CN112686082A - System for counting number of people in vehicle based on visible light polarization - Google Patents
System for counting number of people in vehicle based on visible light polarization Download PDFInfo
- Publication number
- CN112686082A CN112686082A CN201910990791.XA CN201910990791A CN112686082A CN 112686082 A CN112686082 A CN 112686082A CN 201910990791 A CN201910990791 A CN 201910990791A CN 112686082 A CN112686082 A CN 112686082A
- Authority
- CN
- China
- Prior art keywords
- people
- vehicle
- polarization
- subsystem
- counting
- 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.)
- Granted
Links
Images
Landscapes
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a counting system for the number of people in a vehicle based on visible light polarization. The data acquisition and information processing subsystem comprises a polarization module and an information processing module and is used for image acquisition and polarization information fusion vehicle-penetrating film processing; the lighting subsystem is used for supplementing light to the environment in the vehicle; the main control and people counting subsystem is used for detecting passengers and counting the number of people in the vehicle. The invention has high statistical accuracy and good detection effect on the number of people in the film-pasted vehicle.
Description
Technical Field
The invention belongs to the technical field of intelligent security and relates to a system for counting passengers in a film-coated vehicle.
Background
With the gradual development of economy and the gradual progress of science and technology, the living standard of people is continuously improved, automobiles are already used as important transportation tools for people to go out, with the development trend of the country, the centralized construction of certain enterprise parks also increases the difficulty of park access management, and important checkpoints or regions need to be checked for the condition of people entering the automobiles so as to ensure the safety. Taking the chemical industrial area of the sea city as an example, the land occupation area of the garden is 36.1 square kilometers, 153 enterprises are involved, more than 5 thousands of people come in and go out of more than ten roads, and vehicles come in and go out of the garden in the peak of going to and going to work, so that the parking can not be checked one by one. Therefore, the technical means for checking the number of people in the vehicle without stopping the vehicle is urgent, and safe and efficient people checking management is carried out on the entrance and exit of the vehicle certified in the park. The intelligent traffic light has the same significance for clear detection of people in passing vehicles in some important gates or special areas, and through detecting and analyzing the number, the postures, the behaviors and the like of people in running vehicles, whether the important gates and the special areas can pass or not can be prompted or warned.
The detection and identification of the personnel in the running vehicle are influenced by the film sticking of the vehicle, the motion of the vehicle and the external natural environment, so that the difficulty of the detection and identification is greatly increased. The automobile film is a multi-layer polyester composite film material with the functions of improving the optical performance, the safety performance and the like of automobile glass, and is a general name of a solar film, a sun-shading film, an explosion-proof film and the like. The heat insulating film for automobiles has attracted much attention because of its functions of heat insulation and ultraviolet ray insulation. The visible light and infrared reflection are increased, so that the transmittance of solar energy can be effectively reduced, and the purposes of energy conservation and heat insulation are achieved. However, in order to ensure the safety of driving by a driver and to ensure sufficient visible light transmittance, the national mandatory standards stipulate that the visible light transmittance of the front windshield glass of an automobile and the visual area range of the driver should be not less than 70%, and the visible light transmittance of the side windshield glass of an automobile should be not less than 60%. Therefore, blocking the radiation in the infrared band in the sunlight becomes a main means for reducing the temperature in the vehicle to achieve the heat insulation effect. In summary, the main transmission wavelength range of the window glass is 480-760 nm, and accordingly, if the person in the vehicle needs to be clearly detected, the camera should select the wavelength band for imaging.
At present, a main mode for detecting people in a vehicle is to add a sensor in the vehicle for perception, but the accuracy of the people can not be evaluated by further confirming the accuracy of the people outside, the detection outside is affected by glass film sticking and vehicle movement, the application of infrared, millimeter wave, ultrasonic waves and the like is limited, and the detection probability of the common visible light camera affected by film sticking glass, strong light and stray light is not ideal.
Disclosure of Invention
The invention aims to provide a statistical system for accurately counting the number of people in a film-pasted vehicle.
In order to solve the technical problem, the invention provides an in-vehicle people counting system based on visible light polarization, which adopts the following technical scheme:
the statistical system comprises a data acquisition and information processing subsystem, a master control and people counting subsystem and an illumination subsystem.
The data acquisition and information processing subsystem comprises a polarization module and an information processing module;
the information processing module judges whether the system is started or not and sends a synchronous signal to the polarization module and the illumination subsystem; carrying out polarization information fusion vehicle-penetrating film processing on the received image data, and sending the processed image to a main control and people counting subsystem in real time;
the polarization module collects a visible light image after receiving the synchronous signal;
the lighting subsystem receives the synchronous signal and supplements light to the environment in the vehicle;
the main control and people counting subsystem is used for detecting passengers and counting the number of people in the vehicle.
Further, in the above-mentioned case,
the polarization module adopts a polarization camera to collect images.
The polarization information fusion car-penetrating film processing method comprises the following steps: decomposing the image collected by the polarization camera into images in four different vibration directions of 0 degrees, 45 degrees, 90 degrees and 135 degrees according to the position of each pixel point of a 2 x 2 matrix, wherein the size of the images is 1/4 of the original image, extracting the images in the 90 degrees direction and sending the images to the main control and people counting subsystem.
The main control and people counting subsystem detects and counts the number of passengers in the vehicle by adopting a deep learning technology based on big data processing, and the method comprises the following steps:
firstly, a large amount of collected calibrated data is input into a deep network, and a human face and position model of people in the vehicle is obtained after network training. And then, when the algorithm is operated in real time, loading the pre-trained model and the image acquired in real time into an algorithm network, and judging the current image content by the network according to the model and outputting the detected positions and the number of the personnel.
The Cascade-RCNN series detection framework is matched with the HRnet network to serve as a main body structure of the algorithm network, meanwhile, the FPN is matched with the HRnet to conduct feature extraction, and a ROIAlign method is adopted to conduct feature map mapping and dividing.
Compared with the prior art, the invention has the beneficial effects that:
the invention adopts the visible light polarization detector and the polarization transparent film information processing technology to realize the detection of the mobile film sticking vehicle, compared with the light intensity imaging technology, the spectrum imaging technology and the like, the polarization imaging can obtain the information of other dimensions, the polarization device has the function of well removing background noise (strong light, stray light and the like), the difference between a target signal and a background clutter signal can be increased, and the target profile in the imaging is enhanced.
The invention realizes the detection and the people counting of the personnel in the vehicle based on the deep learning technology of the big data, and ensures the accuracy of the detection effect.
The invention has the characteristics of low power consumption, less time consumption, high personnel detection definition in the film-pasted vehicle, high people counting probability and the like, and can meet the requirements on functionality, real-time property and adaptability in industrial application.
Drawings
The figures of the invention are 3 in total.
FIG. 1 is a schematic diagram of a system for counting people in a vehicle according to an embodiment of the invention;
FIG. 2 is a schematic diagram of an installation layout of an in-vehicle occupant detection system according to an embodiment of the present invention;
fig. 3 is a front passenger seat polarization transparent film processed image according to an embodiment of the present invention.
Detailed Description
The technical solution of the present invention will be further described in detail with reference to the accompanying drawings and examples.
In an embodiment, the statistical system includes a data collection and information processing subsystem, a main control and people counting subsystem, and a lighting subsystem, and fig. 1 is referred in detail.
The system layout method refers to the installation layout schematic diagram of the personnel detection system in the vehicle in fig. 2 in detail. And data acquisition and information processing subsystems are erected on two sides of the bayonet, namely in front of the front copilot. The data acquisition and information processing subsystem comprises a polarization module and an information processing module, is used for acquiring images, performing polarization information fusion vehicle-penetrating film processing, and transmitting the processed images to the main control and people counting subsystem in real time.
The polarization module adopts a polarization camera to collect images.
And erecting an illumination subsystem at the position of the vehicle collision line for supplementing light to the environment in the vehicle.
The main control and people counting subsystem is arranged on one side of the front or the secondary driving seat and used for detecting passengers and counting the number of people in the vehicle.
In the embodiment, the automatic rod lifting vehicle release system further comprises a bayonet comprehensive control system, and whether the rod lifting vehicle is released or not is judged by the bayonet comprehensive control system.
The working process of the embodiment is as follows:
and (1) installing the hardware system in place, in a standby state, and waiting for passing vehicles.
And (2) after the front and secondary driving bit data acquisition and information processing system receives a traffic vehicle line collision signal, the system starts to work. The starting work mainly comprises the following contents:
firstly, an information processing module in a data acquisition and information processing subsystem sends a synchronous signal to lighting systems on two sides of a front and auxiliary driver seat to start lighting;
and secondly, the information processing module in the data acquisition and processing subsystem sends a synchronous signal to the polarization camera, and the polarization camera at the front and the auxiliary driving positions is started to shoot and acquire images.
And (3) after receiving the image data, the data acquisition and information processing subsystem information processing module performs polarization information fusion vehicle-penetrating film processing, and sends the processed image to the main control and people counting subsystem in real time, wherein the image processed by the front-and-assistant driving position polarization vehicle-penetrating film processing is shown in detail in fig. 3, the image in fig. 3(a) is a driving position image, and the image in fig. 3(b) is an assistant driving position image.
And (4) carrying out in-vehicle passenger detection and people counting on the transparent film images on the two sides of the front passenger seat by the main control and people counting subsystem.
Firstly, a main control and people counting subsystem inquires data acquisition and information processing subsystems in real time to send data signals, inquires the data sending signals, and starts to receive data of the data acquisition and information processing subsystems on the two sides of the front and auxiliary driver positions;
and secondly, after the data are received, the main control and people counting subsystem detects and counts the number of passengers in the vehicle for the transparent film images on the two sides of the front passenger seat.
And (5) after the main control and people counting subsystem finishes the counting of the number of people in the vehicle, the number of people and the evidence obtaining image are transmitted to the comprehensive bayonet control system, and the comprehensive bayonet control system judges whether the rod raising vehicle is released or not.
In the embodiment, the processing method of the polarization information fusion car-through film comprises the following steps: decomposing the image collected by the polarization camera into images in four different vibration directions of 0 degrees, 45 degrees, 90 degrees and 135 degrees according to the position of each pixel point of a 2 x 2 matrix, wherein the size of the images is 1/4 of the original image, extracting the images in the 90 degrees direction and sending the images to the main control and people counting subsystem.
In a preferred embodiment, the main control and people counting subsystem adopts deep learning technology based on big data processing to detect and count people in the vehicle.
Firstly, a large amount of collected calibrated data is input into a deep network, and a human face and position model of people in the vehicle is obtained after network training. And then, when the algorithm is operated in real time, loading the pre-trained model and the image acquired in real time into an algorithm network, and judging the current image content by the network according to the model and outputting the detected positions and the number of the personnel.
In some embodiments, the Cascade-RCNN series detection framework is matched with the HRnet network to serve as a main body structure of the algorithm network, and meanwhile, the FPN and ROIAlign methods are adopted to improve reliability and detection effect, so that personnel position detection and personnel number counting are finally achieved.
Cascade-RCNN is selected as a main detection framework: the method is characterized in that the method comprises the following steps of training a general single threshold value, training a trained detector with a limited effect, and training the next Stage by using the output of one Stage by adopting a Cascade Stage method in order to ensure the high quality of a result and not reduce training samples. Different thresholds are set, and the higher the threshold is, the better the network has an effect on the candidate box with higher accuracy. Through threshold setting, the trained network has an optimization effect on the input suggestion region.
Selecting HRnet as a feature extraction network: compared to most existing methods that pass input through a network and then improve resolution (e.g., by dilation-convolution, etc.), HRnet is able to maintain high-resolution characterization throughout the process. Starting from the first phase with high resolution subnets, progressively more phases are formed of high resolution to low resolution subnets, and the multi-resolution subnets are connected in parallel. By performing multi-scale fusion for multiple times in the whole process, each high-resolution representation repeatedly receives information from other parallel representations, so that abundant high-resolution representations are obtained.
Selecting FPN to match HRnet for feature extraction: the method is mainly used for solving the defects of the RCNN algorithm in processing multi-scale change, and the FPN method structure constructs a hierarchical structure from top to bottom and with lateral connection to construct high-level semantic features of all scales. The FPN can be used as a universal feature extractor and matched with a detection algorithm.
And (3) selecting a ROIAlign method to map and divide the characteristic diagram: and no rounding operation is performed during feature map mapping and feature map division, and finally non-integer pixel points can be sampled in a double linear interpolation mode, so that the regression position of the feature map is not influenced by rounding.
Claims (10)
1. A system for counting the number of people in a vehicle based on visible light polarization is characterized in that,
the statistical system comprises a data acquisition and information processing subsystem, a main control and people counting subsystem and an illumination subsystem;
the data acquisition and information processing subsystem comprises a polarization module and an information processing module,
the information processing module judges whether the system is started or not and sends a synchronous signal to the polarization module and the illumination subsystem; carrying out polarization information fusion vehicle-penetrating film processing on the received image data, and sending the processed image to a main control and people counting subsystem in real time;
the polarization module collects a visible light image after receiving the synchronous signal;
after the lighting subsystem receives the synchronous signal, the lighting subsystem supplements light to the environment in the vehicle;
the main control and people counting subsystem is used for detecting passengers and counting the number of people in the vehicle.
2. The visible light polarization-based vehicle occupant counting system of claim 1, wherein the polarization module employs a polarization camera to collect images.
3. The system for counting the number of people in the vehicle based on the polarization of the visible light, according to claim 1, wherein the polarization information fusion vehicle-penetrating film processing method comprises:
decomposing the image collected by the polarization camera into images in four different vibration directions of 0 degrees, 45 degrees, 90 degrees and 135 degrees according to the position of each pixel point of a 2 x 2 matrix, wherein the size of the images is 1/4 of the original image, extracting the images in the 90 degrees direction and sending the images to the main control and people counting subsystem.
4. The visible light polarization-based system of claim 1, wherein the sub-system of the main control and people counting employs a deep learning technique based on big data processing to detect and count people in the vehicle.
5. The system according to claim 4, wherein the system is characterized in that a depth network is input by using a large amount of collected calibrated data, a face and position model of the people in the vehicle is obtained after network training, then, when the algorithm is operated in real time, the model obtained through pre-training and the image collected in real time are loaded into the algorithm network, and the network judges the current image content according to the model and outputs the detected positions and the number of the people.
6. The system of claim 5, wherein the Cascade-RCNN series detection framework and the HRnet network are used as main structures of the algorithm network, the FPN and the HRnet are used for feature extraction, and a ROIAlign method is used for feature map mapping and division.
7. The system according to claim 6, wherein the Cascade-RCNN is used as a main detection framework: and (3) training the next Stage by using the output of one Stage by adopting a Cascade Stage method, setting different thresholds, and optimizing the input suggestion region through the trained network.
8. The system of claim 6, wherein the HRnet is used as a feature extraction network: starting from the first phase with high resolution subnets, progressively more phases are formed of high resolution to low resolution subnets, and the multi-resolution subnets are connected in parallel. By performing multi-scale fusion for multiple times in the whole process, each high-resolution representation repeatedly receives information from other parallel representations, and abundant high-resolution representations are obtained.
9. The system of claim 6, wherein the FPN is used for extracting the characteristics in cooperation with the HRnet: a hierarchical structure with lateral connection from top to bottom is constructed by using an FPN method structure to construct high-level semantic features of all scales.
10. The system of claim 6, wherein the ROIAlign method is used for mapping and dividing the feature map: when the feature map mapping and the feature map dividing are carried out, a double linear interpolation mode is adopted to realize the sampling of non-integer pixel points.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990791.XA CN112686082B (en) | 2019-10-18 | 2019-10-18 | In-vehicle people counting system based on visible light polarization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990791.XA CN112686082B (en) | 2019-10-18 | 2019-10-18 | In-vehicle people counting system based on visible light polarization |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112686082A true CN112686082A (en) | 2021-04-20 |
| CN112686082B CN112686082B (en) | 2023-07-14 |
Family
ID=75444722
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910990791.XA Active CN112686082B (en) | 2019-10-18 | 2019-10-18 | In-vehicle people counting system based on visible light polarization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112686082B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001095247A1 (en) * | 2000-06-08 | 2001-12-13 | Honeywell International Inc. | Near-infrared disguise detection |
| US20130027557A1 (en) * | 2011-07-29 | 2013-01-31 | Ricoh Company, Ltd. | Detection apparatus and method |
| US20160061948A1 (en) * | 2013-06-06 | 2016-03-03 | United States Of America, As Represented By The Secretary Of The Army | Moving Multi-Polarization Multi-Transmitter/Receiver Ground Penetrating Radar System and Signal Processing for Buried Target Detection |
| CN107525571A (en) * | 2017-09-05 | 2017-12-29 | 东南大学 | A kind of automobile intelligent overload detection system |
| WO2018165027A1 (en) * | 2017-03-06 | 2018-09-13 | Polaris Sensor Technologies, Inc. | Polarization-based detection and mapping method and system |
| CN108960107A (en) * | 2018-06-25 | 2018-12-07 | 安徽百诚慧通科技有限公司 | A kind of overcrowding recognition methods of small mini van and device |
| CN110111565A (en) * | 2019-04-18 | 2019-08-09 | 中国电子科技网络信息安全有限公司 | A kind of people's vehicle flowrate System and method for flowed down based on real-time video |
-
2019
- 2019-10-18 CN CN201910990791.XA patent/CN112686082B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2001095247A1 (en) * | 2000-06-08 | 2001-12-13 | Honeywell International Inc. | Near-infrared disguise detection |
| US20130027557A1 (en) * | 2011-07-29 | 2013-01-31 | Ricoh Company, Ltd. | Detection apparatus and method |
| US20160061948A1 (en) * | 2013-06-06 | 2016-03-03 | United States Of America, As Represented By The Secretary Of The Army | Moving Multi-Polarization Multi-Transmitter/Receiver Ground Penetrating Radar System and Signal Processing for Buried Target Detection |
| WO2018165027A1 (en) * | 2017-03-06 | 2018-09-13 | Polaris Sensor Technologies, Inc. | Polarization-based detection and mapping method and system |
| CN107525571A (en) * | 2017-09-05 | 2017-12-29 | 东南大学 | A kind of automobile intelligent overload detection system |
| CN108960107A (en) * | 2018-06-25 | 2018-12-07 | 安徽百诚慧通科技有限公司 | A kind of overcrowding recognition methods of small mini van and device |
| CN110111565A (en) * | 2019-04-18 | 2019-08-09 | 中国电子科技网络信息安全有限公司 | A kind of people's vehicle flowrate System and method for flowed down based on real-time video |
Non-Patent Citations (2)
| Title |
|---|
| FARZAN ERLIK NOWRUZI等: "In-Vehicle Occupancy Detection with Convolutional Networks on Thermal Images" * |
| 黄 雯等: "一种基于多分类器融合的人数统计算法" * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112686082B (en) | 2023-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104442924B (en) | All-weather high speed railway vehicle-mounted obstacle detection system and method | |
| CN112581612B (en) | Vehicle-mounted grid map generation method and system based on fusion of laser radar and all-round-looking camera | |
| CN109466442B (en) | School bus student detention alarm system | |
| CN103034843B (en) | Method for detecting vehicle at night based on monocular vision | |
| CN103902976A (en) | Pedestrian detection method based on infrared image | |
| CN105260712A (en) | Method and system for detecting pedestrian in front of vehicle | |
| CN115605777A (en) | Dynamic target point cloud rapid identification and point cloud segmentation method based on road side sensing unit | |
| CN104392629A (en) | Method and device for detecting car distance | |
| CN104952060A (en) | Adaptive segmentation extraction method for infrared pedestrian region of interests | |
| CN110610153A (en) | Lane recognition method and system for automatic driving | |
| DE102020104695A1 (en) | DETECTION OF BLOCKING OF LIGHTING DEVICES | |
| CN109190488A (en) | Front truck car door opening detection method and device based on deep learning YOLOv3 algorithm | |
| CN104407346A (en) | Mobile runway foreign object debris (FOD) monitoring method based on information integration | |
| WO2021145134A1 (en) | Light receiving device, signal processing method for light receiving device, and ranging device | |
| CN112381014A (en) | Illegal parking vehicle detection and management method and system based on urban road | |
| CN103679214A (en) | Vehicle detection method based on online area estimation and multi-feature decision fusion | |
| CN111161160A (en) | Method and device for detecting obstacle in foggy weather, electronic equipment and storage medium | |
| CN117372979A (en) | Road inspection method, device, electronic equipment and storage medium | |
| Qu et al. | Human-vehicle collision detection algorithm based on image processing | |
| CN114445798A (en) | Urban road parking space identification method and system based on deep learning | |
| Peng et al. | PeSOTIF: A challenging visual dataset for perception SOTIF problems in long-tail traffic scenarios | |
| CN112686082A (en) | System for counting number of people in vehicle based on visible light polarization | |
| CN113189590A (en) | FOD-based target identification method and device and storage medium | |
| CN108198164A (en) | A kind of detection device and method of cable tunnel cable integrity | |
| CN106864351A (en) | A kind of vehicle at night traveling auxiliary lighting system based on computer vision |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |