition accuracy of parking space scenes under different lighting conditions are significantly improved. In the passive vision mode, the system uses natural light to capture images, and accurately reconstructs the three-dimensional scene through a binocular camera system and stereo vision technology, ensuring efficient operation when there is sufficient light. In an environment with insufficient light, the active vision mode projects a coded pattern, which can achieve accurate three-dimensional scene reconstruction even in a completely dark environment.

[0028] (2) In the present invention, the semantic segmentation module can accurately identify and distinguish different objects such as vehicles, parking space lines, obstacles, ground markings, etc. in the three-dimensional scene, and assign them corresponding semantic labels, which not only improves the system's understanding of the parking environment, but also provides a solid foundation for subsequent target positioning and navigation; by mapping the target object in the two-dimensional image into the three-dimensional scene, the system can obtain the precise position and posture information of the target object.

[0029] (3) The intelligent three-dimensional visual perception device for parking scenes of the present invention integrates multiple components including an ambient light sensor, a binocular camera system, a pattern projection device, an image processing unit and a semantic segmentation module; the ambient light sensor monitors the light intensity in real time and intelligently controls the switching of visual modes; the binocular camera system and the pattern projection device work in passive and active visual modes respectively to capture necessary image information; the image processing unit and the semantic segmentation module are responsible for the depth analysis and object recognition of the image; the integrated device design not only improves the operating efficiency of the system, but also simplifies installation and maintenance work, thereby reducing the overall cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 is a schematic diagram of a method flow of a preferred embodiment of the present invention;

[0032] Figure 2 It is a schematic diagram of the device mechanism of a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0035] like Figure 1 As shown, the present invention provides an intelligent three-dimensional visual perception method based on parking space scenes, comprising the following steps:

[0036] S1: automatically switching between passive vision mode and active vision mode in response to changes in ambient lighting in the parking space;

[0037] S2: In passive vision mode, a binocular camera is used to capture natural light images and reconstruct the 3D scene;

[0038] S3: In active vision mode, the coded pattern is projected and the 3D scene is reconstructed through pattern changes;

[0039] S4: performing semantic segmentation on different objects in the three-dimensional parking scene image in the passive vision mode or in the passive vision mode, and assigning semantic labels;

[0040] S5: Map the target object in the two-dimensional image into the three-dimensional scene to obtain its position and posture.

[0041] Specifically, in S1, the visual 3D perception technology of the passive vision mode or the mixed active vision mode mainly solves the adverse effects of the illumination difference between day and night on visual perception. By combining the passive and active binocular stereo vision technology, high-precision parking scene measurement under different lighting conditions is achieved.

[0042] Specifically, in S2, the binocular images under natural light are captured by the binocular camera system 102, the binocular images are registered to determine the correspondence between the binocular images, and the three-dimensional parking space scene image in the passive mode is reconstructed by combining the registered binocular images with the stereo vision technology;

[0043] Binocular image registration determines the spatial correspondence between them by identifying and matching the same feature points in the binocular images, thereby providing accurate disparity information for 3D reconstruction;

[0044] Stereo vision technology analyzes the parallax in the images captured by the binocular camera and calculates the depth information of each point in the scene, thereby reconstructing a three-dimensional parking scene map.

[0045] In summary, S2 realizes the conversion from two-dimensional images under natural light to three-dimensional scenes through binocular camera system 102 and stereo vision technology, providing accurate three-dimensional spatial information for intelligent visual perception system.

[0046] Specifically, in S3, by projecting a specific coded optical pattern onto the parking scene, the position and shape changes of the optical pattern in the parking scene are identified and analyzed, and a three-dimensional parking scene image in the active mode is reconstructed;

[0047] Projecting a specific coded optical pattern projects a light beam or pattern with specific geometric or coded information onto the parking space scene using a projection device, so that the binocular camera system 102 can recognize and use changes in the pattern to calculate the three-dimensional structure of the scene.

[0048] In summary, S3 actively projects coded patterns and analyzes their reflections in the scene, enabling the scene-intelligent 3D visual perception method to reconstruct the 3D scene image of the parking space even in environments with poor lighting conditions or complete darkness. The implementation of the active visual mode significantly improves the adaptability of the method under various lighting conditions and the accuracy of 3D reconstruction.

[0049] Specifically, in S4, different objects include but are not limited to: vehicles, parking space lines, obstacles, ground markings, etc., and semantic segmentation can identify and distinguish different objects in the three-dimensional scene; in S5, the original two-dimensional image captured by the binocular camera system 102 is used, and the target object is located in the original two-dimensional image. According to the registration relationship between the original two-dimensional image and the three-dimensional parking space scene image, the target object is mapped to the three-dimensional parking space scene image to obtain the position and posture information of the target object; the original two-dimensional image is the basis for achieving accurate three-dimensional scene reconstruction and target object positioning.

[0050] It should be noted that semantic segmentation is performed on the reconstructed three-dimensional image of the parking scene, different objects in the image (such as ground locks, parking space lines, vehicles, etc.) are distinguished, and their respective semantic labels are assigned, which helps the system to accurately understand the structure and content of the parking scene.

[0051] Based on semantic segmentation, the system accurately locates the target area (such as the parking area for the vehicle). By establishing the registration relationship between the two-dimensional and three-dimensional information of the scene, the system can roughly locate the target area on the two-dimensional level, and then map it to the three-dimensional level for precise positioning, ensuring that the system can accurately identify the vehicle's position and posture information.

[0052] like Figure 2 As shown, an intelligent three-dimensional visual perception device based on parking scene, the device includes:

[0053] Ambient light sensor 101: used to monitor the light intensity of the current environment in real time and control the switching of the visual mode according to the detection result;

[0054] Binocular camera system 102: used to capture natural light images in passive vision mode and projected coded pattern images in active vision mode;

[0055] Pattern projection device 103: used to project a specific coded optical pattern onto the parking space scene in active vision mode;

[0056] Image processing unit 104: used to process the images captured by the binocular camera system 102, perform image registration, parallax analysis, depth information calculation, etc.;

[0057] Semantic segmentation module 105: used to identify and distinguish different objects (such as vehicles, parking space lines, obstacles, ground markings, etc.) in a three-dimensional scene.

[0058] Example:

[0059] S1: The ambient light sensor 101 monitors the light intensity in the parking lot in real time; when the light intensity is detected to be lower than a preset threshold, the system automatically switches to the active vision mode to adapt to the low-light environment; when there is sufficient light, the system maintains the passive vision mode and uses natural light for image capture;

[0060] S2: In the passive vision mode, the binocular camera system 102 automatically captures images under natural light; the image processing unit 104 registers the received binocular images, identifies and matches the same feature points in the images, and determines the spatial correspondence between them; using stereo vision technology, according to the parallax information in the registered binocular images, the depth information of each point in the scene is calculated, thereby reconstructing a three-dimensional parking scene map;

[0061] S3: In the active vision mode, the pattern projection device 103 projects a specific coded optical pattern to the parking scene; the binocular camera system 102 captures the projected coded pattern image; the image processing unit 104 identifies and analyzes the position and shape changes of the optical pattern in the parking scene, and uses these changes to calculate the three-dimensional structure of the scene, and reconstructs the three-dimensional parking scene image in the active mode;

[0062] S4: The semantic segmentation module 105 identifies and distinguishes vehicles, parking space lines, obstacles, ground markings, etc. in the three-dimensional scene image, and assigns corresponding semantic labels; through semantic segmentation, the system can accurately understand the structure and content of the parking space scene, providing a basis for subsequent target positioning and navigation;

[0063] S5: Locate the target object in the original two-dimensional image captured in the binocular camera system 102; map the target object into the three-dimensional parking scene image according to the registration relationship between the original two-dimensional image and the three-dimensional parking scene image; obtain position and posture information: through the mapping process, the system can obtain the precise position and posture information of the target object in the three-dimensional space, providing key data for intelligent parking management.

[0064] To sum up: by implementing an intelligent three-dimensional visual perception method based on parking space scenes, the system can provide high-precision three-dimensional scene reconstruction and target positioning around the clock under various lighting conditions, significantly improving the operational efficiency, safety and user experience of parking lots; by automatically switching visual modes, using binocular cameras and coded pattern projection technology, as well as performing semantic segmentation and precise mapping, it not only optimizes resource utilization and reduces operating costs, but also provides strong support for the construction of intelligent transportation systems, thereby realizing the automation and intelligence of parking lot management.

[0065] The above is based on the ideal embodiment of the present invention. Through the above description, relevant personnel can make various changes and modifications without departing from the technical concept of the present invention. The technical scope of the present invention is not limited to the content in the specification, and the technical scope must be determined according to the scope of the claims.

Claims

1. An intelligent three-dimensional visual perception method based on parking scene, characterized in that: The following steps are involved: S1: automatically switching between passive vision mode and active vision mode in response to changes in ambient lighting in the parking space; S2: In passive vision mode, a binocular camera is used to capture natural light images and reconstruct the 3D scene; S3: In active vision mode, the coded pattern is projected and the 3D scene is reconstructed through pattern changes; S4: performing semantic segmentation on different objects in the three-dimensional parking scene image in the passive vision mode or in the passive vision mode, and assigning semantic labels; S5: Map the target object in the two-dimensional image into the three-dimensional scene to obtain its position and posture.

2. The method of intelligent three-dimensional visual perception based on parking scene according to claim 1, characterized in that: In S2, the binocular camera system 102 is used to capture binocular images under natural light, and the binocular images are registered to determine the correspondence between the binocular images. Stereo vision technology is used to combine the registered binocular images to reconstruct a three-dimensional parking scene image in the passive vision mode.

3. The method of intelligent three-dimensional visual perception based on parking scene according to claim 2, characterized in that: The binocular image registration determines the spatial correspondence between the binocular images by identifying and matching the same feature points in the binocular images, thereby providing accurate disparity information for three-dimensional reconstruction.

4. The method of intelligent three-dimensional visual perception based on parking scene according to claim 2, characterized in that: The stereo vision technology calculates the depth information of each point in the scene by analyzing the parallax in the image captured by the binocular camera, thereby reconstructing a three-dimensional parking space scene map.

5. The method of intelligent three-dimensional visual perception based on parking scene according to claim 1, characterized in that: In S3, a specifically coded optical pattern is projected onto the parking scene, and the position and shape changes of the optical pattern in the parking scene are identified and analyzed to reconstruct a three-dimensional parking scene image in the active vision mode.

6. The method of intelligent three-dimensional visual perception based on parking scene according to claim 5, characterized in that: The projecting of the specifically coded optical pattern is performed by using a projection device to project a light beam or pattern with specific geometric or coded information to the parking space scene, so that the binocular camera system 102 can recognize and use the change of the pattern to calculate the three-dimensional structure of the scene.

7. The method of intelligent three-dimensional visual perception based on parking scene according to claim 1, characterized in that: In S4, the different objects include but are not limited to: vehicles, parking space lines, obstacles, ground markings, etc., and semantic segmentation can identify and distinguish different objects in a three-dimensional scene.

8. The method of intelligent three-dimensional visual perception based on parking scene according to claim 1, characterized in that: In S5, the original two-dimensional image captured by the binocular camera system 102 is used to locate the target object in the original two-dimensional image. According to the registration relationship between the original two-dimensional image and the three-dimensional parking scene image, the target object is mapped into the three-dimensional parking scene image to obtain the position and posture information of the target object.

9. The method of intelligent three-dimensional visual perception based on parking scene according to claim 8, characterized in that: The original two-dimensional image is the basis for achieving accurate three-dimensional scene reconstruction and target object positioning.

10. The intelligent three-dimensional visual perception device based on parking scene according to any one of claims 1 to 9, characterized in that: The device comprises: Ambient light sensor 101: used to monitor the light intensity of the current environment in real time and control the switching of the visual mode according to the detection result; Binocular camera system 102: used to capture natural light images in passive vision mode and projected coded pattern images in active vision mode; Pattern projection device 103: used to project a specific coded optical pattern onto the parking space scene in active vision mode; Image processing unit 104: used to process the images captured by the binocular camera system 102, perform image registration, parallax analysis, depth information calculation, etc.; Semantic segmentation module 105: used to identify and distinguish different objects (such as vehicles, parking space lines, obstacles, ground markings, etc.) in a three-dimensional scene.

Images