WO2018049998A1 - 交通标志牌信息获取方法及装置 - Google Patents
交通标志牌信息获取方法及装置 Download PDFInfo
- Publication number
- WO2018049998A1 WO2018049998A1 PCT/CN2017/100358 CN2017100358W WO2018049998A1 WO 2018049998 A1 WO2018049998 A1 WO 2018049998A1 CN 2017100358 W CN2017100358 W CN 2017100358W WO 2018049998 A1 WO2018049998 A1 WO 2018049998A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- traffic sign
- laser
- points
- image
- data
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
- G06T7/521—Depth or shape recovery from laser ranging, e.g. using interferometry; from the projection of structured light
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/12—Details of acquisition arrangements; Constructional details thereof
- G06V10/14—Optical characteristics of the device performing the acquisition or on the illumination arrangements
- G06V10/143—Sensing or illuminating at different wavelengths
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/582—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of traffic signs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/469—Contour-based spatial representations, e.g. vector-coding
- G06V10/476—Contour-based spatial representations, e.g. vector-coding using statistical shape modelling, e.g. point distribution models
Definitions
- the present application relates to the field of mapping technology, and in particular, to a method and device for acquiring traffic sign information.
- the signage information of traffic signs is one of the important components in high-precision maps.
- the sign information of the traffic sign is generally obtained by using an image collecting device to collect an image of the area containing the traffic sign, and performing image recognition on the collected image to detect the traffic sign therein.
- the embodiment of the present application provides a method and an apparatus for acquiring traffic sign information.
- the optical data includes data of a plurality of laser points
- a traffic sign information acquiring apparatus of an embodiment of the present application may include a processor and a memory, where the computer stores computer readable instructions, and the instructions may cause the processor to perform the following operations:
- the laser data comprising data of a plurality of laser points
- the optical data includes data of a plurality of laser points
- the laser is an active light source
- the accuracy of the subsequent acquisition of the sign information can be improved without being affected by the illumination.
- the plurality of laser points may be spatially clustered according to the data of the plurality of laser points to obtain a point set of the traffic sign, and image data of the traffic sign is determined based on the set of points, and the image data is imaged. Identify and other operations to obtain the sign information of the traffic sign. Since the laser point data includes the coordinates of each laser point, a series of operations based on the coordinates can reduce the error of acquiring the sign information, thereby improving the accuracy, efficiency, and robustness of the sign information for acquiring the traffic sign.
- FIG. 1A is a schematic diagram of an in-vehicle mobile measurement system according to an embodiment of the present application.
- FIG. 1B is a flowchart of a method for acquiring sign information provided by an embodiment of the present application.
- FIG. 2A is a schematic structural diagram of a sign information acquiring apparatus according to an embodiment of the present application.
- FIG. 2B is a schematic structural diagram of a scanning module according to an embodiment of the present application.
- FIG. 2C is a schematic structural diagram of a determining module according to an embodiment of the present application.
- FIG. 2D is a schematic structural diagram of another determining module provided by an embodiment of the present application.
- FIG. 2E is a schematic structural diagram of a determining unit according to an embodiment of the present application.
- FIG. 2F is a schematic structural diagram of an acquisition module according to an embodiment of the present application.
- FIG. 3 is a schematic structural diagram of a sign information acquiring apparatus according to an embodiment of the present application.
- FIG. 4 is a schematic diagram of an information processing system according to an embodiment of the present application.
- FIG. 5 is a flowchart of a method for acquiring traffic sign information provided by an embodiment of the present application.
- High-precision maps are an important aid to autonomous driving technology, and high-precision maps can be used to inform the car in front of the direction of travel and road conditions.
- Traffic signs are an important part of high-precision maps.
- the embodiment of the present application provides a method for acquiring traffic sign information, which can be implemented by an information processing system.
- the information processing system 40 can include a mobile collection device 41 and a data processing center 42.
- the mobile collection device 41 is a mobile device for collecting traffic sign information.
- the mobile collection device 41 can be moved by loading on a movable person or thing (eg, a vehicle, etc.), such as a handheld device, an in-vehicle device, or the like.
- the mobile acquisition device 41 can include a laser scanning system 411 and an imaging imaging system 412.
- the laser scanning system 411 is used to perform laser scanning on the scene to obtain laser data of the scene.
- laser scanning system 411 can include a laser emitting device and a receiving detecting device.
- the laser emitting device is configured to emit laser light to the scene; the receiving detecting device is configured to detect the laser light reflected back from the scene to generate laser data.
- laser scanning system 411 can include a two- or three-dimensional laser scanning device for laser scanning a scene to obtain laser data.
- the laser data can be embodied as a collection of data from multiple laser points, or a point cloud, or other form.
- a laser point cloud is a collection of a large number of laser data points obtained by laser scanning.
- the laser data may include the position (two-dimensional or three-dimensional coordinates) of each laser spot and the reflection characteristic value.
- the image imaging system 412 includes an imaging device for capturing a scene to obtain an image of the scene.
- the camera device may include a lens and an image sensor such as a CCD, a CMOS, or the like. Only one mobile acquisition device 41 is shown in FIG. 4 as an example to illustrate an implementation method of the technical solution of the present application.
- the information processing system 40 may include multiple mobile collection devices 41.
- Data processing center 42 may include one or more data processing devices, and thus data processing center 42 may include one or more processors 421.
- Data processing center 42 may also include a memory 422 and a communication interface 423.
- Data processing center 42 may establish some kind of connection with mobile acquisition device 41 via communication interface 423 for communication.
- This connection can be a wired connection (such as a connection established using a certain cable or socket), or a wireless connection (such as Wi-Fi, cellular mobile communication, radio frequency, Bluetooth, infrared, etc.); it can be a direct connection, It can also be a connection established through some kind of network, and so on.
- the data processing center 42 can acquire images and laser data of the scenes collected by the mobile collection device 41 through the communication interface 423, and process the data to extract information of the traffic signs in the scene.
- an information extraction module 425 can be stored in the memory 422.
- the information extraction module 425 can be stored in the memory 422 in the form of computer readable instructions and can be executed by the processor 421 to implement the traffic sign information acquisition method of various embodiments of the present application.
- FIG. 5 is a flowchart of a method for acquiring traffic sign information according to an embodiment of the present application.
- the method 50 can be performed by the data processing center 42. As shown in FIG. 5, the method 50 can include the following steps.
- Step 501 Acquire an image of the scene obtained by capturing an image of a traffic sign using an imaging device, and acquire laser data of the scene obtained by performing laser scanning on the area, where the laser data includes multiple lasers. Point data.
- Step 502 spatially clustering laser points in the laser data to obtain at least two candidate point sets.
- the spatial clustering means that the laser spots are grouped according to the spatial position based on the position data of each laser spot in the laser data, and each group is used as a set of candidate points. For example, laser points that are closer to each other are grouped into groups, and distant laser points are grouped into different groups.
- the laser spot in each set of candidate points may be formed by a laser reflected from the surface of the same object.
- Each set of candidate points can be embodied as a collection of data for a set of points, or as a point cloud.
- Step 503 Acquire a spatial distribution feature of each candidate point set, and determine a point set corresponding to the traffic sign from the at least two candidate point sets according to the spatial distribution feature.
- Step 504 Extract image data of the traffic sign from the image by using a point set corresponding to the traffic sign, and extract sign information of the traffic sign from the image data.
- the scene including the traffic sign refers to a physical environment (for example, on both sides of the road, above the road, etc.) provided with the traffic sign, also referred to as an area including the traffic sign.
- the traffic sign has the characteristic of high reflectivity to the laser, and the traffic sign in the scene image is positioned by using the laser data of the traffic sign, and then only the location of the traffic sign in the scene image is used for traffic.
- the extraction of the sign information can improve the recognition accuracy of the traffic sign.
- acquiring the laser data of the scene may include: acquiring a plurality of second laser points obtained by the laser scanning and a reflection characteristic value of each of the second laser points; and using the second laser point to select a reflection characteristic value greater than Or a plurality of laser spots equal to a preset threshold form the laser data.
- the laser data is formed at a point with better reflection characteristics for subsequent calculation, which can significantly reduce the data amount of the laser data and improve the calculation efficiency.
- the reflection characteristic value may be a reflection intensity value, a reflectance value, or the like.
- spatially clustering the plurality of laser points to obtain the at least two candidate point sets may include: grouping the plurality of laser points in the laser data with a preset distance threshold to form the at least A set of two candidate points, wherein a spatial distance between laser points in different sets of candidate points is greater than the distance threshold.
- the spatial distance between laser points can be obtained using Euclidean distance or a similar algorithm.
- obtaining a spatial distribution feature of each candidate point set, and determining, according to the spatial distribution feature, the point set corresponding to the traffic sign from the at least two candidate point sets may include: acquiring the at least two a spatial distribution feature value of the laser spot in at least two dimensions of the space in each candidate point set in the candidate point set; using the at least two preset spatial distribution feature thresholds and the spatial distribution feature value of each candidate point set from the And selecting at least one candidate point set among the at least two candidate point sets as the point set corresponding to the traffic sign.
- the spatial distribution feature value may be selected from: a distribution feature value in length, a distribution feature value in width, a distribution feature value in height, a linear distribution feature value, a planar distribution feature value, a body shape distribution feature value, a curvature, Fits the plane feature values.
- the preset spatial distribution feature threshold may be determined according to a size range and a shape feature of the traffic sign. In this way, the set of points corresponding to the traffic signs can be determined from the plurality of candidate point sets.
- extracting image data of the traffic sign from the image by using a point set corresponding to the traffic sign may include: according to a preset parameter and a point set corresponding to the traffic sign
- the location in the set of laser points of the scene determines the traffic
- the position of the image data of the sign in the image of the scene, the image data of the traffic sign is extracted from the position in the image.
- the position of the laser scanning device and the position and posture of the imaging device are relatively fixed, and the conversion relationship between the coordinate systems of the obtained data is called a conversion parameter (also called an external orientation element), and the conversion parameter (or parameter matrix) is utilized.
- the laser point set can be accurately projected onto the image.
- the point set of the traffic sign when determining the position of the image data of the traffic sign in the scene image according to the point cloud of the traffic sign, the point set of the traffic sign may be projected onto the image of the scene by using the conversion parameter, and the point of the traffic sign
- the position of the set projection in the image of the scene can be determined as the position of the traffic sign in the scene image from which the image data of the traffic sign can be obtained.
- the conversion parameters may include spatial position of the main beam, attitude parameters, and the like when performing laser scanning.
- extracting the sign information of the traffic sign from the image data may include performing image recognition on the image data of the traffic sign to obtain sign information of the traffic sign.
- the image data of the traffic sign may be corrected by using a point set corresponding to the traffic sign to eliminate the image data. Imaging tilt angle. Then, the image data of the corrected traffic sign is image-reviewed. By eliminating the imaging tilt angle of the image data, the recognition accuracy of the traffic sign can be improved.
- correcting the image data of the traffic sign by using a set of points corresponding to the traffic sign may include: determining a fitting plane equation of the traffic sign by using a set of points corresponding to the traffic sign And correcting the image data of the traffic sign by using the fitting plane equation to eliminate the imaging tilt of the image data.
- the image data of the obtained traffic sign is relatively close to the image taken from the front of the traffic sign, and the recognition accuracy of the traffic sign can be improved.
- the data processing center 42 may be disposed in a physical device that is different in location from the mobile collection device 41.
- data processing center 42 may be implemented by one or more fixed location computing devices.
- data processing center 42 may also be implemented with mobile acquisition device 41 by one or a set of devices disposed at the same physical location.
- the sign information acquisition method can be applied to a terminal.
- the terminal 1 may include an in-vehicle mobile measurement system 11, and the terminal may perform related operations using a car as a carrier.
- the onboard mobile measurement system 11 can include a laser scanning system 111 and an image imaging system 112 that is coupled to the image imaging system 112.
- the laser scanning system 11 can provide an operation of the laser scanning process, so that the terminal can perform laser scanning processing on the area containing the traffic sign to acquire a plurality of laser spots (also referred to as laser scanning points) of the traffic sign.
- the image imaging system 12 can perform imaging processing on the area of the laser scanning process when the laser scanning system 11 performs laser scanning processing to obtain an image of the area including the traffic sign.
- the terminal 1 may include not only an in-vehicle mobile measurement system, but also other systems and devices, such as a processor, etc., which are not specifically limited in this embodiment of the present application.
- the processor can be coupled to the in-vehicle mobile measurement system, and the processor can be a general purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more An integrated circuit for controlling execution of a program program of the present application, the processor may perform correlation on a plurality of laser spots of a traffic sign obtained by the terminal through laser scanning processing, and an image of a region including a traffic sign obtained by the image imaging system Processing and so on.
- CPU general purpose central processing unit
- ASIC application-specific integrated circuit
- the in-vehicle mobile measurement system can also include not only a laser scanning system and an image imaging system, but also other systems, such as an integrated navigation system.
- the integrated navigation system can be connected to the laser scanning system and the image imaging system respectively, and After the sign information of the traffic sign is obtained, the sign information of the traffic sign is added to the high-precision map, thereby realizing the automatic driving of the car, and the embodiment of the present application also does not specifically limit the same.
- the terminal may perform laser scanning processing on the area including the traffic sign by the above-mentioned laser scanning system, and may also be otherwise limited in the other embodiments. Similarly, the terminal may pass the above.
- the image imaging system obtains an image of the area containing the traffic sign, and the image of the area containing the traffic sign is obtained by other means.
- the embodiment of the present application also does not specifically limit the same.
- the terminal may acquire the sign information of the traffic sign by using the above-mentioned vehicle-mounted mobile measurement system to improve the accuracy, efficiency, and robustness of acquiring the sign information.
- FIG. 1B is a flowchart of a method for acquiring sign information provided by an embodiment of the present application. Referring to FIG. 1B, the method includes:
- Step 101 The terminal performs laser scanning processing on the area including the traffic sign to obtain an image of the traffic sign, and a plurality of laser points of the traffic sign and point cloud data of the plurality of laser points, points of each laser point Cloud data is used to describe the various laser points.
- the traffic sign Since the traffic sign has a strong reflection intensity to the light, there is a significant difference in the reflectance value compared with other objects after the laser scanning process. Therefore, the traffic sign can be utilized to include the traffic sign.
- the area performs laser scanning processing to obtain an image of the traffic sign, and a plurality of laser spots of the traffic sign and point cloud data of the plurality of laser points.
- the terminal may perform laser scanning processing on the area including the traffic sign, obtain an image of the traffic sign, and point cloud data of the plurality of area points and the plurality of area points, where the point cloud data of each area point is Include a reflectance value for each of the regional points; select a plurality of regional points from which the reflectance value is greater than or equal to a reflectance threshold; The plurality of regional points are determined as a plurality of laser points of the traffic sign, and the point cloud data of the selected plurality of regional points is determined as point cloud data of the plurality of laser points.
- the reflectance threshold refers to a criterion for screening a laser spot from a plurality of regional points, and the reflectance threshold may be set in advance.
- the reflectance threshold may be 60%, 70%, or the like. This example does not specifically limit this.
- the point cloud data may include not only reflectance values of the respective area points, but also other parameters, for example, the point cloud data may further include three-dimensional coordinates of the respective area points in the specified three-dimensional coordinate system, etc., The embodiment does not specifically limit this.
- the area where the terminal performs laser scanning processing may include a traffic sign, and may also include multiple traffic signs.
- the area includes multiple traffic signs, the multiple traffic signs
- the method for obtaining the sign information is the same as the method for obtaining the sign information of a traffic sign. The embodiment of the present application does not specifically limit this.
- the number of the obtained regional points is relatively large, and the traffic signboard accounts for a small proportion in the area, so the traffic is utilized.
- the strong reflection characteristics of the sign on the light and the reflectance threshold, the number of laser spots determined from the large number of regional points is small, thereby reducing the calculation amount and improving the efficiency of the subsequent terminal acquiring the sign information of the traffic sign.
- the laser since the laser is an active light source, when the laser scanning process is performed on the area including the traffic sign, it can be unaffected by the illumination, thereby improving the accuracy of the subsequent acquisition of the sign information.
- the terminal may include an in-vehicle mobile measurement system 11 , which in turn may include a laser scanning system 111 and an image imaging system 112 , so that the terminal can pass the laser scanning system 111
- the area including the traffic sign performs laser scanning processing, and when the terminal performs laser scanning processing, the image of the traffic sign is generated by the image forming system 112.
- Step 102 The terminal clicks on the plurality of lasers based on point cloud data of the multiple laser points. Row space clustering, the three-dimensional point cloud structure of the traffic sign is obtained.
- the terminal determines that the number of the plurality of laser points is m, assigns each laser point of the m laser points to 0, and then sequentially traverses each laser point; when determining that one of the laser points is assigned a value of 0 Establishing an initial three-dimensional point cloud structure, adding the laser point to a component constituting the initial three-dimensional point cloud structure, and assigning the laser point to a value of 1; determining the number of laser points constituting the current initial three-dimensional point cloud structure is Determining whether the number n of laser points included in the initial three-dimensional point cloud structure is less than or equal to the number of the m laser points after incrementing by one; when the number n of laser points included in the initial three-dimensional point cloud structure is incremented by one When the number of the m laser points is less than or equal to, the n-1th laser point constituting the initial point cloud structure is used as a reference point, and the distance between the m laser points and the n-1th laser point is determined to be smaller than a threshold laser point
- the distance threshold is a criterion for deleting the points constituting the initial three-dimensional point cloud structure, and the distance threshold may be set in advance.
- the distance threshold may be 10 cm, 20 cm, or the like. No specific restrictions.
- the three-dimensional point cloud structure is a three-dimensional structure of traffic signs composed of a plurality of laser points, and the three-dimensional point cloud structure can carry point cloud data of each laser point.
- the terminal may perform operations such as segmentation and extraction of multiple laser points by using the Euclidean distance clustering method, and may also be used in other methods.
- Step 103 The terminal determines, according to the three-dimensional point cloud structure, that the traffic sign is fitted flat. Surface equation.
- the terminal performs principal component analysis on the three-dimensional point cloud structure to obtain respective spatial feature values of the three-dimensional point cloud structure, and determines a feature vector of the smallest spatial feature value among the spatial feature values of the three-dimensional point cloud structure, and determines The three-dimensional coordinate of the three-dimensional point cloud structure is specified in the three-dimensional coordinate system; and the fitting plane equation of the traffic sign is determined based on the feature vector and the three-dimensional coordinates of the centroid of the three-dimensional point cloud structure in the specified three-dimensional coordinate system .
- the terminal performs principal component analysis on the three-dimensional point cloud structure, that is, the terminal can extract a laser point that can reflect the three-dimensional point cloud structure feature from the three-dimensional point cloud structure, and process the laser point to obtain a projection.
- the matrix is subjected to a dimensionality reduction process to obtain a spatial feature value, and each of the spatial feature values may include a first feature value, a second feature value, and a third feature value, which are not specifically limited in this embodiment of the present application. .
- the terminal can determine the feature vector of the smallest spatial feature value of each of the spatial feature values of the three-dimensional point cloud structure by using a principal component analysis method, and of course, it can be determined by other methods in practical applications. Make specific limits.
- the terminal analyzes the three-dimensional point cloud structure through the principal component analysis method, obtains the operation of each spatial feature value, and the smallest spatial feature among the spatial feature values of the three-dimensional point cloud structure by the principal component analysis method.
- the value is analyzed to obtain the operation of the feature vector of the minimum spatial feature value, which simplifies the calculation process, improves the terminal operation efficiency, and the efficiency of the terminal acquiring the feature vector of each spatial feature value and the minimum spatial feature value.
- the terminal may determine the centroid of the three-dimensional point cloud structure according to the structural shape of the three-dimensional point cloud structure and the mass of the plurality of laser points constituting the three-dimensional point cloud mechanism, and thereby determine the three-dimensional coordinates of the centroid in the specified three-dimensional coordinate system.
- the terminal can also determine the centroid of the three-dimensional point cloud structure and the three-dimensional coordinates of the centroid according to other methods, which is not specifically limited in this embodiment of the present application.
- the specified three-dimensional coordinate system may be a spatial rectangular coordinate system, a spherical coordinate system, or the like, which is not specifically limited in this embodiment of the present application.
- other forms of fitting plane equations may also be used. limited.
- the coefficients a, b, c, and d in the fitting plane equation are all determined by the terminal through the feature vector and the three-dimensional coordinates of the centroid.
- the terminal since the terminal performs laser scanning processing on the area including the traffic sign, the area may also include other interference objects that are also more reflective to light, such as glass billboards, etc.
- the interference structure may bring unnecessary operations to the terminal in subsequent processing, affecting the acquisition of the sign information. Therefore, the terminal determines the feature vector of the smallest spatial feature value among the spatial feature values of the three-dimensional point cloud structure, and determines that the centroid of the three-dimensional point cloud structure is in the three-dimensional coordinate of the specified three-dimensional coordinate system, Steps (1)-(3) determine whether the three-dimensional point cloud structure is an interference structure.
- the terminal may further determine, according to each spatial feature value of the three-dimensional point cloud structure, a first distribution value, a second distribution value, and a curvature value of the three-dimensional point cloud structure, the first distribution value and the second distribution value. It is used to describe the distribution shape of the three-dimensional point cloud structure in three-dimensional space, and the curvature value is used to describe the planarity of the three-dimensional point cloud structure in the three-dimensional space.
- the spatial feature values are sequentially sorted in descending order; the ratio between the largest spatial feature value and the intermediate spatial distribution value is determined, and the largest spatial feature value and the spatial feature value in the middle are sequentially selected.
- the ratio between the ratios is determined as the first distribution value; the ratio between the spatial eigenvalues in the middle and the smallest spatial eigenvalues is determined, and the ratio between the spatial eigenvalues in the middle and the smallest spatial eigenvalues is determined as a second distribution value; determining a ratio between the smallest spatial feature value and the sum of the individual spatial feature values, and determining a ratio between the smallest spatial feature value and the sum of the individual spatial feature values as the curvature value.
- the terminal sequentially sorts the spatial feature values in descending order.
- the sorting result is that the first feature value is greater than the second feature value, and the second feature value is greater than the third feature value, determining the first feature value and a ratio between the second eigenvalues, determining a ratio between the first eigenvalue and the second eigenvalue as the first distribution value; determining a ratio between the second eigenvalue and the third eigenvalue, the second feature
- the ratio between the value and the third eigenvalue is determined as the second distribution value; the ratio between the third eigenvalue and the sum of the individual spatial eigenvalues is determined, and the ratio between the third eigenvalue and the sum of the respective spatial eigenvalues is determined as Curvature value.
- determining the three-dimensional spatial distribution shape of the three-dimensional point cloud structure is a body-shaped distribution; when the first distribution value is smaller than a second distribution value, and when the difference between the first distribution value and the second distribution value is not in the difference range, determining that the three-dimensional point cloud structure has a planar distribution in a three-dimensional space; when the first distribution value When the difference between the first distribution value and the second distribution value is not located in the difference range, the three-dimensional spatial distribution shape of the three-dimensional point cloud structure is determined to be a linear distribution.
- difference range may be set in advance, and the difference range may be 1-10, 10-20, etc., which is not specifically limited in this embodiment of the present application.
- the curvature value when the curvature value is small, it is determined that the planarity of the three-dimensional point cloud structure in three-dimensional space is better, that is, when the curvature value is 0, the planarity of the three-dimensional point cloud structure in three-dimensional space is determined. the best.
- the terminal determines whether the three-dimensional point cloud structure is an interference structure based on the first distribution value, the second distribution value, and the curvature value of the three-dimensional point cloud structure.
- the terminal may be based on the first distribution value and the second distribution of the three-dimensional point cloud structure.
- the value and the curvature value determine whether the three-dimensional point cloud structure is an interference structure.
- the terminal is based on the first distribution value, the second distribution value, and the curve of the three-dimensional point cloud structure.
- the operation may be: determining whether the three-dimensional point cloud structure is an interference structure, determining: a length, a width, and a height of the three-dimensional point cloud structure; if the length of the three-dimensional point cloud structure is not within a specified length, the three-dimensional point cloud The width of the structure is not within the specified width, the height of the three-dimensional point cloud structure is not within the specified height range, the first distribution value is greater than the first distribution threshold, the second distribution value is less than the second distribution threshold, or the curvature value is greater than The curvature threshold determines that the three-dimensional point cloud structure is an interference structure; otherwise, determining that the three-dimensional point cloud structure is not an interference structure.
- the terminal may determine the length, width, and height of the three-dimensional point cloud structure according to the three-dimensional coordinates of the three-dimensional point cloud structure in the specified three-dimensional coordinate system.
- the terminal can be determined not only by the foregoing manner, but also by other methods.
- the specified length range refers to a criterion for determining whether the length of the three-dimensional point cloud structure meets the requirements, and the specified length range may be set in advance.
- the specified length may range from 20 cm to 50 cm, etc. This is not specifically limited.
- the specified width range is a standard for determining whether the width of the three-dimensional point cloud structure satisfies the requirements, and the specified width range may be set in advance.
- the specified width may range from 20 cm to 50 cm, etc. No specific restrictions.
- the specified height range is used to determine whether the height of the three-dimensional point cloud structure meets the required standard, and the specified height range can be set in advance.
- the specified height range may be 20 cm to 50 cm, etc. This is not specifically limited.
- the first distribution threshold is a criterion for determining whether the first distribution value meets the requirement, and the first distribution threshold may be set in advance.
- the first distribution value may be 2, 5, etc. No specific restrictions.
- the second distribution threshold is a criterion for determining whether the second distribution value meets the requirements, and the second distribution threshold may be set in advance.
- the second distribution value may be 50, 100, or the like. This is also not specifically limited.
- the curvature threshold is a criterion for determining whether the curvature value meets the requirements, and the curvature threshold may be set in advance.
- the curvature threshold may be 50%, 30%, etc., which is not specifically limited in the embodiment of the present application.
- the terminal executes the foregoing terminal to determine a feature vector of the smallest spatial feature value among the spatial feature values of the three-dimensional point cloud structure, and determines a centroid of the three-dimensional point cloud structure.
- the three-dimensional structure when the three-dimensional point cloud structure is not an interference structure, that is, when the three-dimensional point cloud structure meets the requirements of the traffic sign, the three-dimensional structure may be determined as a three-dimensional structure of the traffic sign, and then the terminal may be based on the The three-dimensional point cloud structure acquires the fitting plane equation of the traffic sign.
- the terminal may continue to perform the operations of step 104 described below.
- the interference structure does not meet the requirements of the traffic sign
- the three-dimensional point cloud structure is an interference structure
- the operation of other 3D point cloud structures are examples of other 3D point cloud structures.
- the terminal may include not only the in-vehicle mobile measurement system 11 but also a processor. Therefore, when the terminal acquires multiple laser points of the traffic sign through the laser scanning process, the terminal may pass through the processor. The laser spot of the traffic sign is subjected to the operations as described in steps 102 and 103 above.
- Step 104 The terminal acquires the sign information of the traffic sign based on the image of the traffic sign, the three-dimensional point cloud structure, and the fitting plane equation.
- the terminal may project the three-dimensional point cloud structure by using a preset outer orientation element.
- the outer orientation element includes a spatial position and a posture parameter of the main beam when the image is acquired; determining an image area occupied by the three-dimensional point cloud structure on the image;
- the coordinates of the pixel in the specified plane coordinate system are projected into the three-dimensional space by the fitting plane equation to obtain image data of the traffic sign;
- the image data of the traffic sign is corrected to eliminate the imaging tilt of the image data.
- performing image recognition on the image data of the corrected traffic sign to obtain the sign information of the traffic sign.
- the terminal may project the three-dimensional point cloud structure to a position corresponding to the three-dimensional point cloud structure in the traffic sign image by using a preset outer orientation element.
- the outer orientation element is basic data for determining the geometric relationship of the main beam of the photographing in the region containing the traffic sign, and is used to characterize the spatial position and spatial attitude of the main beam of the photographing at the photographing moment.
- the spatial position of the outer orientation element is mainly three linear elements for indicating the coordinate position of the main beam in the area containing the traffic sign, and the attitude parameter is mainly three corner elements for expressing the spatial posture in the image.
- the specified plane coordinate system may be a plane coordinate system established in an image of a traffic sign with any point as an origin, and the specified plane coordinate system may be a plane rectangular coordinate system, a polar coordinate system, or the like. This embodiment of the present application does not specifically limit this.
- the terminal since the terminal obtains the image data of the traffic sign, the image data of the traffic sign may have an imaging tilt angle during imaging, and image recognition of the image data having the imaging tilt angle may obtain an inaccurate flag.
- the card information therefore, the terminal needs to correct the image data of the traffic sign when obtaining the image data of the traffic sign.
- the operation of the terminal to correct the image data of the traffic sign may be: the terminal determines a tilt angle between the image data and a plane of any two coordinate axes in the specified three-dimensional coordinate system, and projects the image data along the tilt direction to the designated Any two coordinates in the three-dimensional coordinate system
- the plane in which the axis is located, or the plane parallel to the plane determines the projected projection data as the corrected influence data.
- the terminal can correct the image data of the traffic sign by the above method, and can also be corrected by other methods, which is not specifically limited in the embodiment of the present application.
- the terminal when it performs image recognition on the corrected traffic sign, it may adopt a recognition method based on color, shape, and sift (scale_invariant feature transform) feature, or based on HOG (Histogram of Oriented Gradient,
- HOG Histogram of Oriented Gradient
- the vehicle-mounted mobile measurement system 11 included in the terminal may include not only the laser scanning system 111 and the image imaging system 112, but also an integrated navigation system. Therefore, when the terminal acquires the sign of the traffic sign, When the information is used, the sign information can be added to the high-precision map through the integrated navigation system to realize the application of the sign information of the traffic sign.
- the terminal may perform laser scanning processing on the area including the traffic sign to obtain images of multiple area points and traffic signs. Since the traffic signs have strong reflection intensity to light, therefore, The terminal can use the feature to determine a plurality of laser points of the traffic sign and point cloud data of the plurality of laser points from the plurality of regional points.
- the laser is an active light source, when the laser scanning process is performed on the area including the traffic sign, the accuracy of the subsequent acquisition of the sign information can be improved without being affected by the illumination.
- the terminal may perform spatial clustering of the plurality of laser points according to the point cloud data of the plurality of laser points to obtain a three-dimensional point cloud structure of the traffic sign, and determine a spatial distribution characteristic of the three-dimensional point cloud structure, according to The spatial distribution feature determines whether the three-dimensional point cloud structure is an interference structure, thereby eliminating some interference structures and reducing the subsequent operational burden of the terminal.
- the fitting plane equation of the traffic sign can be determined, and the image data is image-recognized based on the image of the traffic sign, the three-dimensional point cloud structure and the fitting plane equation. Wait for the operation to obtain the sign information of the traffic sign.
- the point cloud data of the plurality of laser points can accurately express the coordinates of the three-dimensional point cloud structure in the specified three-dimensional coordinate system, a series of operations performed by the terminal based on the accurate coordinates are accurate operations, and the acquisition of the signboard is reduced. The error of the information, thereby improving the accuracy, efficiency and robustness of the sign information of the traffic sign.
- FIG. 2A is a schematic structural diagram of a sign information acquiring apparatus according to an embodiment of the present application.
- the apparatus includes: a scanning module 201, a clustering module 202, a determining module 203, and an obtaining module 204.
- the scanning module 201 is configured to perform laser scanning processing on the area including the traffic sign, obtain an image of the traffic sign, and a plurality of laser points of the traffic sign and point cloud data of the plurality of laser points, each laser point Point cloud data is used to describe the individual laser points;
- the clustering module 202 is configured to spatially cluster the plurality of laser points based on the point cloud data of the plurality of laser points to obtain a three-dimensional point cloud structure of the traffic sign;
- a determining module 203 configured to determine a fitting plane equation of the traffic sign based on the three-dimensional point cloud structure
- the obtaining module 204 is configured to obtain the sign information of the traffic sign based on the image of the traffic sign, the three-dimensional point cloud structure, and the fitting plane equation.
- the scanning module 201 includes:
- the scanning unit 2011 is configured to perform laser scanning processing on the area including the traffic sign, obtain an image of the traffic sign, and point cloud data of the plurality of area points and the plurality of area points, and point cloud data of each area point The reflectance values of the points of the respective regions are included;
- the selecting unit 2022 is configured to select, from the plurality of regional points, a reflectance value greater than or equal to the inverse a plurality of regional points of the radiance threshold;
- the first determining unit 2033 is configured to determine the selected plurality of regional points as the plurality of laser points of the traffic sign, and determine the point cloud data of the selected plurality of regional points as the point cloud of the multiple laser points. data.
- the determining module 203 includes:
- the analyzing unit 2031 is configured to perform principal component analysis on the three-dimensional point cloud structure to obtain respective spatial feature values of the three-dimensional point cloud structure;
- a second determining unit 2032 configured to determine a feature vector of the smallest spatial feature value among the spatial feature values of the three-dimensional point cloud structure, and determine a three-dimensional coordinate of the centroid of the three-dimensional point cloud structure in the specified three-dimensional coordinate system;
- the third determining unit 2033 is configured to determine a fitting plane equation of the traffic sign based on the feature vector and the three-dimensional coordinates of the centroid of the three-dimensional point cloud structure in the specified three-dimensional coordinate system.
- the determining module 203 further includes:
- a fourth determining unit 2034 configured to determine, according to each spatial feature value of the three-dimensional point cloud structure, a first distribution value, a second distribution value, and a curvature value of the three-dimensional point cloud structure, the first distribution value and the second distribution
- the value is used to describe a distribution shape of the three-dimensional point cloud structure in a three-dimensional space
- the curvature value is used to describe the planarity of the three-dimensional point cloud structure in the three-dimensional space
- the determining unit 2035 determines whether the three-dimensional point cloud structure is an interference structure based on the first distribution value, the second distribution value, and the curvature value of the three-dimensional point cloud structure;
- the triggering unit 2036 is configured to: when the three-dimensional point cloud structure is not an interference structure, trigger the second determining unit to determine a feature vector of a smallest spatial feature value among the spatial feature values of the three-dimensional point cloud structure, and determine the three-dimensional point The centroid of the cloud structure is in three-dimensional coordinates in a specified three-dimensional coordinate system.
- the determining unit 2035 includes:
- a first determining subunit 20351 configured to determine a length, a width, and a height of the three-dimensional point cloud structure
- a second determining subunit 20352 configured to: if the length of the three-dimensional point cloud structure is not within a specified length range, the width of the three-dimensional point cloud structure is not within a specified width, and the height of the three-dimensional point cloud structure is not within a specified height range If the first distribution value is greater than the first distribution threshold, the second distribution value is smaller than the second distribution threshold, or the curvature value is greater than the curvature threshold, determining the three-dimensional point cloud structure as an interference structure, otherwise determining the three-dimensional point cloud structure Not to interfere with the structure.
- the obtaining module 204 includes:
- the first projection unit 2041 is configured to project the three-dimensional point cloud structure onto the image of the traffic sign by a preset outer orientation element, where the outer orientation element includes a spatial position and a posture parameter of the main beam when the image is acquired;
- a fifth determining unit 2042 configured to determine an image area occupied by the three-dimensional point cloud structure on the image
- a second projection unit 2043 configured to project coordinates of each pixel in the specified plane coordinate system into the three-dimensional space by using the fitting plane equation to obtain image data of the traffic sign;
- the correcting unit 2044 is configured to correct the image data of the traffic sign to eliminate the imaging tilt of the image data
- the identification unit 2045 is configured to perform image recognition on the corrected image data of the traffic sign to obtain the sign information of the traffic sign.
- the terminal may perform laser scanning processing on the area including the traffic sign to obtain images of multiple area points and traffic signs. Since the traffic signs have strong reflection intensity to light, therefore, The terminal can use the feature to determine a plurality of laser points of the traffic sign and point cloud data of the plurality of laser points from the plurality of regional points. In addition, since the laser is an active light source, a laser sweep is performed on the area containing the traffic sign. When the drawing process is performed, it is possible to improve the accuracy of the subsequent acquisition of the sign information without being affected by the illumination.
- the terminal may perform spatial clustering of the plurality of laser points according to the point cloud data of the plurality of laser points to obtain a three-dimensional point cloud structure of the traffic sign, and determine a spatial distribution characteristic of the three-dimensional point cloud structure, according to The spatial distribution feature determines whether the three-dimensional point cloud structure is an interference structure, thereby eliminating some interference structures and reducing the subsequent operational burden of the terminal.
- the fitting plane equation of the traffic sign can be determined, and the image data is performed based on the image of the traffic sign, the three-dimensional point cloud structure, and the fitting plane equation. An operation such as image recognition to obtain the sign information of the traffic sign.
- the point cloud data of the plurality of laser points can accurately express the coordinates of the three-dimensional point cloud structure in the specified three-dimensional coordinate system, a series of operations performed by the terminal based on the accurate coordinates are accurate operations, and the acquisition of the signboard is reduced. The error of the information, thereby improving the accuracy, efficiency and robustness of the sign information of the traffic sign.
- the sign information obtaining device provided by the above embodiment only uses the division of the above functional modules when obtaining the sign information, and in actual applications, the functions may be assigned different functions according to needs.
- the module is completed, dividing the internal structure of the device into different functional modules to perform all or part of the functions described above.
- the embodiment of the embodiment of the embodiment of the present invention is not described herein.
- FIG. 3 is a schematic structural diagram of a terminal of a sign information acquiring apparatus according to an embodiment of the present application.
- the terminal 300 can include a communication unit 310, a memory 320 including one or more computer readable storage media, an input unit 330, a display unit 340, a sensor 350, an audio circuit 360, and a WIFI (Wireless Fidelity).
- the module 370 includes a processor 380 having one or more processing cores, and a power supply 390 and the like. It will be understood by those skilled in the art that the terminal structure shown in FIG. 3 does not constitute a limitation of the terminal, and may include more or less components than those illustrated, or combine some components, or Different parts are arranged.
- the terminal further includes one or more programs, the one or more programs being stored in the memory and configured to be executed by one or more processors, the one or more programs including The instruction of the sign information obtaining method shown in FIG. 1B provided by the embodiment of the present application is performed.
- a person skilled in the art may understand that all or part of the steps of implementing the above embodiments may be completed by hardware, or may be instructed by a program to execute related hardware, and the program may be stored in a computer readable storage medium.
- the storage medium mentioned may be a read only memory, a magnetic disk or an optical disk or the like.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Optics & Photonics (AREA)
- Traffic Control Systems (AREA)
- Image Processing (AREA)
Abstract
一种交通标志牌信息获取方法及装置,数据处理设备处获取摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像;获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激光数据包括多个激光点的数据(501);对所述多个激光点进行空间聚类,得到至少两个候选点集合(502);获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合(503);利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息(504)。
Description
本申请要求于2016年09月13日提交中国专利局、申请号为201610822521.4、名称为“标志牌信息获取方法及装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本申请涉及测绘技术领域,特别涉及一种交通标志牌信息获取方法及装置。
背景
随着技术的发展,自动驾驶技术作为人工智能的一种典型模式,越来越受到人们的关注。其中,实现自动驾驶技术中的关键技术之一就是高精度地图。交通标志牌的标识牌信息是高精度地图中的重要组成部分之一。目前,获取交通标志牌的标志牌信息一般是利用图像采集设备对包含交通标志牌的区域进行图像采集,对采集的图像进行图像识别来检测其中的交通标志牌。
技术内容
为了解决相关技术的问题,本申请实施例提供了一种交通标志牌信息获取方法及装置。
本申请实施例的一种交通标志牌信息获取方法可以包括:
在数据处理设备处,
获取摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像;
获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激
光数据包括多个激光点的数据;
对所述多个激光点进行空间聚类,得到至少两个候选点集合;
获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合;
利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息。
本申请实施例的一种交通标志牌信息获取装置可以包括处理器和存储器,所述存储器中存储有计算机可读指令,所述指令可以使处理器执行以下操作:
获取摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像;
获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激光数据包括多个激光点的数据;
对所述多个激光点进行空间聚类,得到至少两个候选点集合;
获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合;
利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息。
本申请实施例的一种计算机可读存储介质存储有计算机可读指令,所述指令可以使处理器执行以下操作:
获取摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像;
获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激
光数据包括多个激光点的数据;
对所述多个激光点进行空间聚类,得到至少两个候选点集合;
获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合;
利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息。
根据本申请实施例,由于激光是主动式光源,因此,在对包含该交通标志牌的区域进行激光扫描处理时,可以不受光照影响,提高后续获取标志牌信息的准确度。可以根据该多个激光点的数据,将该多个激光点进行进空间聚类得到该交通标志牌的点集合,并基于该点集合确定该交通标志牌的图像数据,对该图像数据进行图像识别等操作,以获取该交通标志牌的标志牌信息。由于激光点数据包括各激光点的坐标,基于该坐标进行的一系列操作可以减少了获取标志牌信息的误差,从而提高了获取交通标志牌的标志牌信息的准确度、效率以及鲁棒性。
附图简要说明
为了更清楚地说明本申请实施例中的技术方案,下面将对实施例描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1A是本申请实施例提供的一种车载移动测量***的示意图。
图1B是本申请实施例提供的一种标志牌信息获取方法的流程图。
图2A是本申请实施例提供的一种标志牌信息获取装置的结构示意图。
图2B是本申请实施例提供的一种扫描模块的结构示意图。
图2C是本申请实施例提供的一种确定模块的结构示意图。
图2D是本申请实施例提供的另一种确定模块的结构示意图。
图2E是本申请实施例提供的一种判断单元的结构示意图。
图2F是本申请实施例提供的一种获取模块的结构示意图。
图3是本申请实施例提供的一种标志牌信息获取装置的结构示意图。
图4是本申请实施例提供的一种信息处理***的示意图。
图5是本申请实施例提供的一种交通标志牌信息获取方法的流程图。
实施方式
为使本申请的目的、技术方案和优点更加清楚,下面将结合附图对本申请实施方式作进一步地详细描述。
高精度地图是自动驾驶技术的重要辅助,通过高精度地图可以告知汽车前方行走方向和路况。交通标志牌作为高精度地图的重要组成部分。为了获取交通标志牌的标志牌信息,本申请实施例提供了的一种交通标志牌信息获取方法,该方法可以由一个信息处理***实现。如图4所示,该信息处理***40可以包括移动采集设备41和数据处理中心42。
移动采集设备41为可移动的、用于采集交通标志牌信息的设备。一些例子中,移动采集设备41可以通过装载在可移动的人或物(例如交通工具,等)上实现移动,例如可以为手持设备、车载设备,等。移动采集设备41可以包括激光扫描***411与影像成像***412。
激光扫描***411用于对场景进行激光扫描,得到场景的激光数据。一些例子中,激光扫描***411可以包括激光发射装置和接收检测装置。
激光发射装置用于向场景发射激光;接收检测装置用于检测场景反射回的激光,生成激光数据。一些例子中,激光扫描***411可以包括二维或三维激光扫描设备,用于对场景进行激光扫描,得到激光数据。激光数据可以体现为多个激光点(laser point)的数据的集合,或者激光点云(point cloud),或其它形式。激光点云是激光扫描得到的大量激光数据点的集合。激光数据可以包括各个激光点的位置(二维或三维坐标)以及反射特性值。
影像成像***412包括摄像装置,用于对场景进行拍摄得到场景的图像。摄像装置可以包括镜头和图像传感器,例如CCD、CMOS等。图4中仅示出一个移动采集设备41作为例子来说明本申请技术方案的实现方法,一些实施例中,信息处理***40可以包括多个移动采集设备41。
数据处理中心42可以包括一个或多个数据处理设备,因此,数据处理中心42可以包括一个或多个处理器421。数据处理中心42还可以包括存储器422和通信接口423。数据处理中心42可以通过通信接口423与移动采集设备41建立某种连接以进行通信。这种连接可以是有线连接(如使用某种缆线或插口建立的连接),也可以是无线连接(如Wi-Fi、蜂窝移动通信、射频、蓝牙、红外等连接);可以是直接连接,也可以是通过某种网络建立的连接,等。数据处理中心42可以通过通信接口423获取移动采集设备41采集的场景的图像和激光数据,对这些数据进行处理以提取出该场景中的交通标志牌的信息。一些实施例中,存储器422中可以存储有信息提取模块425。信息提取模块425可以以计算机可读指令的形式存储在存储器422中,并且可被处理器421执行,以实现本申请各实施例的交通标志牌信息获取方法。图5为本申请实施例的一种交通标志牌信息获取方法的流程图。该方法50可以由数据处理中心42执行。如图5所示,该方法50可以包括以下步骤。
步骤501,获取使用摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像,并获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激光数据包括多个激光点的数据。
步骤502,对激光数据中的激光点进行空间聚类,得到至少两个候选点集合。
这里,空间聚类是指,根据激光数据中各激光点的位置数据,将激光点根据空间位置进行分组,每组作为一个候选点集合。例如,将相互距离较近的激光点分为一组,距离远的激光点分到不同的组中。这样,每个候选点集合中的激光点可能是同一物体的表面反射的激光形成的。每个候选点集合可以体现为一组点的数据集合,也可以体现为点云。
步骤503,获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合。
步骤504,利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息。
这里,包括交通标志牌的场景是指设置有交通标志牌的物理环境(例如,道路两旁、道路上方,等),也称为包括交通标志牌的区域。
这样,利用交通标志牌具有对激光的高反射率的特点,通过使用交通标志牌的激光数据来对场景图像中的交通标志牌进行定位,然后仅对场景图像中交通标志牌所在的位置进行交通标志牌信息提取,可以提高交通标志牌的识别准确度。
一些实施例中,获取所述场景的激光数据可以包括:获取激光扫描得到的多个第二激光点以及各个第二激光点的反射特性值;利用所述第二激光点中选择反射特性值大于或等于预设阈值的多个激光点形成所述激光数据。这样,利用交通标志牌的高反光性,从得到的激光点中选
取反射特性较好的点形成激光数据,用于后续计算,可以显著减小激光数据的数据量,提高计算效率。反射特性值可以是反射强度值,反射率值,等。
一些实施例中,对多个激光点进行空间聚类得到至少两个候选点集合可以包括:利用预设的距离阈值对所述激光数据中的所述多个激光点进行分组以形成所述至少两个候选点集合,其中不同候选点集合中的激光点之间的空间距离大于所述距离阈值。激光点之间的空间距离可以采用欧式距离或类似的算法得到。这样,通过聚类,可以将激光点分为多个候选点集合,每个候选点集合可以认为是来自一个物体的激光点的集合。后续,可以仅对每个候选点集合进行处理,高效地找出交通标志牌对应的点集合。
一些实施例中,获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合可以包括:获取所述至少两个候选点集合中各候选点集合中激光点在空间至少两个维度上的空间分布特征值;利用至少两个预设的空间分布特征阈值和各候选点集合的所述空间分布特征值从所述至少两个候选点集合中选择至少一个候选点集合作为所述交通标志牌对应的点集合。上述空间分布特征值可以选自:长度上的分布特征值、宽度上的分布特征值、高度上的分布特征值、线状分布特征值、面状分布特征值、体状分布特征值、曲率、拟合平面特征值。上述预设的空间分布特征阈值可以根据交通标志牌的尺寸范围、形状特征确定。这样,就可以从多个候选点集合中确定出交通标志牌对应的点集合。
一些实施例中,利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据可以包括:根据预设的参数和所述交通标志牌对应的点集合在所述场景的激光点集合中的位置确定所述交通
标志牌的图像数据在所述场景的图像中的位置,从所述图像中的所述位置提取所述交通标志牌的图像数据。通常,激光扫描装置的位置和摄像装置的位置和姿态是相对固定的,它们得到的数据的坐标系之间的转换关系称为转换参数(也称外方位元素),利用转换参数(或参数矩阵)可以将激光点集合准确投影到图像上。一些例子中,在根据交通标志牌的点云确定交通标志牌的图像数据在场景图像中的位置时,可以利用转换参数将交通标志牌的点集合投影到场景的图像上,交通标志牌的点集合投射在场景的图像中的位置可以确定为交通标志牌在场景图像中的位置,即可从该位置上获取交通标志牌的图像数据。转换参数可以包括进行激光扫描时,主光束的空间位置、姿态参数,等。
一些实施例中,从所述图像数据中提取出所述交通标志牌的标志牌信息可以包括:对所述交通标志牌的图像数据进行图像识别,以得到所述交通标志牌的标志牌信息。
一些实施例中,在对所述交通标志牌的图像数据进行图像识别之前,还可以利用所述交通标志牌对应的点集合对所述交通标志牌的图像数据进行矫正,以消除所述影像数据的成像倾角。然后,再对矫正后的所述交通标志牌的图像数据进行图像识别。通过消除所述影像数据的成像倾角,可以提高对交通标志牌的识别准确率。
一些实施例中,利用所述交通标志牌对应的点集合对所述交通标志牌的图像数据进行矫正可以包括:利用所述交通标志牌对应的点集合确定所述交通标志牌的拟合平面方程;利用该拟合平面方程对所述交通标志牌的图像数据进行矫正,以消除该图像数据的成像倾角。通过矫正,得到的交通标志牌的图像数据比较接近从交通标志牌的正面拍摄的图像,可以提高对交通标志牌的识别准确率。
本申请的一些实施例中,数据处理中心42可以设置在与移动采集设备41位置不同的物理设备中。例如,数据处理中心42可以由一台或多台位置固定的计算设备实现。
本申请的另一些实施例中,数据处理中心42也可以与移动采集设备41由设置在相同物理位置上的一个或一套设备实现。例如,标志牌信息获取方法可以应用到一种终端中。如图1A所示,该终端1可以包括车载移动测量***11,且该终端可以以汽车为载体进行相关操作。
其中,该车载移动测量***11可以包括激光扫描***111和影像成像***112,该激光扫描***111与该影像成像***112连接。该激光扫描***11可以提供激光扫描处理的操作,使终端可以对包含交通标志牌的区域进行激光扫描处理,以获取交通标志牌的多个激光点(也称为激光扫描点)。该影像成像***12可以在激光扫描***11进行激光扫描处理时,对激光扫描处理的区域进行成像处理,得到该包含交通标志牌的区域的图像。
另外,该终端1中不仅可以包括车载移动测量***,还可以包括其他***和器件,比如,处理器等,本申请实施例对此不做具体限定。该处理器可以与该车载移动测量***连接,且该处理器可以是一个通用中央处理器(CPU),微处理器,特定应用集成电路(application-specific integrated circuit,ASIC),或一个或多个用于控制本申请方案程序执行的集成电路,该处理器可以对终端通过激光扫描处理获取的交通标志牌的多个激光点,以及通过影像成像***获取的包含交通标志牌的区域的图像进行相关处理等。
一些实施例中,该车载移动测量***中同样不仅可以包括激光扫描***和影像成像***,还可以包括其他***,比如,组合导航***等,
该组合导航***可以分别与激光扫描***和影像成像***连接,并在获
取到该交通标志牌的标志牌信息后,将该交通标志牌的标志牌信息加入到高精度地图中,从而实现汽车的自动化驾驶,本申请实施例对此同样不做具体限定。
还需要说明的是,终端可以通过上述激光扫描***对该包含交通标志牌的区域进行激光扫描处理,也可以通过其他方式,本申请实施例对此不做具体限定,同理,终端可以通过上述影像成像***得到该包含交通标志牌的区域的图像,也可以通过其他方式得到该包含交通标志牌的区域的图像,本申请实施例对此同样不做具体限定。
在本申请实施例中,终端可以利用上述车载移动测量***获取该交通标志牌的标志牌信息,以提高获取标志牌信息的准确度、效率和鲁棒性。
图1B是本申请实施例提供的一种标志牌信息获取方法的流程图。参见图1B,该方法包括:
步骤101:终端对包含交通标志牌的区域进行激光扫描处理,得到该交通标志牌的图像,以及该交通标志牌的多个激光点和该多个激光点的点云数据,各个激光点的点云数据用于对该各个激光点进行描述。
由于交通标志牌对光线的反射强度较强,通过激光扫描处理后与其他物体相比,在反射率值上存在明显的差异,因此,可以利用交通标志牌这一特性,对包含交通标志牌的区域进行激光扫描处理,得到该交通标志牌的图像,以及该交通标志牌的多个激光点和该多个激光点的点云数据。
具体地,终端可以对包含该交通标志牌的区域进行激光扫描处理,得到该交通标志牌的图像,以及多个区域点和该多个区域点的点云数据,各个区域点的点云数据中包括该各个区域点的反射率值;从该多个区域点中选择反射率值大于或等于反射率阈值的多个区域点;将选择的
多个区域点确定为该交通标志牌的多个激光点,并将选择的多个区域点的点云数据确定为该多个激光点的点云数据。
需要说明的是,反射率阈值是指从多个区域点中筛选出激光点的标准,且该反射率阈值可以事先设置,比如,该反射率阈值可以为60%、70%等,本申请实施例对此不做具体限定。
另外,点云数据中不仅可以包括各个区域点的反射率值,还可以包括其他参数,比如,该点云数据中还可以包括该各个区域点在指定三维坐标系中的三维坐标等,本申请实施例对此不做具体限定。
在本申请实施例中,该终端进行激光扫描处理的区域中可能包括一个交通标志牌,也可能包括多个交通标志牌,当该区域中包括多个交通标志牌时,该多个交通标志牌的标志牌信息的获取方法与一个交通标志牌的标志牌信息的获取方法相同,本申请实施例对此不做具体限定。
其中,在本申请实施例中,由于对包含该交通标志牌的区域进行激光扫描处理后,得到的区域点的数量较为庞大,而交通标志牌在区域中所占比例很小,因此,利用交通标志牌对光线反射较强的特性以及反射率阈值,从该大量的区域点中确定的激光点的数量较少,从而减少了计算量,提高了后续终端获取交通标志牌的标志牌信息的效率。同时,由于激光是一个主动式光源,在对包含该交通标志牌的区域进行激光扫描处理时,可以不受光照影响,从而提高了后续获取标志牌信息的准确度。
值得说明的是,基于上述描述,终端中可以包括车载移动测量***11,该车载移动测量***11又可以包括激光扫描***111和影像成像***112,因此,终端可以通过该激光扫描***111对该包含交通标志牌的区域进行激光扫描处理,并当终端进行激光扫描处理时,通过影像成像***112生成该交通标志牌的图像。
步骤102:终端基于该多个激光点的点云数据,对该多个激光点进
行空间聚类,得到该交通标志牌的三维点云结构。
具体地,该终端确定该多个激光点的数量为m,将m个激光点中的每个激光点赋值为0,之后依次遍历每个激光点;当确定其中一个激光点的赋值为0时,建立一个初始三维点云结构,将该激光点加入到构成该初始三维点云结构的组成部分中,并将该激光点赋值为1;确定构成当前初始三维点云结构中的激光点数量为n;判断该初始三维点云结构包括的激光点的数量n在递增1后是否小于或等于该m个激光点的数量;当该初始三维点云结构包括的激光点的数量n在递增1后小于或等于该m个激光点的数量时,将构成该初始点云结构的第n-1个激光点作为参考点,确定m个激光点中与第n-1个激光点之间的距离小于距离阈值的激光点,并将确定的激光点加入到该初始三维点云结构中;之后从判断该初始三维点云结构包括的激光点的数量在递增1后是否小于或等于该m个激光点的数量的操作开始进行重复操作,直至该初始三维点云结构包括的激光点的数量在递增1后大于该m个激光点的数量,并确定最终的三维点云结构。
需要说明的是,距离阈值是指对构成初始三维点云结构的点进行删选的标准,且该距离阈值可以事先设置,比如,该距离阈值可以为10cm、20cm等,本申请实施例对此不做具体限定。
另外,三维点云结构为多个激光点构成的交通标志牌的一个三维结构,该三维点云结构中可以携带每个激光点的点云数据。
需要说明的是,上述对该多个激光点进行空间聚类的操作是以欧式距离聚类方法为例进行说明,并不对本申请实施例构成限定。在本申请实施例中,终端可以通过欧式距离聚类方法对多个激光点进行分割和提取等操作,也可以通过其他方法,本申请实施例对此不做具体限定。
步骤103:终端基于该三维点云结构,确定该交通标志牌的拟合平
面方程。
具体地,终端对该三维点云结构进行主成分分析,得到该三维点云结构的各个空间特征值;确定该三维点云结构的各个空间特征值中最小的空间特征值的特征向量,并确定该三维点云结构的质心在指定三维坐标系中的三维坐标;基于该特征向量和该三维点云结构的质心在该指定三维坐标系中的三维坐标,确定该交通标志牌的拟合平面方程。
需要说明的是,终端通过对该三维点云结构进行主成分分析,也即是,终端可以从三维点云结构中提取可以体现三维点云结构特征的激光点,对该激光点进行处理得到投影矩阵,对该投影矩阵进行降维处理,从而得到各个空间特征值,该各个空间特征值可以包括第一特征值、第二特征值和第三特征值,本申请实施例对此不做具体限定。
其中,终端可以通过主成分分析方法确定该三维点云结构的各个空间特征值中最小的空间特征值的特征向量,当然,在实际应用中还可以通过其他方法确定,本申请实施例对此不做具体限定。
需要说明的是,终端通过主成分分析方法对该三维点云结构进行分析,得到各个空间特征值的操作,以及通过主成分分析方法对该三维点云结构的各个空间特征值中最小的空间特征值进行分析,得到该最小的空间特征值的特征向量的操作,简化了计算过程,提高了终端运行效率,以及终端获取各个空间特征值和该最小的空间特征值的特征向量的效率。
另外,终端可以根据三维点云结构的结构形状和构成三维点云机构的多个激光点的质量,确定该三维点云结构的质心,进而可以在指定三维坐标系中确定该质心的三维坐标。当然终端也可以根据其他方式确定该三维点云结构的质心以及该质心的三维坐标,本申请实施例对此不做具体限定。
还需要说明的是,该指定三维坐标系可以为空间直角坐标系、球坐标系等,本申请实施例对此不做具体限定。
再者,终端确定的拟合平面方程可以为a*x+b*y+c*z+d=0,当然,也可以为其他形式的拟合平面方程,本申请实施例对此不做具体限定。
需要说明的是,该拟合平面方程中的系数a、b、c和d均由终端通过该特征向量和该质心的三维坐标确定。
进一步地,由于终端在对包含交通标志牌的区域进行激光扫描处理时,该区域中可能还包括同样对光线反射能力较强的其他干扰物体,比如,玻璃材质的广告牌等,这些干扰物体的干扰结构在后续处理中可能给终端带来多余的操作,影响对标志牌信息的获取。因此,终端在确定该三维点云结构的各个空间特征值中最小的空间特征值的特征向量,并确定该三维点云结构的质心在指定三维坐标系中的三维坐标之前,还可以通过下述步骤(1)-(3)来对该三维点云结构是否为干扰结构进行判断。
(1)、终端还可以基于该三维点云结构的各个空间特征值,确定该三维点云结构的第一分布值、第二分布值和曲率值,该第一分布值和该第二分布值用于描述该三维点云结构在三维空间中的分布形状,该曲率值用于描述该三维点云结构在该三维空间中的平面性。
具体地将各个空间特征值按照从大到小的顺序依次进行排序;确定最大的空间特征值与中间的空间分布值之间的比值,将最大的空间特征值与顺序位于中间的空间特征值之间的比值确定为第一分布值;确定该顺序位于中间的空间特征值与最小的空间特征值之间的比值,将顺序位于中间的空间特征值与最小的空间特征值之间的比值确定为第二分布值;确定最小的空间特征值与各个空间特征值总和之间的比值,将最小的空间特征值与各个空间特征值总和之间的比值确定为曲率值。
比如,终端将各个空间特征值按照从大到小的顺序依次进行排序,当排序结果为第一特征值大于第二特征值,第二特征值大于第三特征值时,确定第一特征值与第二特征值之间的比值,将第一特征值与第二特征值之间的比值确定为该第一分布值;确定第二特征值与第三特征值之间的比值,将第二特征值与第三特征值之间的比值确定为第二分布值;确定第三特征值与各个空间特征值总和之间的比值,将第三特征值与各个空间特征值总和之间的比值确定为曲率值。
需要说明的是,当第一分布值与第二分布值之间的差值位于差值范围内时,确定该三维点云结构的在三维空间分布形状为体状分布;当第一分布值小于第二分布值,且该第一分布值与第二分布值之间的差值不位于差值范围时,确定该三维点云结构的在三维空间分布形状为面状分布;当第一分布值大于第二分布值,且第一分布值与第二分布值之间的差值不位于差值范围时,确定该三维点云结构的在三维空间分布形状为线状分布。
还需要说明的是,差值范围可以事先设置,该差值范围可以为1-10、10-20等,本申请实施例对此不做具体限定。
另外,当曲率值越小时,确定该三维点云结构的在三维空间中的平面性越好,也即是,当曲率值为0时,确定该三维点云结构的在三维空间中的平面性最好。
(2)、终端基于该三维点云结构的第一分布值、第二分布值和曲率值,判断该三维点云结构是否为干扰结构。
其中,由于交通标志牌具有一定的尺寸范围,且交通标志牌呈面状分布,同时交通标志牌的平面性较好,因此,终端可以基于该三维点云结构的第一分布值、第二分布值和曲率值,判断该三维点云结构是否为干扰结构。而终端基于该三维点云结构的第一分布值、第二分布值和曲
率值,判断该三维点云结构是否为干扰结构的操作可以为:确定该三维点云结构的长度、宽度和高度;如果该三维点云结构的长度不位于指定长度范围内、该三维点云结构的宽度不位于指定宽度范围内、该三维点云结构的高度不位于指定高度范围内、该第一分布值大于第一分布阈值、该第二分布值小于第二分布阈值或该曲率值大于曲率阈值,则确定该三维点云结构为干扰结构,否则,确定该三维点云结构不为干扰结构。
需要说明的是,终端可以根据三维点云结构在指定三维坐标系中的三维坐标,确定该三维点云结构的长度、宽度和高度。当然,该终端不仅可以通过上述方式进行确定,还可以通过其他方式进行确定,本申请实施例对此不做具体限定。
还需要说明的是,指定长度范围是指衡量三维点云结构的长度是否满足要求的标准,且该指定长度范围可以事先设置,比如,该指定长度范围可以为20cm-50cm等,本申请实施例对此不做具体限定。
同理,指定宽度范围是指衡量三维点云结构的宽度是否满足要求的标准,且该指定宽度范围可以事先设置,比如,该指定宽度范围可以为20cm-50cm等,本申请实施例同样对此不做具体限定。
同理,指定高度范围和是指衡量三维点云结构的高度是否满足要求的标准,且该指定高度范围可以事先设置,比如,该指定高度范围可以为20cm-50cm等,本申请实施例同样对此不做具体限定。
另外,该第一分布阈值是指衡量第一分布值是否满足要求的标准,且该第一分布阈值可以事先设置,比如,该第一分布值可以为2、5等,本申请实施例对此不做具体限定。
同理,该第二分布阈值是指衡量第二分布值是否满足要求的标准,且该第二分布阈值可以事先设置,比如,该第二分布值可以为50、100等,本申请实施例对此同样不做具体限定。
再者,该曲率阈值是指衡量曲率值是否满足要求的标准,且该曲率阈值可以事先设置,比如,该曲率阈值可以为50%、30%等,本申请实施例对此不做具体限定。
(3)、当该三维点云结构不为干扰结构时,终端执行上述终端确定该三维点云结构的各个空间特征值中最小的空间特征值的特征向量,并确定该三维点云结构的质心在指定三维坐标系中的三维坐标的步骤。
其中,当该三维点云结构不为干扰结构时,也即是,当该三维点云结构符合交通标志牌的要求时,可以确定该三维结构为交通标志牌的三维结构,继而终端可以基于该三维点云结构获取该交通标志牌的拟合平面方程。
需要说明的是,当终端在获取到该交通标志牌的拟合平面方程之后,终端可以继续执行下述步骤104的操作。
另外,由于干扰结构不符合交通标志牌的要求,因此,当该三维点云结构为干扰结构时,可以确定该干扰结构不为交通标志牌,此时可以停止对该三维点云结构的操作,无需获取该干扰结构的拟合平面方程,继而无需再进行下述步骤104中获取交通标志牌的标志牌信息的操作,或者,终端还可以将该三维点云结构进行剔除,避免干扰结构影响到对其他三维点云结构的操作。
值得说明的是,基于上述描述,终端中不仅可以包括车载移动测量***11,还可以包括处理器,因此,当终端通过激光扫描处理获取到交通标志牌的多个激光点时,可以通过处理器对该交通标志牌的激光点进行如上述步骤102和步骤103所述的操作。
步骤104:终端基于该交通标志牌的图像、三维点云结构和拟合平面方程,获取该交通标志牌的标志牌信息。
具体地,终端可以通过预设的外方位元素,将该三维点云结构投影
到该交通标志牌的图像上,该外方位元素包括获取该图像时主光束的空间位置和姿态参数;确定该三维点云结构在该图像上所占的图像区域;将该图像区域内每个像素点在指定平面坐标系中的坐标通过该拟合平面方程投影到三维空间中,得到该交通标志牌的影像数据;对该交通标志牌的影像数据进行矫正,以消除该影像数据的成像倾角;对矫正后的交通标志牌的影像数据进行图像识别,得到该交通标志牌的标志牌信息。
其中,终端可以通过预设的外方位元素,将该三维点云结构投影到交通标志牌图像中该三维点云结构对应的位置。
需要说明的是,外方位元素是确定摄影的主光束在包含交通标志牌的区域的几何关系的基本数据,用于表征摄影的主光束在摄影瞬间的空间位置和空间姿态。该外方位元素包括的空间位置主要为三个直线元素,用于指示主光束在包含交通标志牌的区域的坐标位置,姿态参数主要为三个角元素,用于表达图像中的空间姿态。
还需要说明的是,该指定平面坐标系可以是以任一点为原点,在交通标志牌的图像中建立的平面坐标系,且该指定平面坐标系可以为平面直角坐标系、极坐标系等,本申请实施例对此不做具体限定。
另外,由于终端在得到该交通标志牌的影像数据时,该交通标志牌的影像数据在成像时可能存在成像倾角,对该具有成像倾角的影像数据进行图像识别,可能会获取到不准确的标志牌信息,因此,终端在得到该交通标志牌的影像数据时,需要对该交通标志牌的影像数据进行矫正。
其中,终端矫正该交通标志牌的影像数据的操作可以为:终端确定该影像数据相对指定三维坐标系中任意两个坐标轴所在平面之间的倾角,将该影像数据沿着倾角方向投影到指定三维坐标系中任意两个坐标
轴所在平面,或者与该平面平行的平面中,确定投影后的投影数据为矫正后的影响数据。
需要说明的是,终端可以通过上述方法对该交通标志牌的影像数据进行矫正,也可以通过其他方法进行矫正,本申请实施例对此不做具体限定。
再者,终端对矫正后的交通标志牌进行图像识别时,可以通过基于颜色、形状和sift(scale_invariant feature transform,尺度不变特征变换)特征的识别方法,或者,基于HOG(Histogram of Oriented Gradient,方向梯度直方图)或纹理的识别方法,或者,基于深度学习的识别方法等,本申请实施例对此不做具体限定。
值得说明的是,基于上述描述,终端包括的车载移动测量***11不仅可以包括激光扫描***111和影像成像***112,还可以包括组合导航***,因此,当终端获取到该交通标志牌的标志牌信息时,可以将该标志牌信息通过组合导航***加入到高精度地图中,实现对该交通标志牌的标志牌信息的应用。
在本申请实施例中,终端可以对包含交通标志牌的区域进行激光扫描处理,得到多个区域点和交通标志牌的图像,由于交通标志牌具有对光线的反射强度较强的特点,因此,终端可以利用该特点从多个区域点中,确定该交通标志牌的多个激光点和该多个激光点的点云数据。另外,由于激光是一个主动式光源,在对包含该交通标志牌的区域进行激光扫描处理时,可以不受光照影响,提高后续获取标志牌信息的准确度。之后,终端可以根据该多个激光点的点云数据,将该多个激光点进行进空间聚类得到该交通标志牌的三维点云结构,并确定该三维点云结构的空间分布特征,根据该空间分布特征,判断该三维点云结构是否为干扰结构,从而可以剔除一些干扰结构,减轻了终端后续的操作负担。另外,
当该三维点云结构不为干扰结构时,可以确定该交通标志牌的拟合平面方程,并基于该交通标志牌的图像、三维点云结构和拟合平面方程,对该影像数据进行图像识别等操作,以获取该交通标志牌的标志牌信息。由于该多个激光点的点云数据可以准确的表达该三维点云结构在指定三维坐标系中的坐标,终端基于该准确的坐标进行的一系列操作均为准确的操作,减少了获取标志牌信息的误差,从而提高了获取交通标志牌的标志牌信息的准确度、效率以及鲁棒性。
图2A是本申请实施例提供的一种标志牌信息获取装置的结构示意图,参见图2A,该装置包括:扫描模块201、聚类模块202、确定模块203和获取模块204。
扫描模块201,用于对包含交通标志牌的区域进行激光扫描处理,得到该交通标志牌的图像,以及该交通标志牌的多个激光点和该多个激光点的点云数据,各个激光点的点云数据用于对该各个激光点进行描述;
聚类模块202,用于基于该多个激光点的点云数据,对该多个激光点进行空间聚类,得到该交通标志牌的三维点云结构;
确定模块203,用于基于该三维点云结构,确定该交通标志牌的拟合平面方程;
获取模块204,用于基于该交通标志牌的图像、三维点云结构和拟合平面方程,获取该交通标志牌的标志牌信息。
可选地,参见图2B,该扫描模块201包括:
扫描单元2011,用于对包含该交通标志牌的区域进行激光扫描处理,得到该交通标志牌的图像,以及多个区域点和该多个区域点的点云数据,各个区域点的点云数据中包括该各个区域点的反射率值;
选择单元2022,用于从该多个区域点中选择反射率值大于或等于反
射率阈值的多个区域点;
第一确定单元2033,用于将选择的多个区域点确定为该交通标志牌的多个激光点,并将该选择的多个区域点的点云数据确定为该多个激光点的点云数据。
可选地,参见图2C,该确定模块203包括:
分析单元2031,用于对该三维点云结构进行主成分分析,得到该三维点云结构的各个空间特征值;
第二确定单元2032,用于确定该三维点云结构的各个空间特征值中最小的空间特征值的特征向量,并确定该三维点云结构的质心在指定三维坐标系中的三维坐标;
第三确定单元2033,用于基于该特征向量和该三维点云结构的质心在该指定三维坐标系中的三维坐标,确定该交通标志牌的拟合平面方程。
可选地,参见图2D,该确定模块203还包括:
第四确定单元2034,用于基于该三维点云结构的各个空间特征值,确定该三维点云结构的第一分布值、第二分布值和曲率值,该第一分布值和该第二分布值用于描述该三维点云结构在三维空间中的分布形状,该曲率值用于描述该三维点云结构在该三维空间中的平面性;
判断单元2035,由于基于该三维点云结构的第一分布值、第二分布值和曲率值,判断该三维点云结构是否为干扰结构;
触发单元2036,用于当该三维点云结构不为干扰结构时,触发该第二确定单元确定该三维点云结构的各个空间特征值中最小的空间特征值的特征向量,并确定该三维点云结构的质心在指定三维坐标系中的三维坐标。
可选地,参见图2E,该判断单元2035包括:
第一确定子单元20351,用于确定该三维点云结构的长度、宽度和高度;
第二确定子单元20352,用于如果该三维点云结构的长度不位于指定长度范围内、该三维点云结构的宽度不位于指定宽度范围内、该三维点云结构的高度不位于指定高度范围内、该第一分布值大于第一分布阈值、该第二分布值小于第二分布阈值或该曲率值大于曲率阈值,则确定该三维点云结构为干扰结构,否则,确定该三维点云结构不为干扰结构。
可选地,参见图2F,该获取模块204包括:
第一投影单元2041,用于通过预设的外方位元素,将该三维点云结构投影到该交通标志牌的图像上,该外方位元素包括获取该图像时主光束的空间位置和姿态参数;
第五确定单元2042,用于确定该三维点云结构在该图像上所占的图像区域;
第二投影单元2043,用于将该图像区域内每个像素点在指定平面坐标系中的坐标通过该拟合平面方程投影到三维空间中,得到该交通标志牌的影像数据;
矫正单元2044,用于对该交通标志牌的影像数据进行矫正,以消除该影像数据的成像倾角;
识别单元2045,用于对矫正后的该交通标志牌的影像数据进行图像识别,得到该交通标志牌的标志牌信息。
在本申请实施例中,终端可以对包含交通标志牌的区域进行激光扫描处理,得到多个区域点和交通标志牌的图像,由于交通标志牌具有对光线的反射强度较强的特点,因此,终端可以利用该特点从多个区域点中,确定该交通标志牌的多个激光点和该多个激光点的点云数据。另外,由于激光是一个主动式光源,在对包含该交通标志牌的区域进行激光扫
描处理时,可以不受光照影响,提高后续获取标志牌信息的准确度。之后,终端可以根据该多个激光点的点云数据,将该多个激光点进行进空间聚类得到该交通标志牌的三维点云结构,并确定该三维点云结构的空间分布特征,根据该空间分布特征,判断该三维点云结构是否为干扰结构,从而可以剔除一些干扰结构,减轻了终端后续的操作负担。另外,当该三维点云结构不为干扰结构时,可以确定该交通标志牌的拟合平面方程,并基于该交通标志牌的图像、三维点云结构和拟合平面方程,对该影像数据进行图像识别等操作,以获取该交通标志牌的标志牌信息。由于该多个激光点的点云数据可以准确的表达该三维点云结构在指定三维坐标系中的坐标,终端基于该准确的坐标进行的一系列操作均为准确的操作,减少了获取标志牌信息的误差,从而提高了获取交通标志牌的标志牌信息的准确度、效率以及鲁棒性。
需要说明的是:上述实施例提供的标志牌信息获取装置在获取标志牌信息时,仅以上述各功能模块的划分进行举例说明,实际应用中,可以根据需要而将上述功能分配由不同的功能模块完成,即将装置的内部结构划分成不同的功能模块,以完成以上描述的全部或者部分功能。另外,上述实施例提供的标志牌信息获取装置与标志牌信息获取方法实施例属于同一构思,其具体实现过程详见方法实施例,这里不再赘述。
图3是本申请实施例提供的一种标志牌信息获取装置的终端结构示意图。参见图3,终端300可以包括通信单元310、包括有一个或一个以上计算机可读存储介质的存储器320、输入单元330、显示单元340、传感器350、音频电路360、WIFI(Wireless Fidelity,无线保真)模块370、包括有一个或者一个以上处理核心的处理器380、以及电源390等部件。本领域技术人员可以理解,图3中示出的终端结构并不构成对终端的限定,可以包括比图示更多或更少的部件,或者组合某些部件,或
者不同的部件布置。在本实施例中,终端还包括有一个或者一个以上的程序,这一个或者一个以上程序存储于存储器中,且经配置以由一个或者一个以上处理器执行,所述一个或者一个以上程序包含用于进行本申请实施例提供的如图1B所示的标志牌信息获取方法的指令。
本领域普通技术人员可以理解实现上述实施例的全部或部分步骤可以通过硬件来完成,也可以通过程序来指令相关的硬件完成,所述的程序可以存储于一种计算机可读存储介质中,上述提到的存储介质可以是只读存储器,磁盘或光盘等。
以上所述仅为本申请的较佳实施例,并不用以限制本申请,凡在本申请的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本申请的保护范围之内。
Claims (20)
- 一种交通标志牌信息获取方法,其特征在于,包括:在数据处理设备处,获取摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像;获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激光数据包括多个激光点的数据;对所述多个激光点进行空间聚类,得到至少两个候选点集合;获取各候选点集合中激光点的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合;利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息。
- 如权利要求1所述的方法,其特征在于,获取对所述区域进行激光扫描得到的所述场景的激光数据包括:获取激光扫描得到的多个第二激光点的数据,其中包括各个第二激光点的反射特性值;利用所述第二激光点中选择反射特性值大于或等于预设阈值的多个激光点的数据形成所述激光数据。
- 如权利要求1所述的方法,其特征在于,对所述多个激光点进行空间聚类,得到至少两个候选点集合包括:利用预设的距离阈值对所述激光数据中的所述多个激光点进行分组 以形成所述至少两个候选点集合,其中不同候选点集合中的激光点之间的空间距离大于所述距离阈值。
- 如权利要求1所述的方法,其特征在于,获取各候选点集合中激光点的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合包括:获取所述至少两个候选点集合中各候选点集合中激光点在至少两个维度上的空间分布特征值;利用至少两个预设的空间分布特征阈值和各候选点集合的所述空间分布特征值从所述至少两个候选点集合中选择至少一个候选点集合作为所述交通标志牌对应的点集合。
- 如权利要求4所述的方法,其特征在于,所述空间分布特征值选自:长度上的分布特征值、宽度上的分布特征值、高度上的分布特征值、线状分布特征值、面状分布特征值、体状分布特征值、曲率、拟合平面特征值;所述至少两个预设的空间分布特征阈值根据交通标志牌的尺寸范围、形状特征确定。
- 如权利要求1所述的方法,其特征在于,利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据包括:获取转换参数,所述转换参数用于进行激光点和场景图像之间的坐标转换;根据所述转换参数和所述交通标志牌对应的点集合确定所述交通标志牌的图像数据在所述图像中的位置,从所述图像中的所述位置提取所 述交通标志牌的图像数据。
- 如权利要求1所述的方法,其特征在于,从所述图像数据中提取出所述交通标志牌的标志牌信息包括:对所述交通标志牌的图像数据进行图像识别,以得到所述交通标志牌的标志牌信息。
- 如权利要求7所述的方法,其特征在于,进一步包括:利用所述交通标志牌对应的点集合对所述交通标志牌的图像数据进行矫正,以消除所述图像数据的成像倾角;对所述交通标志牌的图像数据进行图像识别包括:对矫正后的所述交通标志牌的图像数据进行图像识别。
- 如权利要求8所述的方法,其特征在于,利用所述交通标志牌对应的点云对所述交通标志牌的图像数据进行矫正包括:利用所述交通标志牌对应的点集合确定所述交通标志牌的拟合平面方程;利用该拟合平面方程对所述交通标志牌的图像数据进行矫正,以消除该图像数据的成像倾角。
- 一种标志牌信息获取装置,其特征在于,包括处理器和存储器,所述存储器中存储有计算机可读指令,所述指令可以使处理器执行以下操作:获取摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像;获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激光数据包括多个激光点的数据;对所述多个激光点进行空间聚类,得到至少两个候选点集合;获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合;利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息。
- 如权利要求10所述的装置,其特征在于,所述指令可以使处理器执行以下操作:获取激光扫描得到的多个第二激光点的数据,其中包括各个第二激光点的反射特性值;利用所述第二激光点中选择反射特性值大于或等于预设阈值的多个激光点形成所述激光数据。
- 如权利要求10所述的装置,其特征在于,所述指令可以使处理器执行以下操作:利用预设的距离阈值对所述激光数据中的所述多个激光点进行分组以形成所述至少两个候选点集合,其中不同候选点集合中的激光点之间的空间距离大于所述距离阈值。
- 如权利要求10所述的装置,其特征在于,所述指令可以使处理器执行以下操作:获取所述至少两个候选点集合中各候选点集合中激光点在至少两个 维度上的空间分布特征值;利用至少两个预设的空间分布特征阈值和各候选点集合的所述空间分布特征值从所述至少两个候选点集合中选择至少一个候选点集合作为所述交通标志牌对应的点集合。
- 如权利要求10所述的装置,其特征在于,所述指令可以使处理器执行以下操作:获取转换参数,所述转换参数用于进行激光点和场景图像之间的坐标转换;根据所述转换参数和所述交通标志牌对应的点集合确定所述交通标志牌的图像数据在所述图像中的位置,从所述图像中的所述位置提取所述交通标志牌的图像数据。
- 如权利要求10所述的装置,其特征在于,所述指令可以使处理器执行以下操作:利用所述交通标志牌对应的点集合对所述交通标志牌的图像数据进行矫正,以消除所述图像数据的成像倾角;对矫正后的所述交通标志牌的图像数据进行图像识别,以得到所述交通标志牌的标志牌信息。
- 一种计算机可读存储介质,其特征在于,存储有计算机可读指令,所述指令可以使处理器执行以下操作:获取摄像装置对包含交通标志牌的场景进行拍摄得到的所述场景的图像;获取对所述区域进行激光扫描得到的所述场景的激光数据,所述激 光数据包括多个激光点的数据;对所述多个激光点进行空间聚类,得到至少两个候选点集合;获取各候选点集合的空间分布特征,根据所述空间分布特征从所述至少两个候选点集合中确定所述交通标志牌对应的点集合;利用所述交通标志牌对应的点集合从所述图像中提取出所述交通标志牌的图像数据,从所述图像数据中提取出所述交通标志牌的标志牌信息。
- 如权利要求16所述的存储介质,其特征在于,所述指令可以使处理器执行以下操作:利用预设的距离阈值对所述激光点集合中的所述多个激光点进行分组以形成所述至少两个候选点集合,其中不同候选点集合中的激光点之间的空间距离大于所述距离阈值。
- 如权利要求16所述的存储介质,其特征在于,所述指令可以使处理器执行以下操作:获取所述至少两个候选点集合中各候选点集合中激光点在至少两个维度上的空间分布特征值;利用至少两个预设的空间分布特征阈值和各候选点集合的所述空间分布特征值从所述至少两个候选点集合中选择至少一个候选点集合作为所述交通标志牌对应的点集合。
- 如权利要求16所述的存储介质,其特征在于,所述指令可以使处理器执行以下操作:获取转换参数,所述转换参数用于进行激光点和场景图像之间的坐 标转换;根据所述转换参数和所述交通标志牌对应的点集合确定所述交通标志牌的图像数据在所述图像中的位置,从所述图像中的所述位置提取所述交通标志牌的图像数据。
- 如权利要求16所述的存储介质,其特征在于,所述指令可以使处理器执行以下操作:利用所述交通标志牌对应的点集合对所述交通标志牌的图像数据进行矫正,以消除所述图像数据的成像倾角;对矫正后的所述交通标志牌的图像数据进行图像识别,以得到所述交通标志牌的标志牌信息。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US16/016,981 US10891500B2 (en) | 2016-09-13 | 2018-06-25 | Method and apparatus for acquiring traffic sign information |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610822521.4 | 2016-09-13 | ||
CN201610822521.4A CN107818288B (zh) | 2016-09-13 | 2016-09-13 | 标志牌信息获取方法及装置 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US16/016,981 Continuation US10891500B2 (en) | 2016-09-13 | 2018-06-25 | Method and apparatus for acquiring traffic sign information |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2018049998A1 true WO2018049998A1 (zh) | 2018-03-22 |
Family
ID=61600642
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2017/100358 WO2018049998A1 (zh) | 2016-09-13 | 2017-09-04 | 交通标志牌信息获取方法及装置 |
Country Status (3)
Country | Link |
---|---|
US (1) | US10891500B2 (zh) |
CN (1) | CN107818288B (zh) |
WO (1) | WO2018049998A1 (zh) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2560799A (en) * | 2017-01-23 | 2018-09-26 | Ford Global Tech Llc | Sign recognition for autonomous vehicles |
CN109685893A (zh) * | 2019-01-11 | 2019-04-26 | 中煤航测遥感集团有限公司 | 空间一体化建模方法及装置 |
CN110148196A (zh) * | 2018-09-12 | 2019-08-20 | 腾讯大地通途(北京)科技有限公司 | 一种图像处理方法、装置以及相关设备 |
CN111652050A (zh) * | 2020-04-20 | 2020-09-11 | 宁波吉利汽车研究开发有限公司 | 一种交通标志的定位方法、装置、设备和介质 |
CN113960572A (zh) * | 2021-10-20 | 2022-01-21 | 北京轻舟智航科技有限公司 | 一种过滤埋地灯噪声点云的处理方法和装置 |
Families Citing this family (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6985207B2 (ja) * | 2018-05-09 | 2021-12-22 | トヨタ自動車株式会社 | 自動運転システム |
CN108846333B (zh) * | 2018-05-30 | 2022-02-18 | 厦门大学 | 标志牌地标数据集生成及车辆定位方法 |
CN109633681A (zh) * | 2018-12-05 | 2019-04-16 | 芜湖智久机器人有限公司 | 一种反光板识别方法及装置 |
CN109856643B (zh) * | 2018-12-15 | 2022-10-04 | 国网福建省电力有限公司检修分公司 | 基于3d激光的可移动式无感全景感知方法 |
US11055546B2 (en) * | 2018-12-18 | 2021-07-06 | Here Global B.V. | Automatic positioning of 2D image sign sightings in 3D space |
CN111664845B (zh) * | 2019-03-05 | 2023-06-16 | 千寻位置网络有限公司 | 交通标志定位、视觉地图制作方法及装置、定位*** |
CN111724472B (zh) * | 2019-03-19 | 2024-07-02 | 北京四维图新科技股份有限公司 | 地图要素空间位置的确定方法和装置 |
CN110018491B (zh) * | 2019-04-16 | 2021-06-22 | 广州大学 | 激光扫描方法、装置及激光雷达 |
CN112069856B (zh) * | 2019-06-10 | 2024-06-14 | 商汤集团有限公司 | 地图生成方法、驾驶控制方法、装置、电子设备及*** |
CN112348876B (zh) * | 2019-08-08 | 2024-06-11 | 北京地平线机器人技术研发有限公司 | 一种标志牌的空间坐标获取方法及装置 |
JP7313998B2 (ja) * | 2019-09-18 | 2023-07-25 | 株式会社トプコン | 測量データ処理装置、測量データ処理方法および測量データ処理用プログラム |
CN111126211B (zh) * | 2019-12-13 | 2023-08-29 | 北京四维图新科技股份有限公司 | 标牌识别方法和装置、电子设备 |
CN111243310B (zh) * | 2020-01-10 | 2021-08-31 | 广州大学 | 交通标志识别方法、***、介质和设备 |
CN113496145B (zh) * | 2020-03-19 | 2023-06-16 | 武汉四维图新科技有限公司 | 标牌角点获取方法、装置及设备 |
CN111383337B (zh) * | 2020-03-20 | 2023-06-27 | 北京百度网讯科技有限公司 | 用于识别对象的方法和装置 |
CN111667545B (zh) * | 2020-05-07 | 2024-02-27 | 东软睿驰汽车技术(沈阳)有限公司 | 高精度地图生成方法、装置、电子设备及存储介质 |
CN112184736B (zh) * | 2020-10-10 | 2022-11-11 | 南开大学 | 一种基于欧式聚类的多平面提取方法 |
CN112990034B (zh) * | 2021-03-23 | 2023-09-26 | 武汉大学 | 一种影像优先的交通标牌变化检测方法 |
CN115128626A (zh) * | 2021-03-26 | 2022-09-30 | 西安四维图新信息技术有限公司 | 交通标牌矢量化方法、***、终端及存储介质 |
CN113098791B (zh) * | 2021-03-30 | 2022-05-06 | 中山大学 | 一种多业务网络流的成分占比分析方法 |
CN113298866B (zh) * | 2021-06-11 | 2024-01-23 | 梅卡曼德(北京)机器人科技有限公司 | 物体分类方法及装置 |
CN114004951B (zh) * | 2021-10-29 | 2024-07-12 | 武汉大学 | 联合点云强度和几何结构的道路标志提取方法及*** |
CN117689832B (zh) * | 2024-02-04 | 2024-05-17 | 腾讯科技(深圳)有限公司 | 交通标牌生成方法、装置、设备和存储介质 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102779280A (zh) * | 2012-06-19 | 2012-11-14 | 武汉大学 | 一种基于激光传感器的交通信息提取方法 |
CN104280036A (zh) * | 2013-07-05 | 2015-01-14 | 北京四维图新科技股份有限公司 | 一种交通信息的检测与定位方法、装置及电子设备 |
CN104866840A (zh) * | 2015-06-04 | 2015-08-26 | 广东中城规划设计有限公司 | 一种从机载激光点云数据识别架空电力线的方法 |
US9193355B2 (en) * | 2012-09-05 | 2015-11-24 | Google Inc. | Construction zone sign detection using light detection and ranging |
CN105678318A (zh) * | 2015-12-31 | 2016-06-15 | 百度在线网络技术(北京)有限公司 | 交通标牌的匹配方法及装置 |
CN106560835A (zh) * | 2015-09-30 | 2017-04-12 | 高德软件有限公司 | 一种路牌识别方法及装置 |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101907081B1 (ko) * | 2011-08-22 | 2018-10-11 | 삼성전자주식회사 | 3차원 점군의 물체 분리 방법 |
US9383753B1 (en) * | 2012-09-26 | 2016-07-05 | Google Inc. | Wide-view LIDAR with areas of special attention |
DE102012110595A1 (de) * | 2012-11-06 | 2014-05-08 | Conti Temic Microelectronic Gmbh | Verfahren und Vorrichtung zur Erkennung von Verkehrszeichen für ein Fahrzeug |
CN103499343B (zh) * | 2013-09-25 | 2015-06-24 | 武汉海达数云技术有限公司 | 基于三维激光反射强度的道路标线点云自动分类识别方法 |
US9368936B1 (en) * | 2013-09-30 | 2016-06-14 | Google Inc. | Laser diode firing system |
JP6203208B2 (ja) * | 2015-02-18 | 2017-09-27 | 株式会社東芝 | 道路構造物管理システム、及び道路構造物管理方法 |
CN105184852B (zh) * | 2015-08-04 | 2018-01-30 | 百度在线网络技术(北京)有限公司 | 一种基于激光点云的城市道路识别方法及装置 |
US10528055B2 (en) * | 2016-11-03 | 2020-01-07 | Ford Global Technologies, Llc | Road sign recognition |
US20180211119A1 (en) * | 2017-01-23 | 2018-07-26 | Ford Global Technologies, Llc | Sign Recognition for Autonomous Vehicles |
-
2016
- 2016-09-13 CN CN201610822521.4A patent/CN107818288B/zh active Active
-
2017
- 2017-09-04 WO PCT/CN2017/100358 patent/WO2018049998A1/zh active Application Filing
-
2018
- 2018-06-25 US US16/016,981 patent/US10891500B2/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102779280A (zh) * | 2012-06-19 | 2012-11-14 | 武汉大学 | 一种基于激光传感器的交通信息提取方法 |
US9193355B2 (en) * | 2012-09-05 | 2015-11-24 | Google Inc. | Construction zone sign detection using light detection and ranging |
CN104280036A (zh) * | 2013-07-05 | 2015-01-14 | 北京四维图新科技股份有限公司 | 一种交通信息的检测与定位方法、装置及电子设备 |
CN104866840A (zh) * | 2015-06-04 | 2015-08-26 | 广东中城规划设计有限公司 | 一种从机载激光点云数据识别架空电力线的方法 |
CN106560835A (zh) * | 2015-09-30 | 2017-04-12 | 高德软件有限公司 | 一种路牌识别方法及装置 |
CN105678318A (zh) * | 2015-12-31 | 2016-06-15 | 百度在线网络技术(北京)有限公司 | 交通标牌的匹配方法及装置 |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2560799A (en) * | 2017-01-23 | 2018-09-26 | Ford Global Tech Llc | Sign recognition for autonomous vehicles |
CN110148196A (zh) * | 2018-09-12 | 2019-08-20 | 腾讯大地通途(北京)科技有限公司 | 一种图像处理方法、装置以及相关设备 |
CN110148196B (zh) * | 2018-09-12 | 2022-03-25 | 腾讯大地通途(北京)科技有限公司 | 一种图像处理方法、装置以及相关设备 |
CN109685893A (zh) * | 2019-01-11 | 2019-04-26 | 中煤航测遥感集团有限公司 | 空间一体化建模方法及装置 |
CN111652050A (zh) * | 2020-04-20 | 2020-09-11 | 宁波吉利汽车研究开发有限公司 | 一种交通标志的定位方法、装置、设备和介质 |
CN111652050B (zh) * | 2020-04-20 | 2024-04-02 | 宁波吉利汽车研究开发有限公司 | 一种交通标志的定位方法、装置、设备和介质 |
CN113960572A (zh) * | 2021-10-20 | 2022-01-21 | 北京轻舟智航科技有限公司 | 一种过滤埋地灯噪声点云的处理方法和装置 |
CN113960572B (zh) * | 2021-10-20 | 2024-05-03 | 北京轻舟智航科技有限公司 | 一种过滤埋地灯噪声点云的处理方法和装置 |
Also Published As
Publication number | Publication date |
---|---|
US10891500B2 (en) | 2021-01-12 |
US20180307924A1 (en) | 2018-10-25 |
CN107818288A (zh) | 2018-03-20 |
CN107818288B (zh) | 2019-04-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2018049998A1 (zh) | 交通标志牌信息获取方法及装置 | |
US11255973B2 (en) | Method and apparatus for extracting lane line and computer readable storage medium | |
JP7190842B2 (ja) | 情報処理装置、情報処理装置の制御方法及びプログラム | |
CN111566437B (zh) | 三维测量***以及三维测量方法 | |
WO2020102944A1 (zh) | 点云处理方法、设备及存储介质 | |
US10872227B2 (en) | Automatic object recognition method and system thereof, shopping device and storage medium | |
US20190206071A1 (en) | Method and apparatus for recovering point cloud data | |
CN112613378B (zh) | 3d目标检测方法、***、介质及终端 | |
WO2019042426A1 (zh) | 增强现实场景的处理方法、设备及计算机存储介质 | |
US11030763B1 (en) | System and method for identifying items | |
CN111815707B (zh) | 点云确定方法、点云筛选方法、装置、计算机设备 | |
CN111709923B (zh) | 一种三维物体检测方法、装置、计算机设备和存储介质 | |
WO2021114776A1 (en) | Object detection method, object detection device, terminal device, and medium | |
WO2021114777A1 (en) | Target detection method, terminal device, and medium | |
WO2022217988A1 (zh) | 传感器配置方案确定方法、装置、计算机设备、存储介质及程序 | |
CN114611635B (zh) | 一种对象的识别方法、装置、存储介质及电子装置 | |
CN115147333A (zh) | 一种目标检测方法及装置 | |
CN111626241A (zh) | 一种人脸检测方法及装置 | |
CN113808202B (zh) | 一种多目标检测和空间定位方法及其*** | |
WO2021114775A1 (en) | Object detection method, object detection device, terminal device, and medium | |
KR102249381B1 (ko) | 3차원 영상 정보를 이용한 모바일 디바이스의 공간 정보 생성 시스템 및 방법 | |
CN109376653B (zh) | 用于定位车辆的方法、装置、设备和介质 | |
JP4546155B2 (ja) | 画像処理方法、画像処理装置および画像処理プログラム | |
CN113281770A (zh) | 坐标系关系获取方法及装置 | |
CN114897968B (zh) | 车辆视野的确定方法、装置、计算机设备、存储介质 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 17850200 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 17850200 Country of ref document: EP Kind code of ref document: A1 |