CN116184438A - Data processing method for identifying bad weather based on laser radar - Google Patents
Data processing method for identifying bad weather based on laser radar Download PDFInfo
- Publication number
- CN116184438A CN116184438A CN202310101249.0A CN202310101249A CN116184438A CN 116184438 A CN116184438 A CN 116184438A CN 202310101249 A CN202310101249 A CN 202310101249A CN 116184438 A CN116184438 A CN 116184438A
- Authority
- CN
- China
- Prior art keywords
- weather
- point cloud
- data
- rain
- rainfall
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
- G01S17/95—Lidar systems specially adapted for specific applications for meteorological use
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Radar, Positioning & Navigation (AREA)
- Computer Networks & Wireless Communication (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Electromagnetism (AREA)
- Remote Sensing (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
The invention relates to a data processing method for identifying bad weather based on a laser radar, and belongs to the technical field of intelligent traffic radar data processing. The method for identifying the type of severe weather by utilizing the point cloud characteristics of the laser radar is provided by combining the laser radar to obtain the track data of sunny days, rainfall and snowfall weather, so that the application range of the laser radar is enlarged, good weather reference conditions are provided for the laser radar to monitor the road side data of the vehicle, the spatial characteristics of target track information are enriched, and the practicability is strong. According to the invention, the extracted laser radar point cloud data is fully extracted, the purposes of inputting pictures, extracting and comparing features, outputting weather types and weather intensity levels, directly outputting weather results end to end and predicting duration time of severe weather are realized, the weather conditions and duration time of specific roads can be finely distinguished, and the method has practicability.
Description
Technical Field
The invention relates to a data processing method for identifying bad weather based on a laser radar, and belongs to the technical field of intelligent traffic radar data processing.
Background
Weather identification and classification are one of important classification problems in the technical field of computer data processing. Weather conditions are important factors affecting the running safety of vehicles, and traffic accidents caused by severe weather such as rain, snow and the like account for more than 30% of the total accidents. The normal operation of road traffic is seriously affected by severe rain and snow weather, and especially the commute peak time causes the congestion index to rise straight up. The expressway road surface weather identification is to identify whether the weather conditions are bad or not specifically aiming at the weather conditions of traffic roads, and judging the weather conditions timely through road surface images is helpful for better traffic management. In recent years, many researchers have developed in the fields of intelligent transportation and automatic driving, but safety driving still cannot be completely guaranteed in severe weather, severe weather type judgment is monitored in real time and fed back to drivers in time, so that driving safety can be further improved, and a foundation is laid for related research on vehicle operation safety in severe weather.
The first method is to set up static weather monitoring station in order to collect the data such as road ponding and snow depth, temperature, visibility, etc. on the road, monitor and identify the weather condition on the highway section, this kind of method is from the overall effect of the rain and snow weather, neglect the rain and snow change of the dynamic course in the rainfall process, it is difficult to describe the dynamic association of the traffic characteristic and rainfall and snowfall in the rainfall and snowfall accurately, this method is not suitable for the relevant study under the severe weather environment in driving; the second method is to use digital images to process information and classify the road surface weather of the expressway through image processing, deep learning and other methods, especially in terms of man-machine interaction, and to use monitoring videos or detect the current weather category, but the method is difficult to quantify the severity of the weather, and still faces the fine challenges in the road surface weather image classification task.
Disclosure of Invention
Aiming at the defects of the existing bad weather identification algorithm, the invention aims to solve the defects of the identification algorithm, and provides a bad weather identification method based on a laser radar.
The laser radar is widely applied to the aspects of automatic driving, road side perception, high-precision image construction and the like as an important sensor, and can quickly acquire three-dimensional position point information (namely point cloud data) of a target in space by transmitting and receiving laser beams and further determine key characteristic data such as the position, the size, the external contour and the like of the target.
The technical scheme of the invention is as follows:
a data processing method for identifying bad weather based on a laser radar comprises the following steps:
s1: acquiring laser radar point cloud signal intensity data of different weather types, and preprocessing the data, wherein the weather types comprise sunny days, rainfall and snowfall, the rainfall is divided into heavy rain, medium rain, light rain and capillary rain, and the snowfall is divided into heavy snow, medium snow and light snow;
s2: constructing a training set and a testing set according to the laser radar point cloud signal intensity data obtained in the step S1, and constructing corresponding state data association characteristics of different weather types, namely sunny days, rainfall snowfall intensities and point cloud patterns;
s3: clustering the state characteristics under different weather types to obtain point cloud clusters, and determining the data association characteristics of the corresponding point cloud clusters under different weather types;
s4: judging the consistency condition of the data association characteristic of a cloud picture sample and a trained sunny point cloud cluster, judging the consistency condition of the data association characteristic of a cloud picture sample and a trained rainfall or snowfall point cloud cluster, acquiring the rainfall or snowfall degree level and predicting the duration of the rainfall or snowfall.
Preferably, in step S1, the process of preprocessing data includes:
s1-1: interpolation is carried out on the lost data, and unreasonable data outside the set response threshold value is removed;
s1-2: performing combined noise reduction processing on the point cloud data to realize point cloud data processing and characteristic restoration of a target organism in severe weather;
the combined noise reduction method is realized by a combined filter, the combined filter is composed of spatial filtering, ground filtering, voxel filtering and statistical outlier filtering, and the characteristic that the filtering algorithm generally needs to scan the whole point cloud spatial data is considered, and the combined noise reduction method is firstly used for removing irrelevant data in space by the spatial filtering algorithm. According to the position of the target point cloud, unnecessary walls, trees and the like in the point cloud space are removed, and the actually required point cloud data of the target point cloud are reserved in sunny days, raining days and snowing days.
The invention can improve the image quality and remove high-frequency noise and interference; filtering noise points returned by the ground surface and the ground object; the complexity of data processing time is reduced while maintaining the shape characteristics of the point cloud; and filtering out point clouds outside the threshold value by using statistics. The combined filter can effectively remove leakage points caused by a single filter.
Preferably, step S2 includes:
s2-1: selecting a known weather as sunny days, a rainfall weather as heavy rain, medium rain, light rain and capillary rain, and respectively collecting three-dimensional point cloud charts of the laser radar when snowfall weather is heavy snow, medium snow and small snow, and constructing a plurality of groups of sample data sets according to continuous time periods of the same weather of the same place and different time periods of different weather of the same place to train and test a convolutional neural network;
s2-2: in each three-dimensional point cloud image acquired in S2-1, the center of the image is taken as the origin of coordinates, 5 key square points are respectively positioned at the left upper part, the left lower part, the right upper part, the right lower part and the center, cubes with the side length of n units are taken as the interested areas, the position of each cube is represented by two three-dimensional diagonal point coordinates, each point is formed by three-dimensional vectors, namely n1 (x 1, y1, z 1), n2 (x 2, y2, z 2) …, namely the position of the first cube is represented as n12 (x 1-x2, y1-y2, z1-z 2) …, the method comprises the steps of inputting a plurality of groups of sample information into a convolutional neural network model (CNN), extracting and constructing a data association characteristic relation corresponding to heavy rain, medium rain, light rain, capillary rain, heavy snow, medium snow, light snow and the like in rainfall weather on a sunny day according to a laser radar point cloud image of the same weather in the same place in a continuous time period of the same weather and in different time periods of different weather in the same place, and determining data association characteristics according to weather conditions, wherein the data association characteristic comprises the sum of the number of point clouds, the reflection intensity value of the point clouds, the threshold range of the point cloud sparse characteristics and the like.
Convolutional Neural Networks (CNNs) are suitable for computer vision tasks, such as image processing, image classification, etc., to solve image classification problems. The step of convolving the neural network generally comprises convolving, pooling, and color map multi-channel input of the data set, and outputting a final result after multiple convolutions and pooling.
The total number of the point clouds, the reflection intensity value of the point clouds, the sparse characteristics (sparseness and space distribution condition) of the point clouds and the like in the step are determined by the result output through the neural network, and the same weather type has a dynamic fluctuation range, namely a threshold range.
Preferably, the step S3 specifically includes:
s3-1: acquiring a point cloud image data set of the extracted associated information of different weather types, wherein the point cloud image data set comprises classified weather type point cloud data, determining the sum of target point cloud quantity data from the classified point cloud data, and determining a processing weather type corresponding to the target point cloud data;
s3-2: the method comprises the steps of processing weather types including heavy rain, medium rain, light rain, capillary rain in sunny days, and heavy snow, medium snow and small snow in snowfall weather, marking data according to the weather types, for example, marking 1,2 and 3 … backwards in sequence, marking the same number through the data threshold of the related characteristics in the same range, namely, the same weather type, clustering the same type, and clustering cloud data of target points to obtain a clustering result;
s3-3: after the clustering result is obtained, carrying out point cloud data visualization, showing and marking weather types of 5 regions of interest in the same point cloud image, and counting dynamic change data association characteristics (i.e. instant frequency state information) of duration intervals of the same region of interest, wherein the dynamic change data association characteristics comprise point cloud number sum characteristics of different regions of interest, point cloud reflection intensity characteristics of different regions of interest and point cloud sparse characteristic distribution rules;
s3-4: and determining the consistency of the data association characteristics of 5 regions of interest in the same map, and then defining the weather type of the point cloud map, wherein the weather type comprises sunny days, heavy rain, medium rain, light rain, capillary rain of rainfall weather, and heavy snow, medium snow and small snow of snowfall weather, the point cloud characteristic range changing along with time and the defined rainfall/snowfall grade index.
Preferably, step S4 specifically includes:
s4-1: preprocessing a cloud image sample of a certain point obtained by the current laser radar, and performing the step S1;
s4-2: 5 key square points and interested areas are defined by the preprocessed current laser radar point cloud diagram, and the key square points and the interested areas are compared with a threshold range of data association features in sunny weather, wherein the threshold range comprises dynamic change data association features of duration time intervals of the same interested area, the sum total feature of point clouds of different interested areas, the point cloud reflection intensity and the point cloud sparse feature distribution rule of different interested areas, and whether the feature range is consistent in sunny weather is judged;
s4-3: if the current laser radar point cloud picture is within the threshold range of the data association characteristic in sunny weather, the sunny weather is the sunny weather, and if not, the step S4-4 is executed;
s4-4: 5 key square points and interested areas are defined by the preprocessed current laser radar point cloud image, the current laser radar point cloud image is compared with a threshold range of data association characteristics in rainfall weather and snowfall weather obtained in the step S2, the threshold range comprises dynamic change data association characteristics of duration intervals of the same interested area, namely point cloud number sum characteristics of different interested areas, point cloud reflection intensity and point cloud sparse characteristic distribution rules of different interested areas, rainfall and snowfall types of the image are judged according to the consistency of the data association characteristics of the 5 interested areas of the same image, and rainfall and snowfall grades are determined, namely heavy rain, medium rain, light rain, capillary rain, riot snow, heavy snow, medium snow and light snow of the rainfall weather;
step S4-5: and (3) after determining the rainfall and snowfall grades, repeating the step (S3-3) to perform point cloud data visual display, training the time-frequency state information of the weather type of the current place, forming the current rainfall trend according to the time interval and the rainfall intensity, and predicting the continuous duration time of the rainfall type.
Many training sample results are visualized to form a trend from which the duration of rainfall can be determined (e.g., declining).
The law and data obtained in the step S2 are large-batch, and then the step S3 is carried out: the weather type classification determines whether the weather is rainy or snowy or sunny, then refines, for example, compares the rainfall data in the same type to determine whether the weather is heavy rain or light rain, ensures that the image is displayed as the same weather type when the consistency result appears through 5 key square points in the same image, and enters the step S4 to input data to obtain the final determination of heavy rain, medium rain, light rain and wool rain.
The invention is not exhaustive and can be seen in the prior art.
The beneficial effects of the invention are as follows:
1. the invention combines the laser radar to scan the track data of the road in sunny days, rainfall, snowfall and other weather, and provides a method for identifying bad weather types by utilizing the point cloud characteristics of the laser radar, the method fully utilizes the signal attenuation degree (the power of the emitted laser beam and the received laser beam can be compared and attenuation change) of the laser radar when the bad weather occurs, thereby identifying the weather types and the bad weather degree level;
2. according to the invention, the extracted laser radar point cloud data is fully extracted, the purposes of inputting pictures, extracting features, comparing, outputting weather types and weather intensity levels, directly outputting weather results end to end and predicting severe weather duration are realized, and compared with the case of judging the road surface of the expressway by weather forecast, the weather condition and duration of a specific road can be judged in a fine granularity mode, so that the method has practicability.
Drawings
FIG. 1 is a flow chart of a data processing method based on the identification of bad weather by a laser radar;
FIG. 2 is a schematic view of a three-dimensional point cloud image selecting 5 orientations of a region of interest;
FIG. 3 is a schematic diagram of cloud data of a certain rainfall weather;
FIG. 4 is a schematic diagram of some snowfall point cloud data;
FIG. 5 is a schematic view of point cloud data prior to snowfall pretreatment;
fig. 6 is a schematic diagram of point cloud data after pretreatment in snowfall weather.
Detailed Description
In order to make the technical problems, technical solutions and advantages to be solved by the present invention more apparent, the following detailed description will be given with reference to the accompanying drawings and specific embodiments, but not limited thereto, and the present invention is not fully described and is according to the conventional technology in the art.
Example 1
A data processing method for identifying bad weather based on a laser radar is shown in fig. 1-6, and comprises the following steps:
s1: acquiring laser radar point cloud signal intensity data of different weather types, and preprocessing the data, wherein the weather types comprise sunny days, rainfall and snowfall, the rainfall is divided into heavy rain, medium rain, light rain and capillary rain, and the snowfall is divided into heavy snow, medium snow and light snow;
s2: constructing a training set and a testing set according to the laser radar point cloud signal intensity data obtained in the step S1, and constructing corresponding state data association characteristics of different weather types, namely sunny days, rainfall snowfall intensities and point cloud patterns;
s3: clustering the state characteristics under different weather types to obtain point cloud clusters, and determining the data association characteristics of the corresponding point cloud clusters under different weather types;
s4: judging the consistency condition of the data association characteristic of a cloud picture sample and a trained sunny point cloud cluster, judging the consistency condition of the data association characteristic of a cloud picture sample and a trained rainfall or snowfall point cloud cluster, acquiring the rainfall or snowfall degree level and predicting the duration of the rainfall or snowfall.
Example 2
A data processing method for identifying bad weather based on a lidar, as in embodiment 1, except that in step S1, the process of preprocessing the data includes:
s1-1: interpolation is carried out on the lost data, and unreasonable data outside the set response threshold value is removed;
s1-2: performing combined noise reduction processing on the point cloud data to realize point cloud data processing and characteristic restoration of a target organism in severe weather;
the combined noise reduction method is realized by a combined filter, the combined filter is composed of spatial filtering, ground filtering, voxel filtering and statistical outlier filtering, and the characteristic that the filtering algorithm generally needs to scan the whole point cloud spatial data is considered, and the combined noise reduction method is firstly used for removing irrelevant data in space by the spatial filtering algorithm. According to the position of the target point cloud, unnecessary walls, trees and the like in the point cloud space are removed, and the actually required point cloud data of the target point cloud are reserved in sunny days, raining days and snowing days.
The invention can improve the image quality and remove high-frequency noise and interference; filtering noise points returned by the ground surface and the ground object; the complexity of data processing time is reduced while maintaining the shape characteristics of the point cloud; and filtering out point clouds outside the threshold value by using statistics. The combined filter can effectively remove leakage points caused by a single filter.
Example 3
A data processing method for identifying bad weather based on a lidar, as described in embodiment 2, except that step S2 includes:
s2-1: selecting a known weather as sunny days, a rainfall weather as heavy rain, medium rain, light rain and capillary rain, and respectively collecting three-dimensional point cloud charts of the laser radar when snowfall weather is heavy snow, medium snow and small snow, and constructing a plurality of groups of sample data sets according to continuous time periods of the same weather of the same place and different time periods of different weather of the same place to train and test a convolutional neural network;
s2-2: in each three-dimensional point cloud image acquired in S2-1, the center of the image is taken as the origin of coordinates, 5 key square points are respectively positioned at the left upper part, the left lower part, the right upper part, the right lower part and the center, cubes with the side length of n units are taken as the interested areas, the position of each cube is represented by two three-dimensional diagonal point coordinates, each point is formed by three-dimensional vectors, namely n1 (x 1, y1, z 1), n2 (x 2, y2, z 2) …, namely the position of the first cube is represented as n12 (x 1-x2, y1-y2, z1-z 2) …, the method comprises the steps of inputting a plurality of groups of sample information into a convolutional neural network model (CNN), extracting and constructing a data association characteristic relation corresponding to heavy rain, medium rain, light rain, capillary rain, heavy snow, medium snow, light snow and the like in rainfall weather on a sunny day according to a laser radar point cloud image of the same weather in the same place in a continuous time period of the same weather and in different time periods of different weather in the same place, and determining data association characteristics according to weather conditions, wherein the data association characteristic comprises the sum of the number of point clouds, the reflection intensity value of the point clouds, the threshold range of the point cloud sparse characteristics and the like.
Convolutional Neural Networks (CNNs) are suitable for computer vision tasks, such as image processing, image classification, etc., to solve image classification problems. The step of convolving the neural network generally comprises convolving, pooling, and color map multi-channel input of the data set, and outputting a final result after multiple convolutions and pooling.
The total number of the point clouds, the reflection intensity value of the point clouds, the sparse characteristics (sparseness and space distribution condition) of the point clouds and the like in the step are determined by the result output through the neural network, and the same weather type has a dynamic fluctuation range, namely a threshold range.
Example 4
A data processing method for identifying bad weather based on a lidar, as in embodiment 3, except that step S3 specifically includes:
s3-1: acquiring a point cloud image data set of the extracted associated information of different weather types, wherein the point cloud image data set comprises classified weather type point cloud data, determining the sum of target point cloud quantity data from the classified point cloud data, and determining a processing weather type corresponding to the target point cloud data;
s3-2: the method comprises the steps of processing weather types including heavy rain, medium rain, light rain and capillary rain of sunny days, and heavy snow, medium snow and small snow of snowfall weather, processing data labels 1,2 and 3 … according to the weather types, labeling the same number when the data threshold values of the related features are in the same range, namely the same weather type, clustering the same type, and clustering cloud data of target points to obtain a clustering result;
s3-3: after the clustering result is obtained, carrying out point cloud data visualization, showing and marking weather types of 5 regions of interest in the same point cloud image, and counting dynamic change data association characteristics (i.e. instant frequency state information) of duration intervals of the same region of interest, wherein the dynamic change data association characteristics comprise point cloud number sum characteristics of different regions of interest, point cloud reflection intensity characteristics of different regions of interest and point cloud sparse characteristic distribution rules;
s3-4: and determining the consistency of the data association characteristics of 5 regions of interest in the same map, and then defining the weather type of the point cloud map, wherein the weather type comprises sunny days, heavy rain, medium rain, light rain, capillary rain of rainfall weather, and heavy snow, medium snow and small snow of snowfall weather, the point cloud characteristic range changing along with time and the defined rainfall/snowfall grade index.
Example 5
A data processing method for identifying bad weather based on a lidar is as described in embodiment 4, except that step S4 specifically includes:
s4-1: preprocessing a cloud image sample of a certain point obtained by the current laser radar, and performing the step S1;
s4-2: 5 key square points and interested areas are defined by the preprocessed current laser radar point cloud diagram, and the key square points and the interested areas are compared with a threshold range of data association features in sunny weather, wherein the threshold range comprises dynamic change data association features of duration time intervals of the same interested area, the sum total feature of point clouds of different interested areas, the point cloud reflection intensity and the point cloud sparse feature distribution rule of different interested areas, and whether the feature range is consistent in sunny weather is judged;
s4-3: if the current laser radar point cloud picture is within the threshold range of the data association characteristic in sunny weather, the sunny weather is the sunny weather, and if not, the step S4-4 is executed;
s4-4: 5 key square points and interested areas are defined by the preprocessed current laser radar point cloud image, the current laser radar point cloud image is compared with a threshold range of data association characteristics in rainfall weather and snowfall weather obtained in the step S2, the threshold range comprises dynamic change data association characteristics of duration intervals of the same interested area, namely point cloud number sum characteristics of different interested areas, point cloud reflection intensity and point cloud sparse characteristic distribution rules of different interested areas, rainfall and snowfall types of the image are judged according to the consistency of the data association characteristics of the 5 interested areas of the same image, and rainfall and snowfall grades are determined, namely heavy rain, medium rain, light rain, capillary rain, riot snow, heavy snow, medium snow and light snow of the rainfall weather;
step S4-5: and (3) after determining the rainfall and snowfall grades, repeating the step (S3-3) to perform point cloud data visual display, training the time-frequency state information of the weather type of the current place, forming the current rainfall trend according to the time interval and the rainfall intensity, and predicting the continuous duration time of the rainfall type.
While the foregoing is directed to the preferred embodiments of the present invention, it will be appreciated by those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present invention, and such modifications and adaptations are intended to be comprehended within the scope of the present invention.
Claims (5)
1. The data processing method for identifying bad weather based on the laser radar is characterized by comprising the following steps:
s1: acquiring laser radar point cloud signal intensity data of different weather types, and preprocessing the data, wherein the weather types comprise sunny days, rainfall and snowfall, the rainfall is divided into heavy rain, medium rain, light rain and capillary rain, and the snowfall is divided into heavy snow, medium snow and light snow;
s2: constructing a training set and a testing set according to the laser radar point cloud signal intensity data obtained in the step S1, and constructing corresponding state data association characteristics of different weather types, namely sunny days, rainfall snowfall intensities and point cloud patterns;
s3: clustering the state characteristics under different weather types to obtain point cloud clusters, and determining the data association characteristics of the corresponding point cloud clusters under different weather types;
s4: judging the consistency condition of the data association characteristic of a cloud picture sample and a trained sunny point cloud cluster, judging the consistency condition of the data association characteristic of a cloud picture sample and a trained rainfall or snowfall point cloud cluster, acquiring the rainfall or snowfall degree level and predicting the duration of the rainfall or snowfall.
2. The data processing method for identifying bad weather based on laser radar according to claim 1, wherein in step (1), the process of preprocessing the data in step S1 includes:
s1-1: the lost data is interpolated, and the data outside the set response threshold value is removed;
s1-2: performing combined noise reduction processing on the point cloud data to realize point cloud data processing and characteristic restoration of a target organism in severe weather;
the combined noise reduction method is realized by a combined filter, and the combined filter is composed of spatial filtering, ground filtering, voxel filtering and statistical outlier filtering.
3. The data processing method for identifying bad weather based on lidar according to claim 2, wherein step S2 comprises:
s2-1: selecting a known weather as sunny days, a rainfall weather as heavy rain, medium rain, light rain and capillary rain, and respectively collecting three-dimensional point cloud charts of the laser radar when snowfall weather is heavy snow, medium snow and small snow, and constructing a plurality of groups of sample data sets according to continuous time periods of the same weather of the same place and different time periods of different weather of the same place to train and test a convolutional neural network;
s2-2: in each three-dimensional point cloud image acquired in the step S2-1, the center of the image is taken as a coordinate origin, 5 key square points are respectively positioned at the left upper part, the left lower part, the right upper part, the right lower part and the center, cubes with the side length of n unit lengths are taken as interested areas, the position of each cube is represented by two three-dimensional diagonal point coordinates, each point is formed by three-dimensional vectors, multiple groups of sample information are input into a convolutional neural network model, and data association characteristic relations corresponding to heavy rain, medium rain, light rain, capillary rain, heavy snow, medium snow and small snow in rainy weather are extracted and constructed according to the continuous time period of the same weather of the same place and different time periods of different weather of the same place, and the data association characteristics including the sum of the number of point clouds, the reflection intensity value of the point clouds and the threshold range of the point cloud sparse characteristics are determined according to weather conditions.
4. The data processing method for identifying bad weather based on the lidar according to claim 3, wherein the step S3 is specifically:
s3-1: acquiring a point cloud image data set of the extracted associated information of different weather types, determining the sum of the number of target point clouds from the classified point cloud data, and determining the processing weather type corresponding to the target point cloud data;
s3-2: the method comprises the steps of processing weather types including sunny days, heavy rain, medium rain, light rain, capillary rain in rainfall weather, and heavy snow, medium snow and small snow in snowfall weather, marking data according to the weather types, and clustering target point cloud data to obtain clustering results;
s3-3: after a clustering result is obtained, carrying out point cloud data visualization, displaying and labeling weather types of 5 regions of interest in the same point cloud image, and counting dynamic change data association characteristics of duration intervals of the same region of interest, namely, time-frequency state information, wherein the time-frequency state information comprises the sum total characteristics of the point clouds of different regions of interest, the point cloud reflection intensity characteristics of different regions of interest and the distribution rule of the point cloud sparse characteristics;
s3-4: and determining the consistency of the data association characteristics of 5 regions of interest in the same map, and then defining the weather type of the point cloud map, wherein the weather type comprises sunny days, heavy rain, medium rain, light rain, capillary rain of rainfall weather, and heavy snow, medium snow and small snow of snowfall weather, the point cloud characteristic range changing along with time and the defined rainfall/snowfall grade index.
5. The data processing method for identifying bad weather based on the lidar according to claim 4, wherein step S4 is specifically:
s4-1: preprocessing a cloud image sample of a certain point obtained by the current laser radar, and performing the step S1;
s4-2: 5 key square points and interested areas are defined by the preprocessed current laser radar point cloud diagram, and the key square points and the interested areas are compared with a threshold range of data association features in sunny weather, wherein the threshold range comprises dynamic change data association features of duration time intervals of the same interested area, the sum total feature of point clouds of different interested areas, the point cloud reflection intensity and the point cloud sparse feature distribution rule of different interested areas, and whether the feature range is consistent in sunny weather is judged;
s4-3: if the current laser radar point cloud picture is within the threshold range of the data association characteristic in sunny weather, the sunny weather is the sunny weather, and if not, the step S4-4 is executed;
s4-4: defining 5 key square points and interested areas from the preprocessed current laser radar point cloud image, comparing the current laser radar point cloud image with a threshold range of data association characteristics in rainy weather and snowy weather, wherein the threshold range comprises dynamic change data association characteristics of the same interested area at a continuous time interval, namely the total point cloud number characteristics of different interested areas, the point cloud reflection intensity and the point cloud sparse characteristic distribution rule of different interested areas, and judging rainfall and snowfall types of the image according to the consistency of the data association characteristics of the 5 interested areas of the same image, and determining rainfall and snowfall grades, namely heavy rain, medium rain, light rain, capillary rain, heavy snow, medium snow and light snow of the rainy weather;
step S4-5: and (3) after determining the rainfall and snowfall grades, repeating the step (S3-3) to perform point cloud data visual display, training the time-frequency state information of the weather type of the current place, forming the current rainfall trend according to the time interval and the rainfall intensity, and predicting the continuous duration time of the rainfall type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310101249.0A CN116184438A (en) | 2023-02-08 | 2023-02-08 | Data processing method for identifying bad weather based on laser radar |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310101249.0A CN116184438A (en) | 2023-02-08 | 2023-02-08 | Data processing method for identifying bad weather based on laser radar |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116184438A true CN116184438A (en) | 2023-05-30 |
Family
ID=86433996
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310101249.0A Pending CN116184438A (en) | 2023-02-08 | 2023-02-08 | Data processing method for identifying bad weather based on laser radar |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116184438A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740935A (en) * | 2023-06-26 | 2023-09-12 | 河北高速公路集团有限公司 | Expressway environment prediction method, device, equipment and storage medium |
| CN117647852A (en) * | 2024-01-29 | 2024-03-05 | 吉咖智能机器人有限公司 | Weather state detection method and device, electronic equipment and storage medium |
-
2023
- 2023-02-08 CN CN202310101249.0A patent/CN116184438A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740935A (en) * | 2023-06-26 | 2023-09-12 | 河北高速公路集团有限公司 | Expressway environment prediction method, device, equipment and storage medium |
| CN116740935B (en) * | 2023-06-26 | 2024-04-30 | 河北高速公路集团有限公司 | Expressway environment prediction method, device, equipment and storage medium |
| CN117647852A (en) * | 2024-01-29 | 2024-03-05 | 吉咖智能机器人有限公司 | Weather state detection method and device, electronic equipment and storage medium |
| CN117647852B (en) * | 2024-01-29 | 2024-04-09 | 吉咖智能机器人有限公司 | Weather state detection method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116184438A (en) | Data processing method for identifying bad weather based on laser radar | |
| CN101430195B (en) | Method for computing electric power line ice-covering thickness by using video image processing technology | |
| Yang et al. | Image-based visibility estimation algorithm for intelligent transportation systems | |
| CN114783183A (en) | Monitoring method and system based on traffic situation algorithm | |
| CN111027430B (en) | Traffic scene complexity calculation method for intelligent evaluation of unmanned vehicles | |
| CN107067781B (en) | GIS road black spot map generation method for advanced driving assistance system application | |
| CN115081508B (en) | Traffic running risk parallel simulation system based on traffic digital twin | |
| CN108875593A (en) | Visible images weather recognition methods based on convolutional neural networks | |
| Xu et al. | A gradient boost approach for predicting near-road ultrafine particle concentrations using detailed traffic characterization | |
| CN113869196B (en) | Vehicle type classification method and device based on laser point cloud data multi-feature analysis | |
| CN115206103B (en) | Variable speed limit control system based on parallel simulation system | |
| CN112818775B (en) | Forest road rapid identification method and system based on regional boundary pixel exchange | |
| CN111210163B (en) | Public transport vehicle risk evaluation system and method based on multi-source data | |
| CN109448397A (en) | A kind of group's mist monitoring method based on big data | |
| Li et al. | Crash report data analysis for creating scenario-wise, spatio-temporal attention guidance to support computer vision-based perception of fatal crash risks | |
| CN113269960A (en) | Method for analyzing accident risk of highway tunnel group traffic | |
| CN111242223B (en) | Street space quality evaluation method based on streetscape image multi-feature fusion | |
| CN115205796A (en) | Method and system for monitoring foreign matter invasion limit and early warning risk of track line | |
| CN113888850B (en) | Sand-dust meteorological municipal road congestion situation multi-factor early warning method | |
| Hallmark et al. | Examining the effects of winter road maintenance operations on traffic safety through visual analytics | |
| Sun et al. | Objects detection with 3-D roadside LiDAR under snowy weather | |
| CN111539363A (en) | Highway rockfall identification and analysis method | |
| CN113284340A (en) | Road traffic safety grading system | |
| CN112816483A (en) | Group fog recognition early warning method and system based on fog value analysis and electronic equipment | |
| Ghashghaie et al. | Spatial statistics analysis to identify hot spots using accidental event calls services |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |