CN107704837A - The extracting method of road network topological sum geological information - Google Patents
The extracting method of road network topological sum geological information Download PDFInfo
- Publication number
- CN107704837A CN107704837A CN201710981117.6A CN201710981117A CN107704837A CN 107704837 A CN107704837 A CN 107704837A CN 201710981117 A CN201710981117 A CN 201710981117A CN 107704837 A CN107704837 A CN 107704837A
- Authority
- CN
- China
- Prior art keywords
- road
- network structure
- geological information
- topological sum
- extracting method
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/182—Network patterns, e.g. roads or rivers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/34—Smoothing or thinning of the pattern; Morphological operations; Skeletonisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of extracting method of road network topological sum geological information.This method comprises the following steps：Aerial images are just being penetrated to high accuracy using deep learning and are carrying out lane segmentation acquisition bianry image, the bianry image is being pre-processed；The processing obtained image of pretreatment is to grow the whole network structure of road；The whole network structure of the road is modified to obtain correct road network topology information.This method to high-precision aerial images on the basis of fine lane segmentation is carried out based on deep learning, preprocess method is provided, to reduce the influence of flase drop existing for lane segmentation and missing inspection, good robustness is shown, can automatically generate complete, correct road network topological sum geological information.
Description
Technical field
The present invention relates to the extracting method of road topology and geological information, more particularly to road topology and geological information carry
Take method.
Background technology
Road is automatically extracted from satellite remote sensing and aerial images, is one and 20 years of researches classes has been carried out
Topic.Existing mode mainly utilizes computer vision and machine learning techniques, using rim detection, feature extraction, template matches,
The methods of Dynamic Programming, fuzzy set, multiple scale detecting, random field models or neutral net, carry out the extraction of road.But due to road
The complex nature of the problem is extracted on road (such as has close vehicle and pedestrian, road face color and building top color, road and week on road
Blocked without substantially demarcating and trees being present in exclosure face), the degree of accuracy of extraction and robustness are always to realize the obstruction of automation.
The generation of road extraction and road network topological sum geological information at present is still based on artificial.
By network and patent search, closest technical scheme is as follows with the present invention：
(1) patent《A kind of method of the Road network extraction based on self-adaption cluster study》(patent No. application number：
201610273017.3).The target problem that this method needs to solve extracts including road network.Road network extraction is in stage extraction road
On the basis of, the geometry based on section connection is spliced come what is carried out, is then based on what chi structure was spliced again
Amendment.
(2) patent《A kind of high-definition remote sensing map road extracting method based on K means》(number of patent application：
201410219942.9).The target problem that this method needs to solve is road extraction, but exports the lane segmentation for original image
Image, rather than the road network topological sum geological information of structuring.It the method use connected domain filtering and carry out noise reduction, and to image
Skeletal extraction is carried out.
Existing technology mainly carries out road extraction using computer vision and machine learning techniques, and in extraction process
Generate road axis and topology information.The accuracy of road network topological sum geological information is limited to the road that algorithm can reach
The accuracy of road extraction, and have impact on the final value of technology.Meanwhile existing research focuses on the image point of road network mostly
Cut, be seldom related to available for navigation route planning and aid in the road network topological sum geometry letter of high-precision map lane line extraction
Breath automatically generates.
In image segmentation field, having shown traditional computer vision and machine learning techniques can not compare deep learning
The advantage of plan.Especially in the case where magnanimity marker samples data be present, deep learning, which has, obtains the latent of duration performance lifting
Power.
Aerial images are just being penetrated to high accuracy using deep learning convolutional neural networks and are carrying out road fine segmentation, can larger journey
The accuracy rate of degree lifting segmentation, but due to lane segmentation the complex nature of the problem, the situation of false retrieval and missing inspection still occurs.
The content of the invention
The present invention solves the problems, such as it is that the extraction of existing road topological sum geological information has false retrieval and missing inspection, no
It can automatically extract and the problem of poor robustness.
To solve the above problems, the present invention provides a kind of extracting method of road network topological sum geological information, this method bag
Include following steps：Aerial images are just being penetrated to high accuracy using deep learning and are carrying out lane segmentation acquisition bianry image, to described two
Value image is pre-processed；The processing obtained image of pretreatment is to grow the whole network structure of road；To the road
Whole network structure is modified to obtain correct road network topology information.
In further scheme, the pretreatment is to carry out institute using computer vision morphological method and connected domain analysis
State pretreatment.
In further scheme, the pretreatment includes first down-sampled to bianry image progress；Then, closed using morphology
Computing-first expand down-sampled obtained image described in post-etching, to fill minuscule hole and connection adjacent domain；Finally, make after
With connected domain analysis, connected domain all on image is obtained, only retains the connected domain of maximum area.
In further scheme, the whole network structure for growing road includes：Extracted from the image of pretreatment
The node of network structure；Seed fill algorithm is used to all nodes, iteration grows whole network structure.
In further scheme, the node of network structure is extracted in the image of the pretreatment to be included：Using skeletal extraction
Algorithm obtains single pixel skeleton, then carries out convolution algorithm to the single pixel skeleton image by mean filter to extract netted knot
The node of structure.
In the further scheme, methods described is also included after carrying out convolution operation to image, and statistics is all to be more than 255/3
Node of the pixel as network structure, for all neighborhood of nodes, only retain one, delete other all nodes adjacent thereto,
So as to obtain all nodes of network structure.
In further scheme, the whole network structure to the road is modified including designing netted operator
Except the short side in network structure.
In further scheme, the whole network structure to the road is modified including designing netted operator
Except the foot bridge in network structure.
In further scheme, the whole network structure to the road, which is modified, also to be included to network structure
Each edge carries out curve fitting, and carries out denoising and smooth.
In further scheme, the whole network structure to the road, which is modified, also to be included to denoising and smooth
Image carry out vector quantization, and each edge is vacuated, removes unnecessary point.
Compared with prior art, the present invention at least has advantages below：
Because the present invention carries out lane segmentation acquisition bianry image to just penetrating aerial images to high accuracy using deep learning,
The bianry image is pre-processed, the missing inspection and false retrieval to segmentation result are corrected；The figure that processing pretreatment obtains
As automatically to grow the whole network structure of road；The whole network structure of the road is modified to obtain correctly
Road network topology information.So, full-automatic generation road network topological sum geological information is realized.After tested, algorithm has height
Robustness.The road network information of generation can be used for traditional navigation route planning, it can also be used to the road segment segment of high-precision map producing
Segmentation and track line drawing.
Brief description of the drawings
Fig. 1 is the flow chart of the extracting method of road network topological sum geological information of the present invention；
Fig. 2 is the flow chart of iteration growth；
Fig. 3 is the lane segmentation binary map based on deep learning；
Fig. 4 is the road network generated by the method for the present invention；
Fig. 5 is the road bianry image of manual drawing；
Fig. 6 is the road network generated by the method for the present invention.
Embodiment
To describe technology contents, construction feature, institute's reached purpose and effect of the present invention in detail, below in conjunction with embodiment
And accompanying drawing is coordinated to be described in detail.
The present invention is just penetrating the base of aerial images progress lane segmentation using deep learning convolutional neural networks to high accuracy
On plinth, using computer vision morphological method and connected domain analysis, false retrieval in segmentation result and the situation of missing inspection are carried out
Amendment；Then skeletal extraction algorithm is used, is extracted the single pixel skeleton of road image network structure；Pass through mean filter pair again
Single pixel skeleton image carries out convolution algorithm, is extracted the node of network structure；Then all nodes are calculated using seed filling
Method, iteration grow whole network structure；And devise two netted operators：Short side and foot bridge are removed, to the netted knot of generation
Structure is trimmed；The each edge of network structure is carried out curve fitting afterwards, carries out denoising and smooth；Vector quantization is carried out again, and
Each edge is vacuated, removes unnecessary point.So far, road network topological sum geological information, which automatically generates, finishes.
Fig. 1 and Fig. 2 are referred to, the extracting method of road network topological sum geological information of the present invention comprises the following steps：
Automatically generating for road network topological sum geological information is divided into three phases by technical scheme：Lane segmentation
The pretreatment of bianry image, road network structure automatically generating and correcting.
(1) pretreatment of lane segmentation bianry image
In the pretreatment stage of lane segmentation bianry image, the down-sampled of image is carried out first, to lift subsequent algorithm
Perform speed.Down-sampled degree depends on requirement of the subsequent applications for road network geometric positioning accuracy.For navigation way
Planning and the extraction offer application such as road direction and shape information to high-precision track, will for road network geometric positioning accuracy
Ask not high, larger sample rate can be used.By down-sampled, after having obtained the bianry image of small in resolution, morphology is used
Closed operation-first expand post-etching, to fill minuscule hole and connection adjacent domain.Then connected domain analysis is used, obtains image
Upper all connected domains, only retain the connected domain of maximum area, so as to eliminate all isolated zonules.Subsequently into next
Stage.
(2) road network structure automatically generates
The stage is automatically generated in road network structure, skeletal extraction is carried out to the image that pretreatment obtains, has obtained list
The road network skeleton of pixel wide.Mean filter is carried out using 3X3 1/9 unit matrix as core, convolution operation is carried out to image
Afterwards, all nodes (point that 3 sides are intersected) for being more than 255/3 pixel as network structure are counted, for all neighborhood of nodes,
Only retain one, other all nodes adjacent thereto are deleted, so as to obtain all nodes of network structure.
Operate below and all nodes of road are performed, as shown in Figure 2.The adjacent pixel point set of node is obtained, first upper bottom left
The right side, then diagonal.Then all neighbor pixels are circulated, judges whether the point is " use ".If it is not, create
One side for originating in the node, and the pixel is added in the point sequence on side, mark the point that the pixel is " use ".
Start growth iteration, and incoming adjacent pixel point set to the side, the growth as following iteration needs the point excluded.Life on side
In long iteration, last point in the point sequence on the side is obtained, as seed, is obtained
Obtain its all adjacent pixel point sets.Concentrated in neighbor pixel and remove the needs that the iteration is passed to from previous step
The point of exclusion, all points " used " are removed, remove first node on the side.Then to remaining neighbor pixel
Collection circulates, if neighbor pixel is node, by the node join side, and terminates the iteration growth on the side；Otherwise by phase
Adjacent pixel adds the point sequence on side, marks the point that the pixel is " use ", then starts the growth iteration of a new round, and
Incoming adjacent pixel point set, the growth as following iteration need the point excluded.By above iterative algorithm, can grow whole
Individual network structure.
(3) the automatic amendment of road network structure
The network structure generated by above iterative algorithm, because the spy of the flase drop of lane segmentation and skeletal extraction algorithm
Point, short side and foot bridge can be included.They are not belonging to road network information, it should are deleted.The technical program is by constructing
Except short side and foot bridge both network structure operators, all short sides and foot bridge can be effectively removed, so as to obtain correct road
Road network topology information.
Remove being implemented as the network structure operator of short side：All sides are traveled through, length is obtained and is less than predetermined threshold value
And the only side of individual node, the side and node are deleted in network structure.
Remove being implemented as the network structure operator of foot bridge：All sides are traveled through, length is obtained and is less than predetermined threshold value
And have the side of two nodes, the side is deleted in network structure, and two nodes on the side are merged into a node.
In currently acquired network structure, the point sequence in each bar side is still the adjacent pixel point set that end points is node
Close.The technical program has carried out curve matching to the point sequence, generates new point sequence.The process of fitting is to original sequence
Row have carried out noise reduction and fairing, so as to obtain the shape information on more preferable side.
Then vector quantization is carried out to new point sequence, each edge is described with LineString, every two in the point sequence on side
Individual consecutive points constitute LineString line segment.On the basis of vector quantization, the LineString of each edge is vacuated,
To obtain describing the minimum point set needed for shape.
Up to the present, automatically generating for road network topological sum geological information has been fully completed.
Based on the road network topological sum geological information automatically generated, the planning of guidance path can be achieved.And it can give high-precision
A bit in aerial images on road is spent, to obtain the road direction and curvature information at the point.The information can be used for realizing
Lane line automatically extracts the noise reduction process of process in high-precision map, remove it is all through edge filter find with road direction not
The edge of symbol.
Fig. 3 to Fig. 6 is referred to, because the present invention carries out road point to just penetrating aerial images to high accuracy using deep learning
Acquisition bianry image is cut, the bianry image is pre-processed；The processing obtained image of pretreatment is to grow the whole of road
Individual network structure；The whole network structure of the road is modified to obtain correct road network topology information, so, no
The situation of false retrieval and missing inspection occurs, shows good robustness, can automatically generate complete, correct road network and open up
Flutter and geological information.
Claims (10)
1. a kind of extracting method of road network topological sum geological information, it is characterized in that：This method comprises the following steps：
Aerial images are just being penetrated to high accuracy using deep learning and are carrying out lane segmentation acquisition bianry image, the bianry image is being entered
Row pretreatment；
The processing obtained image of pretreatment is to grow the whole network structure of road；
The whole network structure of the road is modified to obtain correct road network topology information.
2. the extracting method of road network topological sum geological information as claimed in claim 1, it is characterized in that：The pretreatment is to adopt
The pretreatment is carried out with computer visual morphological method and connected domain analysis.
3. the extracting method of road network topological sum geological information as claimed in claim 2, it is characterized in that：The pretreatment includes
First bianry image is carried out down-sampled；Then, down-sampled obtained figure described in post-etching is expanded using closing operation of mathematical morphology-first
Picture, to fill minuscule hole and connection adjacent domain；Finally, connected domain all on image is obtained using connected domain analysis afterwards,
Only retain the connected domain of maximum area.
4. the extracting method of road network topological sum geological information as claimed in claim 1, it is characterized in that：It is described to grow road
Whole network structure include：The node of network structure is extracted from the image of pretreatment；Seed filling is used to all nodes
Algorithm, iteration grow whole network structure.
5. the extracting method of road network topological sum geological information as claimed in claim 4, it is characterized in that：The figure of the pretreatment
The node of network structure is extracted as in be included：Single pixel skeleton is obtained using skeletal extraction algorithm, then by mean filter to institute
Single pixel skeleton image is stated to carry out convolution algorithm and extract the node of network structure.
6. the extracting method of road network topological sum geological information as claimed in claim 5, it is characterized in that：Methods described also includes
After convolution operation being carried out to image, all nodes for being more than 255/3 pixel as network structure of statistics, for all adjacent bonds
Point, only retain one, other all nodes adjacent thereto are deleted, so as to obtain all nodes of network structure.
7. the extracting method of road network topological sum geological information as claimed in claim 1, it is characterized in that：It is described to the road
Whole network structure be modified and remove short side in network structure, including all sides of traversal including designing netted operator, obtain
Length be less than predetermined threshold value and the only side of individual node, the side and node are deleted in network structure.
8. the extracting method of the road network topological sum geological information as described in claim 1 or 7, it is characterized in that：It is described to described
The whole network structure of road is modified the foot bridge removed including the netted operator of design in network structure, including traversal owns
Side, obtain that length is less than predetermined threshold value and have the side of two nodes, the side is deleted in network structure, and by two of the side
Node merges into a node.
9. the extracting method of road network topological sum geological information as claimed in claim 8, it is characterized in that：It is described to the road
Whole network structure be modified and also include carrying out curve fitting to each edge of network structure, and carry out denoising and smooth.
10. the extracting method of road network topological sum geological information as claimed in claim 9, it is characterized in that：It is described to the road
The whole network structure on road, which is modified, also to be included carrying out vector quantization to denoising and smooth image, and each edge is taken out
It is dilute, remove unnecessary point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710981117.6A CN107704837A (en) | 2017-10-19 | 2017-10-19 | The extracting method of road network topological sum geological information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710981117.6A CN107704837A (en) | 2017-10-19 | 2017-10-19 | The extracting method of road network topological sum geological information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107704837A true CN107704837A (en) | 2018-02-16 |
Family
ID=61181313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710981117.6A Pending CN107704837A (en) | 2017-10-19 | 2017-10-19 | The extracting method of road network topological sum geological information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107704837A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109472416A (en) * | 2018-11-15 | 2019-03-15 | 索奇智能科技（北京）有限公司 | Indoor path planning method and device based on automatic road network data extraction, and client |
| CN109558801A (en) * | 2018-10-31 | 2019-04-02 | 厦门大学 | Road network extraction method, medium, computer equipment and system |
| CN109948477A (en) * | 2019-03-06 | 2019-06-28 | 东南大学 | Method for extracting road network topology points in picture |
| CN110389991A (en) * | 2018-04-12 | 2019-10-29 | 腾讯大地通途（北京）科技有限公司 | The processing method that reports an error, device and the storage medium in map section |
| CN111507153A (en) * | 2019-01-31 | 2020-08-07 | 斯特拉德视觉公司 | Post-processing method and device for detecting lane line by using segmentation score map and cluster map |
| CN112950665A (en) * | 2021-02-01 | 2021-06-11 | 武汉大学 | Semi-automatic extraction method and system for curved target |
| CN113724385A (en) * | 2021-07-30 | 2021-11-30 | 深圳市普渡科技有限公司 | Robot topology map generation system, method, computer device and storage medium |
| CN114399124A (en) * | 2022-03-24 | 2022-04-26 | 腾讯科技（深圳）有限公司 | Path data processing method, path planning method, path data processing device, path planning device and computer equipment |
| CN114627257A (en) * | 2022-03-15 | 2022-06-14 | 北京百度网讯科技有限公司 | Three-dimensional road network map construction method and device, electronic equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101364259A (en) * | 2008-04-09 | 2009-02-11 | 武汉大学 | Method for extracting road various information of multi-level knowledge driven panchromatic remote sensing image |
| CN101901343A (en) * | 2010-07-20 | 2010-12-01 | 同济大学 | Remote sensing image road extracting method based on stereo constraint |
| CN101976335A (en) * | 2010-09-03 | 2011-02-16 | 浙江大学 | Remote-sensing image road network extract method based on improved C-V model |
| CN104657978A (en) * | 2014-12-24 | 2015-05-27 | 福州大学 | Road extracting method based on shape characteristics of roads of remote sensing images |
| CN105719300A (en) * | 2016-01-22 | 2016-06-29 | 黄河水利委员会信息中心 | Riverway main stream line detection method based on SNE manifold learning |
| CN106548182A (en) * | 2016-11-02 | 2017-03-29 | 武汉理工大学 | Based on deep learning and the causal analytic pavement distress survey method and device of master |
| CN106919915A (en) * | 2017-02-22 | 2017-07-04 | 武汉极目智能技术有限公司 | Map road mark and road quality harvester and method based on ADAS systems |
-
2017
- 2017-10-19 CN CN201710981117.6A patent/CN107704837A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101364259A (en) * | 2008-04-09 | 2009-02-11 | 武汉大学 | Method for extracting road various information of multi-level knowledge driven panchromatic remote sensing image |
| CN101901343A (en) * | 2010-07-20 | 2010-12-01 | 同济大学 | Remote sensing image road extracting method based on stereo constraint |
| CN101976335A (en) * | 2010-09-03 | 2011-02-16 | 浙江大学 | Remote-sensing image road network extract method based on improved C-V model |
| CN104657978A (en) * | 2014-12-24 | 2015-05-27 | 福州大学 | Road extracting method based on shape characteristics of roads of remote sensing images |
| CN105719300A (en) * | 2016-01-22 | 2016-06-29 | 黄河水利委员会信息中心 | Riverway main stream line detection method based on SNE manifold learning |
| CN106548182A (en) * | 2016-11-02 | 2017-03-29 | 武汉理工大学 | Based on deep learning and the causal analytic pavement distress survey method and device of master |
| CN106919915A (en) * | 2017-02-22 | 2017-07-04 | 武汉极目智能技术有限公司 | Map road mark and road quality harvester and method based on ADAS systems |
Non-Patent Citations (6)
| Title |
|---|
| J. AMINI 等: "Automatic road-side extraction from large scale imagemaps", 《INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION》 * |
| J.B. MENA: "Automatic vectorization of segmented road networks by geometrical and topological analysis of high resolution binary images", 《KNOWLEDGE BASED SYSTEMS》 * |
| 刘如意 等: "一种自动的高分辨率遥感影像道路提取方法", 《西安电子科技大学学报（自然科学版）》 * |
| 李峰 等: "机载LiDAR点云提取城市道路网的半自动方法", 《测绘科学》 * |
| 栾学晨 等: "城市道路网主干道提取的形态分析方法", 《武汉大学学报·信息科学版》 * |
| 赵晓峰: "高分辨率遥感影像城区道路提取方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110389991A (en) * | 2018-04-12 | 2019-10-29 | 腾讯大地通途（北京）科技有限公司 | The processing method that reports an error, device and the storage medium in map section |
| CN110389991B (en) * | 2018-04-12 | 2022-02-22 | 腾讯大地通途（北京）科技有限公司 | Error reporting processing method and device for map section and storage medium |
| CN109558801A (en) * | 2018-10-31 | 2019-04-02 | 厦门大学 | Road network extraction method, medium, computer equipment and system |
| CN109472416B (en) * | 2018-11-15 | 2022-05-13 | 索奇智能科技（北京）有限公司 | Indoor path planning method and device based on automatic road network data extraction, and client |
| CN109472416A (en) * | 2018-11-15 | 2019-03-15 | 索奇智能科技（北京）有限公司 | Indoor path planning method and device based on automatic road network data extraction, and client |
| CN111507153A (en) * | 2019-01-31 | 2020-08-07 | 斯特拉德视觉公司 | Post-processing method and device for detecting lane line by using segmentation score map and cluster map |
| CN111507153B (en) * | 2019-01-31 | 2023-12-15 | 斯特拉德视觉公司 | Post-processing method and device for detecting lane lines by using segmentation score graph and cluster graph |
| CN109948477A (en) * | 2019-03-06 | 2019-06-28 | 东南大学 | Method for extracting road network topology points in picture |
| CN109948477B (en) * | 2019-03-06 | 2022-05-13 | 东南大学 | Method for extracting road network topology points in picture |
| CN112950665A (en) * | 2021-02-01 | 2021-06-11 | 武汉大学 | Semi-automatic extraction method and system for curved target |
| CN113724385A (en) * | 2021-07-30 | 2021-11-30 | 深圳市普渡科技有限公司 | Robot topology map generation system, method, computer device and storage medium |
| CN114627257A (en) * | 2022-03-15 | 2022-06-14 | 北京百度网讯科技有限公司 | Three-dimensional road network map construction method and device, electronic equipment and storage medium |
| CN114399124A (en) * | 2022-03-24 | 2022-04-26 | 腾讯科技（深圳）有限公司 | Path data processing method, path planning method, path data processing device, path planning device and computer equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107704837A (en) | The extracting method of road network topological sum geological information | |
| CN109409263B (en) | Method for detecting urban ground feature change of remote sensing image based on Siamese convolutional network | |
| CN111986099B (en) | Tillage monitoring method and system based on convolutional neural network with residual error correction fused | |
| Wegner et al. | Road networks as collections of minimum cost paths | |
| Bastani et al. | Roadtracer: Automatic extraction of road networks from aerial images | |
| CN105931295B (en) | A kind of geologic map Extracting Thematic Information method | |
| CN110853026B (en) | Remote sensing image change detection method integrating deep learning and region segmentation | |
| CN103927526B (en) | Vehicle detecting method based on Gauss difference multi-scale edge fusion | |
| CN113076871B (en) | Fish shoal automatic detection method based on target shielding compensation | |
| CN111738165B (en) | Method for extracting individual plant canopy from high-resolution unmanned aerial vehicle visible light remote sensing image | |
| CN112541483B (en) | Dense face detection method combining YOLO and blocking-fusion strategy | |
| CN107871124A (en) | A kind of Remote Sensing Target detection method based on deep neural network | |
| CN110163213B (en) | Remote sensing image segmentation method based on disparity map and multi-scale depth network model | |
| CN108875595A (en) | A kind of Driving Scene object detection method merged based on deep learning and multilayer feature | |
| CN110443818A (en) | A kind of Weakly supervised semantic segmentation method and system based on scribble | |
| CN111126287B (en) | Remote sensing image dense target deep learning detection method | |
| Sun et al. | Information fusion for rural land-use classification with high-resolution satellite imagery | |
| Ni et al. | HyperSfM | |
| CN108022251A (en) | A kind of extracting method and system of the center line of tubular structure | |
| CN109271928A (en) | A kind of road network automatic update method based on the fusion of vector road network with the verifying of high score remote sensing image | |
| CN111582093A (en) | Automatic small target detection method in high-resolution image based on computer vision and deep learning | |
| CN106169186A (en) | Based on the method obtaining initial profile in level set moving object detection | |
| Attarzadeh et al. | Object-based building extraction from high resolution satellite imagery | |
| Ok | Automated extraction of buildings and roads in a graph partitioning framework | |
| CN107480693A (en) | Condition random field framework is embedded in the Weakly supervised image scene understanding method of registration information |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180216 |
|
| WD01 | Invention patent application deemed withdrawn after publication |