CN101425128A - People stream detecting method based on laser sensor depth image - Google Patents
People stream detecting method based on laser sensor depth image Download PDFInfo
- Publication number
- CN101425128A CN101425128A CNA2007100475903A CN200710047590A CN101425128A CN 101425128 A CN101425128 A CN 101425128A CN A2007100475903 A CNA2007100475903 A CN A2007100475903A CN 200710047590 A CN200710047590 A CN 200710047590A CN 101425128 A CN101425128 A CN 101425128A
- Authority
- CN
- China
- Prior art keywords
- depth image
- people
- target
- stream
- method based
- 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
Abstract
Firstly, a laser sensing camera is used for shooting deep images of a multiframe which is idle in a detected region so as to obtain a topographic map of the ground of the detected region; then, in the real people stream detecting environment, actual height of each detected point in the detected region is obtained. A height lattice image of each point of the detected region is divided into a plurality of layers; in each layer, the lattice image undergoes classification and collection by adopting a self-iteration organization analyzing technique, and a target in each layer is separated. With a start from the first layer, the targets of the layers are accumulated or combined in an up-down manner, and a pedestrian target in each frame depth image is detected. A multiframe image target is tracked so as to finally achieve the aim of detecting information such as density, speed, flow rate, and the like of the people stream in the detected region in real time. When the external detection environment changes, the method can still effectively detect the people stream; in addition, the method also has the advantages of convenient and simple calculation, good instantaneity, synchronous people stream monitoring, and the like.
Description
Technical field
The invention belongs to Flame Image Process, mode identification technology, relate to the people stream detecting method of large-scale public arenas such as can being widely used in railway station, airport, subway station, large-scale conference and exhibition center and walkway.
Background technology
Along with economic growth, large-scale movement of population in quickening of urbanization process and the city or between the city, cause some large-scale public places such as railway station, airport, subway station, local people's current density such as walkway sharply increases, for effective public administration in these public service places has brought huge pressure.Along with reaching its maturity of real-time monitoring technique, electric detective technology, mechanics of communication and information science technology, in real time stream of people's information monitoring and development of Management System be applied in the public management field and come into one's own just day by day, and demonstrate huge economic and social benefit.Real-time stream of people's information monitoring in public domain and management system support and information platform for public management provides important techniques; It in real time, the state of the reflection monitoring stream of people and public place pedestrian traffic stream dynamically, be people streams in public places monitoring and dynamic guiding, stream of people's analysis of causes of blocking up, pedestrian traffic stream is induced decision-making, making rational planning for of walkway provides important evidence.The detection of people streams in public places information (flow, density etc.) is a gordian technique in real-time stream of people's information monitoring in public domain and the management system, and obtaining of it can be the public domain public management direct, the most reliable foundation is provided.Yet detection and detection means to the real-time people's stream information in public place are closely bound up, and different stream of people's detection meanss has determined diverse ways, precision and the suitable condition that information is obtained.
Find through literature search prior art, Halvorson, G.A etc. point out in its Master's thesis (" Automated Real-Time Dimension; Measurement of Moving Vehicles UsingInfrare; Laser Rangefinders ") (" automatically real-time yardstick: infrared; laser range finder is in the application of moving vehicle detection ") literary composition, the target object detection means mainly relies on Video Detection and two kinds of technology of infrared detection at present, these two kinds of technology are widely used in the tracking and the detection of various specific objectives, its biggest advantage is can be real-time, intuitively, accurately the object in the surveyed area is carried out effectively, detect fast, the research of related algorithm at present is comparative maturity also, and universality is stronger.Its weak point is for this bulk area of intensive pedestrian smaller, and move and the target object of irregularities effectively detects poor-performing, in addition, these two kinds of methods seriously rely on external conditions such as weather and illumination, in case external permission testing conditions changes, detecting effect will have a greatly reduced quality, and become the important bottleneck that these two kinds of methods of restriction are used in some specific environments.
Summary of the invention
The objective of the invention is at the deficiencies in the prior art, proposed a kind of people stream detecting method based on laser sensor depth image.In the actual persons stream information detects, this method has solved the problem that detection means relies on weather, illumination and detected object size in Video Detection and infrared detection at all, and this method still can effectively detect pedestrian's stream when external testing environment changes.In addition, this method also have calculate easy, real-time good, realize the stream of people's advantages such as monitoring function synchronously.
For reaching above purpose, solution of the present invention is:
A kind of people stream detecting method based on laser sensor depth image, comprise: at first take the depth image of multiframe when surveyed area is vacant by the laser sensing video camera, obtain the topomap on surveyed area ground by the mode of landform mapping, then under actual stream of people's testing environment, the stream of people's depth image deducts the true altitude on each ground in the topomap in the use actual scene, obtains the true altitude of each check point of surveyed area.Next, true altitude scope according to the surveyed area each point, by a plurality of threshold values surveyed area each point height dot matrix image is divided into a plurality of figure layers, each figure layer is adopted respectively from iteration fabric analysis technology dot matrix image is carried out cluster, be partitioned into the target in each figure layer.Then from first figure layer, adopt top-down mode that the target in each figure layer is added up or merge, detect the pedestrian's target in each frame depth image.By Kalman filtering the multiple image target is followed the tracks of at last, finally reach the real-time purpose that detects of the expert stream of people's of surveyed area information such as density, speed and flow.
Further, collecting device is installed: the laser sensing camera of use comprises sensing head, control module and man-machine interaction parts.Sensing head is exactly a detecting unit, and its three-dimensional dimension is W550mm*D155mm*H150mm, in case be fixed on the monitoring point, sensing head can be rotated at level, vertical both direction, and the maximum angle that allows of level is 150 degree, and the vertical maximum angle that allows is 30 degree.When to 3 d object scanning, the sensor data comprise scanning angle, pendulum angle and depth data.In addition, because detection variable is a polar coordinate system, so raw data must be converted into cartesian coordinate system through necessary processing.
Landform mapping: because of what need in the Cluster Classification method is the height of stream of people's individuality, so must adopt the mode of landform mapping to obtain the Terrain Elevation information of surveyed area.Because the reason of working mechanism, the sweep trace of laser sensing camera is fixed in the rectangular net point unlike common CCD camera, therefore need repeatedly scan idle area, the height value of each point of accumulative total surveyed area, balancedly be assigned to and obtain real floor level image on the default net point, the method that adopts function to fit at last obtains the surveyed area topomap.
Many Threshold Segmentation, cluster: by original depth-map being looked like carry out the landform mapping Filtering Processing, each detects the height image of a surface can to get the surveyed area stream of people, in the image gray scale representation of each pixel should detect a height of surface point.The present invention adopts a plurality of height thresholds that close-packed lattice is cut apart, obtain a plurality of height grade figure layers, dot matrix in each figure layer adopted respectively from iteration fabric analysis algorithm carry out cluster, identify object in each figure layer, then from first figure layer, target in the All Layers is added up or merges processing, finally obtain the pedestrian's target in each frame depth image.
Pedestrian's volume tracing: after detecting the pedestrian's target in each frame depth image, will carry out target following with that, obtain information such as stream of people's average velocity, flow.The present invention adopts the Kalman filter tracking mode to carry out corresponding Tracking Recognition to detecting individuality, and Kalman filter is a set of equations, makes process square error minimum come the state of estimation procedure with effective alternative manner.For tracking pedestrians accurately, each pedestrian has been built 3 Kalman filter, wherein two be on the x direction and the y direction on information in grid, another one is the individual square frame scaling of pedestrian in the image.
Owing to adopted such scheme, the present invention to have following characteristics: the present invention adopts the people stream detecting method based on laser sensor depth image to have lot of advantages.What at first the present invention adopted is the laser sensing camera, and this equipment can not be subjected to the restriction continuous firing of external conditions such as weather, light, has overcome the shortcoming that traditional video and infrared detection means are subjected to factor affecting such as weather, illumination; Secondly, the present invention has used the landform mapping mode based on the function match, has saved storage resources, and only increases a Filtering Processing when computing, has improved data processing speed; Once more, what the present invention adopted detected depth image is the method for many Threshold Segmentation, this method eliminated the threshold value that exists during single threshold is cut apart select difficult, detect the not congruent problem of target, improved the accuracy that detects target greatly; At last, system of the present invention is easy to administer and maintain, the whole stream of people based on the laser depth image detect mainly by the laser sensing camera and the auxiliary electronic equipment that are erected at the walkway top forms, integrated level height, image transmit and can utilize wired or wireless channel.Like this total system be modular construction, volume little, be easy to install, use, and safeguard the normal traffic operation that also can not influence walkway.
Description of drawings
Fig. 1 is the schematic flow sheet of the embodiment of the invention
Fig. 2 is many Threshold Segmentation synoptic diagram of the embodiment of the invention.
Fig. 3 is the figure strata class schematic flow sheet of the embodiment of the invention.
Embodiment
The present invention is further illustrated below in conjunction with the accompanying drawing illustrated embodiment.
The present invention adopts laser sensing camera, identification and tracking pedestrians stream.The present invention at first takes the depth image of the vacant surveyed area of multiframe by the laser sensing video camera, obtain the topomap on surveyed area ground by the mode of landform mapping, then under actual stream of people's testing environment, the stream of people's depth image deducts the true altitude on each ground in the topomap in the use actual scene, obtains the true altitude of each check point of surveyed area.Next, true altitude scope according to the surveyed area each point, by a plurality of threshold values surveyed area each point height dot matrix image is divided into a plurality of figure layers, each figure layer is adopted respectively from iteration fabric analysis technology dot matrix image is carried out cluster, be partitioned into the target in each figure layer.Then from first figure layer, adopt top-down mode that the target in each figure layer is added up or merge, detect the pedestrian's target in each frame depth image.By Kalman filtering the multiple image target is followed the tracks of at last, finally reach the real-time purpose that detects of the expert stream of people's of surveyed area information such as density, speed and flow.
The inventive method comprises following step:
1, laser sensor camera collection depth image:
The laser sensor camera is different from common CCD camera, and the image that it is gathered is depth image rather than common gray level image.Pixel value on one amplitude deepness image and cam lens center are directly proportional to the distance that detects target.When the surface of pulse laser beam direct irradiation at testee, the visual detector that is installed on the laser sensor from the light of body surface scattering detects, counter can be measured light in real time and was issued to by the detected time of visual detector from laser diode digit time, and it is converted into the distance of testee point, thereby obtain the depth image of object to be detected point.Good robustness when the laser sensing camera not only has CCD camera collection image commonly used, and be not subjected to condition restriction such as outside weather and illumination, practicality is stronger.When to 3 d object scanning, the sensor data comprise scanning angle, pendulum angle and depth data.In addition, be polar coordinate system because detect data representation, so raw data must be converted into cartesian coordinate system through necessary processing.
2, landform mapping:
According to the testing mechanism principle, the laser sensing camera is detected to be the degree of depth of tested object point apart from the laser shooting head mirror heart, and in the detection classification to the stream of people, what need is each height that detects individual distance ground, the degree of depth that the height on each individual distance ground can be deducted ground by the degree of depth of this inspected object point obtains, so the landform depth information in tested zone is the condition precedent that is respectively detected individual elevation information, particularly the sort of uneven ground, this process is exactly a landform mapping.
Though can be by the vacant detection zone depth image of scanning, and will detect the data storage form of getting up and carry out landform mapping, but this mode is wasted storage resources, in practical operation, the function match is a kind of well-adapted not only easy but also landform mapping mode fast, and this method is being calculated when detecting individual height, fitting function can be used as a kind of wave filter, very convenient.
Ground shape function fitting method is described below, if function y has n independent variable, by m testing site (x
J1, x
J2, x
J3...., x
Jn), j=1,2 ... m so then has
Wherein u is the match factor, and its account form can be changed into:
Generally speaking, landform mapping generally adopts fitting function 2 times, therefore shape function H
Ground(x, y)=u
0+ u
1X+u
2Y+u
3x
2+ u
4y
2+ u
5Xy, u
jBe to fit the factor.X, y, x
2, y
2, xy is 5 independent variables, by aforementioned approximating method, and the topographic map that obtains fitting.
3, many Threshold Segmentation, cluster:
By original depth-map being looked like carry out the landform mapping Filtering Processing, each detects the height image of a surface can to get the surveyed area stream of people, in the image gray scale representation of each pixel should detect a height of surface point.Can obtain the maximal value and the minimum value of the height of surveyed area target object dot matrix by statistics.If only adopt single height threshold to cut apart, exist threshold value to select the problem of difficulty, if select too much to this image, will the loss detection target, if but select too smallly, detect dot matrix with dense distribution, will seriously strengthen the difficulty of target object clustering algorithm.The clustering algorithm of close-packed lattice still is a problem that is difficult to solution at present, therefore the present invention adopts a plurality of height thresholds that close-packed lattice is cut apart, obtain a plurality of height grade figure layers, dot matrix in each figure layer adopted respectively from iteration fabric analysis algorithm carry out cluster, identify object in each figure layer, then from first figure layer, the target in the All Layers is added up or merges processing, finally obtain the pedestrian's target in each frame depth image.
4, the individual track algorithm of pedestrian
In stream of people's supervisory system, target following plays crucial effects.After detecting the pedestrian's target in each frame depth image, to carry out target following with that, obtain information such as stream of people's average velocity, flow.The present invention adopts the Kalman filter tracking mode to carry out corresponding Tracking Recognition to detecting individuality, and Kalman filter is a set of equations, makes process square error minimum come the state of estimation procedure with effective alternative manner.For tracking pedestrians accurately, each pedestrian has been built 3 Kalman filter, wherein two be on the x direction and the y direction on information in grid, another one is the individual square frame scaling of pedestrian in the image.
Present embodiment adopts the stream of people's detection scheme based on laser sensor depth image shown in Figure 1, and concrete implementation step is as follows:
1, the hardware collecting device is installed
The laser sensing camera that the present invention uses comprises sensing head, control module and man-machine interaction parts (monitor, keyboard and mouse).Sensing head is exactly a detecting unit, and its three-dimensional dimension is W550mm*D155mm*H150mm, in case be fixed on the monitoring point, sensing head can be rotated at level, vertical both direction, and the maximum angle that allows of level is 150 degree, and the vertical maximum angle that allows is 30 degree.When to 3 d object scanning, the sensor data comprise scanning angle, pendulum angle and depth data.In addition, because detection variable is a polar coordinate system, so raw data must be converted into cartesian coordinate system through necessary processing.
2, landform mapping
As foregoing, because what need in the Cluster Classification method is the height of stream of people's individuality, so must adopt the mode of landform mapping to obtain the Terrain Elevation information of surveyed area.Because the reason of working mechanism, the sweep trace of laser sensing camera is fixed in the rectangular net point unlike common CCD camera, therefore need repeatedly scan idle area, the height value of each point of accumulative total surveyed area, balancedly be assigned to and obtain real floor level image on the default net point, the method that adopts function to fit at last obtains the surveyed area topomap.
3, many Threshold Segmentation, cluster
After surveyed area stream of people depth image is carried out landform filtering, just can obtain the surveyed area stream of people's highly dense dot matrix, for each vertical frame dimension degree close-packed lattice, the present invention chooses a plurality of height thresholds to be cut apart close-packed lattice, obtain the dot chart layer of many differing heights rate ranges, dot matrix employing to each figure layer is carried out cluster from iteration fabric analysis algorithm, identifies pedestrian's individuality in each figure layer (as Fig. 2, Fig. 3).After this, the present invention is since the 1st figure layer, and setting the pedestrian who identifies individual is intermediate result, and the object that identifies in this intermediate result and the second layer is compared, if appear at simultaneously on two figure layers at same position object appearing object, then show it is same a group traveling together; If do not have to occur and exist in the second figure layer at the same position first figure layer, explanation is a new target, adds intermediate result; If occur, but do not find that at the second figure layer explanation is a flase drop target, deletes from middle result at the first figure layer; If a plurality of pedestrian's targets in that the first figure layer occurs become pedestrian's target on the second figure layer, just will detect target and permeate individually, and in intermediate result, delete processing.Repeat said process figure layer end to the last, all identifications that finally obtain in the single frames height image are individual.
4, pedestrian's volume tracing
In stream of people's supervisory system, target following plays crucial effects.After detecting the pedestrian's target in each frame depth image, to carry out target following with that, obtain information such as stream of people's average velocity, flow.The present invention adopts the Kalman filter tracking mode to carry out corresponding Tracking Recognition to detecting individuality, and Kalman filter is a set of equations, makes process square error minimum come the state of estimation procedure with effective alternative manner.For tracking pedestrians accurately, each pedestrian has been built 3 Kalman filter, wherein two be on the x direction and the y direction on information in grid, another one is the individual square frame scaling of pedestrian in the image.
From above use-case, can find, present embodiment has overcome the dependence to weather and illumination condition of traditional video, infrared detection means effectively, changed checkout equipment into the laser sensing camera, be erected on the passage that the pedestrian passes through and put the Dynamic Recognition and the detection that just can realize the stream of people, in addition, this detection means also have calculate easy, fast operation, be easy to safeguard, advantages such as monitoring in real time.
The above-mentioned description to embodiment is can understand and apply the invention for ease of those skilled in the art.The person skilled in the art obviously can easily make various modifications to these embodiment, and needn't pass through performing creative labour being applied in the General Principle of this explanation among other embodiment.Therefore, the invention is not restricted to the embodiment here, those skilled in the art should be within protection scope of the present invention for improvement and modification that the present invention makes according to announcement of the present invention.
Claims (10)
1, a kind of people stream detecting method based on laser sensor depth image is characterized in that:
Take the depth image of multiframe when surveyed area is vacant by the laser sensing video camera, obtain the topomap on surveyed area ground by the mode of landform mapping;
Under actual stream of people's testing environment, the stream of people's depth image deducts the true altitude on each ground in the topomap in the use actual scene, obtains the true altitude of each check point of surveyed area;
True altitude scope according to the surveyed area each point, by a plurality of threshold values surveyed area each point height dot matrix image is divided into a plurality of figure layers, each figure layer adopted respectively from iteration fabric analysis technology dot matrix image is carried out cluster, be partitioned into the target in each figure layer; From first figure layer, adopt top-down mode that the target in each figure layer is added up or merge, detect the pedestrian's target in each frame depth image;
By Kalman filtering the multiple image target is followed the tracks of, finally reach the real-time purpose that detects of the expert stream of people's of surveyed area information such as density, speed and flow.
2, the people stream detecting method based on laser sensor depth image according to claim 1 is characterized in that: the laser sensing camera of use comprises sensing head, control module and man-machine interaction parts; Sensing head is a detecting unit, in case be fixed on the monitoring point, sensing head can be rotated at level, vertical both direction.
3, the people stream detecting method based on laser sensor depth image according to claim 2 is characterized in that: it is 150 degree that this sensing head horizontally rotates the maximum angle that allows, and the maximum angle that allows of vertical rotation is 30 degree.
4, the people stream detecting method based on laser sensor depth image according to claim 1 is characterized in that: when to 3 d object scanning, the sensor data comprise scanning angle, pendulum angle and depth data; Detected raw data under the polar coordinate system is converted into data under the cartesian coordinate system.
5, the people stream detecting method based on laser sensor depth image according to claim 1 is characterized in that: the mode of employing landform mapping obtains the Terrain Elevation information of surveyed area; Idle area is repeatedly scanned, and the height value of each point of accumulative total surveyed area balancedly is assigned to and obtains real floor level image on the default net point, and the method that adopts function to fit at last obtains the surveyed area topomap.
6, the people stream detecting method based on laser sensor depth image according to claim 1, it is characterized in that: by original depth-map being looked like carry out the landform mapping Filtering Processing, each detects the height image of a surface can to get the surveyed area stream of people, in the image gray scale representation of each pixel should detect a height of surface point; Adopt a plurality of height thresholds that close-packed lattice is cut apart, obtain a plurality of height grade figure layers, dot matrix in each figure layer adopted respectively from iteration fabric analysis algorithm carry out cluster, identify object in each figure layer, then from first figure layer, target in the All Layers is added up or merges processing, finally obtain the pedestrian's target in each frame depth image.
7, the people stream detecting method based on laser sensor depth image according to claim 1 is characterized in that: after detecting the pedestrian's target in each frame depth image, then carry out target following to obtain information such as stream of people's average velocity, flow.
8, the people stream detecting method based on laser sensor depth image according to claim 7 is characterized in that: carry out corresponding Tracking Recognition to detecting individuality, make process square error minimum come the state of estimation procedure with effective alternative manner.
9, the people stream detecting method based on laser sensor depth image according to claim 8 is characterized in that: adopt the Kalman filter tracking mode to carry out corresponding Tracking Recognition to detecting individuality.
10, the people stream detecting method based on laser sensor depth image according to claim 9, it is characterized in that: each pedestrian is built 3 Kalman filter, wherein two be on the x direction and the y direction on information in grid, another one is the individual square frame scaling of pedestrian in the image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007100475903A CN101425128A (en) | 2007-10-30 | 2007-10-30 | People stream detecting method based on laser sensor depth image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2007100475903A CN101425128A (en) | 2007-10-30 | 2007-10-30 | People stream detecting method based on laser sensor depth image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101425128A true CN101425128A (en) | 2009-05-06 |
Family
ID=40615737
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2007100475903A Pending CN101425128A (en) | 2007-10-30 | 2007-10-30 | People stream detecting method based on laser sensor depth image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101425128A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102074059A (en) * | 2010-12-13 | 2011-05-25 | 北京北大千方科技有限公司 | Method and equipment for improving laser-based passenger flow detection accuracy by using decision sequence |
| CN102521646A (en) * | 2011-11-11 | 2012-06-27 | 浙江捷尚视觉科技有限公司 | Complex scene people counting algorithm based on depth information cluster |
| CN102542566A (en) * | 2010-12-08 | 2012-07-04 | 微软公司 | Orienting the position of a sensor |
| CN102622578A (en) * | 2012-02-06 | 2012-08-01 | 中山大学 | Passenger counting system and passenger counting method |
| CN103198327A (en) * | 2013-04-25 | 2013-07-10 | 重庆师范大学 | Open type scene-oriented entrance and exit population counting method and device |
| CN105628951A (en) * | 2015-12-31 | 2016-06-01 | 北京小孔科技有限公司 | Method and device for measuring object speed |
| CN107016350A (en) * | 2017-04-26 | 2017-08-04 | 中科唯实科技（北京）有限公司 | A kind of Falls Among Old People detection method based on depth camera |
| CN107016349A (en) * | 2017-03-10 | 2017-08-04 | 中科唯实科技（北京）有限公司 | A kind of crowd's flow analysis method based on depth camera |
| CN107264797A (en) * | 2016-04-06 | 2017-10-20 | 成都积格科技有限公司 | Crowd massing early warning unmanned plane |
| CN110806588A (en) * | 2019-10-17 | 2020-02-18 | 北醒（北京）光子科技有限公司 | Pedestrian flow detection system based on laser radar |
-
2007
- 2007-10-30 CN CNA2007100475903A patent/CN101425128A/en active Pending
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542566A (en) * | 2010-12-08 | 2012-07-04 | 微软公司 | Orienting the position of a sensor |
| CN102542566B (en) * | 2010-12-08 | 2014-08-20 | 微软公司 | Orienting the position of a sensor |
| CN102074059B (en) * | 2010-12-13 | 2013-05-22 | 北京北大千方科技有限公司 | Method and equipment for improving laser-based passenger flow detection accuracy by using decision sequence |
| CN102074059A (en) * | 2010-12-13 | 2011-05-25 | 北京北大千方科技有限公司 | Method and equipment for improving laser-based passenger flow detection accuracy by using decision sequence |
| CN102521646A (en) * | 2011-11-11 | 2012-06-27 | 浙江捷尚视觉科技有限公司 | Complex scene people counting algorithm based on depth information cluster |
| CN102521646B (en) * | 2011-11-11 | 2015-01-21 | 浙江捷尚视觉科技股份有限公司 | Complex scene people counting algorithm based on depth information cluster |
| CN102622578A (en) * | 2012-02-06 | 2012-08-01 | 中山大学 | Passenger counting system and passenger counting method |
| CN103198327B (en) * | 2013-04-25 | 2016-12-28 | 重庆师范大学 | Gateway demographic method and device towards open scene |
| CN103198327A (en) * | 2013-04-25 | 2013-07-10 | 重庆师范大学 | Open type scene-oriented entrance and exit population counting method and device |
| CN105628951A (en) * | 2015-12-31 | 2016-06-01 | 北京小孔科技有限公司 | Method and device for measuring object speed |
| US20170193310A1 (en) * | 2015-12-31 | 2017-07-06 | Pinhole (Beijing) Technology Co., Ltd. | Method and apparatus for detecting a speed of an object |
| US10289918B2 (en) | 2015-12-31 | 2019-05-14 | Pinhole (Beijing) Technology Co., Ltd. | Method and apparatus for detecting a speed of an object |
| CN105628951B (en) * | 2015-12-31 | 2019-11-19 | 北京迈格威科技有限公司 | The method and apparatus of speed for measurement object |
| CN107264797A (en) * | 2016-04-06 | 2017-10-20 | 成都积格科技有限公司 | Crowd massing early warning unmanned plane |
| CN107264797B (en) * | 2016-04-06 | 2019-12-17 | 成都积格科技有限公司 | Crowd gathers early warning unmanned aerial vehicle |
| CN107016349A (en) * | 2017-03-10 | 2017-08-04 | 中科唯实科技（北京）有限公司 | A kind of crowd's flow analysis method based on depth camera |
| CN107016349B (en) * | 2017-03-10 | 2020-11-06 | 中科唯实科技（北京）有限公司 | Crowd flow analysis method based on depth camera |
| CN107016350A (en) * | 2017-04-26 | 2017-08-04 | 中科唯实科技（北京）有限公司 | A kind of Falls Among Old People detection method based on depth camera |
| CN110806588A (en) * | 2019-10-17 | 2020-02-18 | 北醒（北京）光子科技有限公司 | Pedestrian flow detection system based on laser radar |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101425128A (en) | People stream detecting method based on laser sensor depth image | |
| CN108710875B (en) | A kind of take photo by plane road vehicle method of counting and device based on deep learning | |
| Guan et al. | Use of mobile LiDAR in road information inventory: A review | |
| CN106570454B (en) | Pedestrian traffic parameter extracting method based on mobile laser scanning | |
| CN104005325B (en) | Based on pavement crack checkout gear and the method for the degree of depth and gray level image | |
| CN103714363B (en) | A kind of motor vehicle exhaust smoke video identification system | |
| CN103456172B (en) | A kind of traffic parameter measuring method based on video | |
| CN109644360B (en) | Method for distinguishing pedestrian flow direction by using WI-FI probe | |
| CN105812733B (en) | A kind of scene monitoring guiding system of air traffic control | |
| Guan et al. | Automated road information extraction from mobile laser scanning data | |
| CN104143194B (en) | A kind of point cloud segmentation method and device | |
| CN105427619B (en) | Vehicle following distance automatic production record and method | |
| CN101514993B (en) | Vehicle speed measurement device based on linear array CCD camera | |
| CN103500338A (en) | Road zebra crossing automatic extraction method based on vehicle-mounted laser scanning point cloud | |
| CN108960060A (en) | A kind of automatic driving vehicle pavement texture identifying system and method | |
| CN101975951A (en) | Field environment barrier detection method fusing distance and image information | |
| CN102682602B (en) | Collecting method for road traffic parameters based on video technique | |
| CN113345237A (en) | Lane-changing identification and prediction method, system, equipment and storage medium for extracting vehicle track by using roadside laser radar data | |
| Zhang et al. | Background filtering and vehicle detection with roadside lidar based on point association | |
| CN106778655A (en) | A kind of entrance based on human skeleton is trailed and enters detection method | |
| CN109272482A (en) | A kind of urban road crossing vehicle queue detection system based on sequence image | |
| CN102789686B (en) | Road traffic flow detecting method based on road surface brightness composite mode recognition | |
| CN105184274A (en) | Method of obtaining passenger flow velocity and density parameters based on depth image | |
| CN108717540A (en) | The method and device of pedestrian and vehicle are distinguished based on 2D laser radars | |
| Lehtomäki et al. | Performance analysis of a pole and tree trunk detection method for mobile laser scanning data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20090506 |