CN108038447A - A kind of pedestrian detecting system based on Zynq-7000 - Google Patents
A kind of pedestrian detecting system based on Zynq-7000 Download PDFInfo
- Publication number
- CN108038447A CN108038447A CN201711321125.4A CN201711321125A CN108038447A CN 108038447 A CN108038447 A CN 108038447A CN 201711321125 A CN201711321125 A CN 201711321125A CN 108038447 A CN108038447 A CN 108038447A
- Authority
- CN
- China
- Prior art keywords
- zynq
- usb interface
- isomery
- pedestrian
- soc
- 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
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
Abstract
The invention discloses a kind of pedestrian detecting system based on Zynq 7000, it is characterized in that, including 7000 Soc of isomery embeded processor Zynq and the camera with USB interface being attached thereto, the mouse with USB interface, band USB interface keyboard and HDMI display, camera with USB interface, the mouse with USB interface and be connected respectively with the PS of 7000 Soc of isomery embeded processor Zynq part with USB interface keyboard, HDMI display is partly connected with the PL of 7000 Soc of isomery embeded processor Zynq.This system is not only compact and flexible with embedded system, can compare favourably in process performance with PC machine, while pedestrian detection rate accuracy is high, small, easy for installation, cost is low.
Description
Technical field
The present invention relates to computer vision and protection and monitor field, more particularly to a kind of pedestrian's inspection based on Zynq-7000
Examining system.
Background technology
Pedestrian detection is to search for definite pedestrian using a kind of ad hoc approach to image to be detected in certain time to deposit
Whether, detect the quantity of pedestrian if it is present returning, and pedestrian density in the range of zoning.In recent years, with net
Network technology is maked rapid progress, and the application development of pedestrian detection is more and more extensive, due to the otherness between pedestrian's individual, movement appearance
The diversity of state, pedestrian occur the complexity of scene and block certainly, visual angle, illumination, the influence of the factor such as scale so that Hang Renjian
Examining system realizes that difficulty is larger.Current pedestrian detecting system is mostly based on PC machine realization, and there are of high cost, the big grade of volume lacks
Point.Reliability and all good pedestrian detecting system of real-time are realized on embedded platform, becomes the hot spot of current research.
Zynq-7000 is to be integrated with based on the complete programmable expansible platform structures of Xlinux, the structure in single-chip
There is the processing system (Processing System, abbreviation PS) of abundant double-core ARM Cortex-A9 polycaryon processors
With Xilinx programmable logic cells (Programmable Logic, abbreviation PL), double-core ARM Cortex-A9 multi-core CPUs are
" heart " of PS, it includes on-chip memory, external memory interface and a set of abundant I/O peripheral hardwares.PL in Zynq-7000
Using the FPGA technology of Xilinx7 series, for expanded function, to meet specific functional requirement, Zynq-7000 is used as first
The product that high-performance ARMCortex A series processors are combined closely with high-performance FPGA in single-chip, with other independent ARM
For Cortex-A9 compared with Xilinx FPGA are on veneer, it can have following advantage：Design cost reduces, design overall power drop
Low, design volume is reduced, design risk reduces and flexible design.In order to realize these advantages, Xilix is in design Zynq-7000
When the processor of different process feature and FPGA fusions on a single die and are not only ensured into its yields, more to design height
Path is interconnected between high-performance processor and FPGA in the piece of effect, which is realized by AXI buses, and AXI is a kind of
Bus protocol, the agreement be ARM companies AMBA agreements in most important part, be it is a kind of towards high-performance, high bandwidth, low prolong
Slow on-chip bus, meets the design requirement of very-high performance and system-on-chip, and AXI is to ensure to communicate with piece interconnect high-speed in piece
Basis, the internal unit of Zynq-7000 has AXI bus interface, and by AXI bus protocols, inside, which is started to write, to be accused
Tell the communication of low latency, i.e. ARM and FPGA can ensure the data transfer of high speed.
Traditional pedestrian detecting system has the problems such as bulky, equipment is heavy, is not easy to installation, is of high cost, exploitation
A pedestrian detection rate accuracy is high, and the pedestrian detecting system small, easy for installation, cost is low is applied to hair with good
Exhibition prospect.
The content of the invention
The purpose of the present invention is face the deficiencies in the prior art, and provide a kind of pedestrian detection system based on Zynq-7000
System.This system is not only compact and flexible with embedded system, can compare favourably in process performance with PC machine, while pedestrian examines
Survey rate accuracy is high, small, easy for installation, cost is low.
Realizing the technical solution of the object of the invention is：
A kind of pedestrian detecting system based on Zynq-7000, including isomery embeded processor Zynq-7000Soc
(System on Chip, system-on-chip, abbreviation Soc) and be attached thereto the camera with USB interface, the mouse with USB interface
Mark, with USB interface keyboard and HDMI display, the camera with USB interface, the mouse with USB interface and with USB interface keyboard
It is connected respectively with the PS parts of isomery embeded processor Zynq-7000Soc, HDMI display and isomery embeded processor
The PL parts of Zynq-7000Soc connect.
Hardware-accelerated IP is equipped with the isomery embeded processor Zynq-7000Soc, hardware-accelerated IP is used for image
Pre-processed, including image scaling, gray processing, realized at the PS ends of Zynq-7000 with programmable logic.
The camera with USB interface is used to gather pedestrian image information.
VMDA transmission IP in isomery embeded processor Zynq-7000Soc is realized by programmable logic, total using AXI
Line completes PL ends and communicates with PC ends.
Camera collection pedestrian image video information with USB interface is simultaneously uploaded to based on Zynq-7000 by USB interface
Zedboard development boards, Zedboard development boards integrate double-core ARM Cortex-A9 processor systems Processing
The USB interface of System and Xilinx programmable logic Programmable Logic, Zedboard development boards obtains picture number
According to image pretreatment operation and pass through by the hardware-accelerated IP in programmable logic Programmable Logic
Image information is uploaded to processing system Processing System, processing system Processing System fortune by AXI buses
Row (SuSE) Linux OS, pedestrian detection is carried out to the image pre-processed by hardware-accelerated IP, and the pedestrian detected is used
Rectangle frame marks, and is finally shown testing result by HDMI display.
Further, pedestrian detection algorithm is realized by the PS ends Linux application program modules of Zynq-7000.
Further, the pedestrian detection method is using HOG extraction features and svm classifier algorithm.
Linux device drives in isomery embeded processor Zynq-7000Soc are Linux user-application modules
System hardware is provided and the configuration and data interaction of IP, Linux user-application modules are realized pedestrian detection algorithm and provided
The human-computer interaction interface of friendly interface, the camera with USB interface, the mouse with USB interface, band USB interface keyboard and HDMI
Display provides user mutual, is used in the technical program isomery embeded processor Zynq-7000Soc in the progress of AXI buses
Portion's data transfer, has efficient pedestrian detection ability.
This system is not only compact and flexible with embedded system, can compare favourably in process performance with PC machine, at the same time
Pedestrian detection rate accuracy is high, small, easy for installation, cost is low.
Brief description of the drawings
Fig. 1 is the structure diagram of embodiment.
In figure, the hardware-accelerated IP 3.VMDA transmission IP 4.Linux device drives of 1. cameras 2. with USB interface
5.Linux user-application modules 6. are with mouse 8.HDMI display of the USB interface keyboard 7. with USB interface.
Embodiment
Present invention is further elaborated with reference to the accompanying drawings and examples, but is not limitation of the invention.
Embodiment：
With reference to Fig. 1, a kind of pedestrian detecting system based on Zynq-7000, including isomery embeded processor Zynq-7000
Soc and the camera 1 with USB interface being attached thereto, the mouse 7 with USB interface, band USB interface keyboard 6 and HDMI display
8, the camera 1 with USB interface, the mouse 7 with USB interface and with USB interface keyboard 6 respectively with isomery embeded processor
The PS parts of Zynq-7000 Soc connect, the PL parts of HDMI display 8 and isomery embeded processor Zynq-7000 Soc
Connection.
It is equipped with hardware-accelerated IP2 in the isomery embeded processor Zynq-7000 Soc, hardware-accelerated IP2 is used for pair
Image is pre-processed, including image scaling, gray processing, is realized at the PS ends of Zynq-7000 with programmable logic.
The camera 1 with USB interface is used to gather pedestrian image information.
VMDA transmission IP3 in isomery embeded processor Zynq-7000 Soc is realized by programmable logic, utilizes AXI
Bus completes PL ends and communicates with PC ends.
Camera 1 with USB interface gathers pedestrian image video information and is uploaded to by USB interface based on Zynq-
7000 Zedboard development boards, Zedboard development boards integrate double-core ARM Cortex-A9 processor systems Processing
The USB interface of System and Xilinx programmable logic Programmable Logic, Zedboard development boards obtains picture number
According to image pretreatment operation and pass through by the hardware-accelerated IP2 in programmable logic Programmable Logic
Image information is uploaded to processing system Processing System, processing system Processing System fortune by AXI buses
Row (SuSE) Linux OS, pedestrian detection is carried out to the image pre-processed by hardware-accelerated IP, and the pedestrian detected is used
Rectangle frame marks, and is finally shown testing result by HDMI display.
Further, pedestrian detection algorithm is realized by the PS ends Linux application program modules 5 of Zynq-7000.
Further, the pedestrian detection method is using HOG extraction features and svm classifier algorithm.
Linux device drives in isomery embeded processor Zynq-7000 Soc are Linux user-application modules
The configuration and data interaction of 5 offer system hardwares and IP, Linux user-application modules 5 are realized pedestrian detection algorithm and are carried
For the human-computer interaction interface of friendly interface, the camera 1 with USB interface, the mouse 7 with USB interface, 6 and of band USB interface keyboard
HDMI display 8 provides user mutual, and AXI buses are used in the technical program isomery embeded processor Zynq-7000 Soc
Internal data transfer is carried out, there is efficient pedestrian detection ability.
It is the function of several modules and realization in embodiment below：
Camera 1 with USB interface：For video acquisition, USB that this part passes through hardening at the PS ends of Zynq-7000
Interface is realized；
Hardware-accelerated IP2：Image is zoomed in and out, two Image Pretreatment Algorithm operations of gray processing, wherein image scaling
It is to be standardized to the picture size of processing, facilitate subsequent treatment, gray processing is carried out to image to be calculated to reduce
Measure and facilitate follow-up HOG feature extractions；
VMDA transmits IP3：Xilix for solve big data as similar image external memory storage and inside FPGA it
Between high speed data transfer problem and design, which is added by the XPS instruments in ISE external members；
Linux device drives 4：As the bridge between connection hardware and kernel, Linux device drives be operating system with
The passage of data interaction between hardware, therefore after the design of Hardware I P is completed, it is also necessary to select or design corresponding equipment
Driving and application layer program interface, could be provided as the behaviour such as configuration and reading and writing data in system user space to Hardware I P or outer
Make, Xilinx provides polytype IP kernel driving, according to the needs of the system, adds corresponding driving, adds herein
The video pipeline driving of offer, when configuring linux kernel, needs enabled parameter CONFIG_VIDEO_XILIX, and bind
Into Linux device trees, the video device driver of Xilix IP realizes that V4L2 is on video equipment in kernel based on V4L2
Kernel-driven frame, provides unified interface for the video equipment of upper access bottom, in user's space, can pass through ioctl
Order is interacted with driver.The configuration and access of hardware are carried out to system from user's space；
Linux user-application modules 5：It is divided into two parts, is respectively the realization of pedestrian detection algorithm and man-machine friendship
The realization at mutual interface, wherein pedestrian detection algorithm use HOG feature extractions and svm classifier algorithm to be realized, wherein algorithm steps
It is rapid as follows：
S1：Calculate gradient：The gradient of pixel (x, y) and gradient magnitude and direction are in image：
Gx(x, y)=f (x+1, y)-f (x-1, y)
Gy(x, y)=f (x, y+1)-f (x, y-1)
Wherein Gx(x, y) and Gy(x, y) represents level, the vertical gradient at pixel (x, y) place in input picture respectively
Value, M (x, y) represent gradient magnitude, and θ (x, y) represents gradient direction；
S2., input picture is divided into the small lattice of equal sizes, and several small lattice are merged into a fritter；
S3：The selection of direction passage：0 ° -180 ° or 0 ° -360 ° are divided into n passage；
S4：The acquisition of histogram：To each pixels statistics their gradient orientation histogram in each small lattice, histogram
Abscissa be n direction passage choosing in step S24, histogram ordinate is the ladder for the pixel for belonging to some direction passage
Spend the cumulative of size and finally obtain one group of vector；
S5：Normalized：In units of the fritter where the corresponding pixel of vector, vector is normalized,
Its formula is as follows：
Wherein, v*Representing the vector after normalization, v represents the vector before normalization,Represent 2 rank models of vector
Number, ε represent the constant of a very little, take 0.01 here；Its value does not interfere with result, it is therefore an objective to which denominator is 0 in order to prevent；
S6：Form HOG features：Processed all n vectors above are connected, form one group of vector, are HOG
Feature.
S7：Determine parameter σ={ σ of Nonlinear Support Vector Machines gaussian kernel function1, C={ C1}；
S8：SVM study is carried out using the pedestrian detection data set after HOG feature extractions, one is obtained and is based on Gauss
The non-linear SVM classifier f (x) of kernel function；
S9：Pedestrian detection is carried out using the non-linear SVM classifier f (x) based on gaussian kernel function.
Human-computer interaction interface has used the graphic user interface based on Qt, is shown by HDMI display, Yong Huke
To be interacted by way of USB mouse or USB keyboard input with system.User can observe pedestrian's inspection in display window
The effect and bus service condition of survey, complete the interaction between user and system.
When the user interface for opening pedestrian detecting system, pedestrian detecting system is enabled by mouse, user's observable is at this time
System resource utilization situation, such as cpu busy percentage, the data throughout of processing system port, goes out in the range of camera head monitor
During existing people, LCD display can mark pedestrian with rectangle frame.
Claims (2)
1. a kind of pedestrian detecting system based on Zynq-7000, it is characterised in that including isomery embeded processor Zynq-
7000 Soc and the camera with USB interface being attached thereto, the mouse with USB interface, band USB interface keyboard and HDMI are shown
Device, the camera with USB interface, the mouse with USB interface and with USB interface keyboard respectively with isomery embeded processor
The PS parts of Zynq-7000 Soc connect, and the PL parts of HDMI display and isomery embeded processor Zynq-7000 Soc connect
Connect.
2. the pedestrian detecting system according to claim 1 based on Zynq-7000, it is characterised in that the isomery insertion
Hardware-accelerated IP is equipped with formula processor Zynq-7000 Soc, hardware-accelerated IP is used to pre-process image, including image
Scaling, gray processing, are realized at the PS ends of Zynq-7000 with programmable logic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711321125.4A CN108038447A (en) | 2017-12-12 | 2017-12-12 | A kind of pedestrian detecting system based on Zynq-7000 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711321125.4A CN108038447A (en) | 2017-12-12 | 2017-12-12 | A kind of pedestrian detecting system based on Zynq-7000 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108038447A true CN108038447A (en) | 2018-05-15 |
Family
ID=62102196
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711321125.4A Pending CN108038447A (en) | 2017-12-12 | 2017-12-12 | A kind of pedestrian detecting system based on Zynq-7000 |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108038447A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344801A (en) * | 2018-10-26 | 2019-02-15 | 京东方科技集团股份有限公司 | A kind of object detecting method and device |
| CN109978752A (en) * | 2019-03-06 | 2019-07-05 | 南京航空航天大学 | A kind of image scaling IP kernel system based on SoC |
| CN111079669A (en) * | 2019-12-20 | 2020-04-28 | 京东方科技集团股份有限公司 | Image processing method, device and storage medium |
| CN112257490A (en) * | 2020-07-19 | 2021-01-22 | 扬州大学 | Embedded pedestrian density statistical system and method based on FPGA |
| CN112991382A (en) * | 2019-12-02 | 2021-06-18 | 中国科学院国家空间科学中心 | PYNQ frame-based heterogeneous visual target tracking system and method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140277827A1 (en) * | 2013-03-13 | 2014-09-18 | Service Solutions U.S. Llc | Vehicle measurement apparatus having a system-on-a-chip device and a sensor |
| CN105847766A (en) * | 2016-05-30 | 2016-08-10 | 福州大学 | Zynq-7000 based moving object detecting and tracking system |
| US9810736B2 (en) * | 2015-12-17 | 2017-11-07 | Raytheon Company | System and apparatus for trusted and secure test ports of integrated circuit devices |
-
2017
- 2017-12-12 CN CN201711321125.4A patent/CN108038447A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140277827A1 (en) * | 2013-03-13 | 2014-09-18 | Service Solutions U.S. Llc | Vehicle measurement apparatus having a system-on-a-chip device and a sensor |
| US9810736B2 (en) * | 2015-12-17 | 2017-11-07 | Raytheon Company | System and apparatus for trusted and secure test ports of integrated circuit devices |
| CN105847766A (en) * | 2016-05-30 | 2016-08-10 | 福州大学 | Zynq-7000 based moving object detecting and tracking system |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344801A (en) * | 2018-10-26 | 2019-02-15 | 京东方科技集团股份有限公司 | A kind of object detecting method and device |
| WO2020082936A1 (en) * | 2018-10-26 | 2020-04-30 | 京东方科技集团股份有限公司 | Object detection method and apparatus |
| US20200356801A1 (en) * | 2018-10-26 | 2020-11-12 | Boe Technology Group Co., Ltd. | Method and device for object detection |
| US11508144B2 (en) * | 2018-10-26 | 2022-11-22 | Beijing Boe Technology Development Co., Ltd. | Method and device for object detection |
| CN109978752A (en) * | 2019-03-06 | 2019-07-05 | 南京航空航天大学 | A kind of image scaling IP kernel system based on SoC |
| CN112991382A (en) * | 2019-12-02 | 2021-06-18 | 中国科学院国家空间科学中心 | PYNQ frame-based heterogeneous visual target tracking system and method |
| CN112991382B (en) * | 2019-12-02 | 2024-04-09 | 中国科学院国家空间科学中心 | Heterogeneous visual target tracking system and method based on PYNQ framework |
| CN111079669A (en) * | 2019-12-20 | 2020-04-28 | 京东方科技集团股份有限公司 | Image processing method, device and storage medium |
| CN112257490A (en) * | 2020-07-19 | 2021-01-22 | 扬州大学 | Embedded pedestrian density statistical system and method based on FPGA |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108038447A (en) | A kind of pedestrian detecting system based on Zynq-7000 | |
| CN105957090B (en) | A kind of monocular vision pose measuring method and system based on Davinci technology | |
| CN104038738B (en) | Intelligent monitoring system and intelligent monitoring method for extracting coordinates of human body joint | |
| CN110096933A (en) | The method, apparatus and system of target detection | |
| CN109029381A (en) | A kind of detection method of tunnel slot, system and terminal device | |
| CN110909611A (en) | Method and device for detecting attention area, readable storage medium and terminal equipment | |
| CN105844654A (en) | Augmented reality system and method based on Zynq software and hardware coprocessing | |
| CN102663758A (en) | Image acquiring and processing method based on FPGA (field programmable gate array) serving as control core | |
| CN108564008A (en) | A kind of real-time pedestrian and method for detecting human face based on ZYNQ | |
| JP6909829B2 (en) | Information processing methods, devices, systems and storage media | |
| CN110598715A (en) | Image recognition method and device, computer equipment and readable storage medium | |
| Weiyao et al. | Human action recognition using multilevel depth motion maps | |
| Safaei et al. | System-on-a-chip (soc)-based hardware acceleration for foreground and background identification | |
| CN110503083A (en) | A kind of critical point detection method, apparatus and electronic equipment | |
| CN111079536A (en) | Behavior analysis method based on human body key point time sequence, storage medium and equipment | |
| Hoang et al. | 3D skeleton-based action recognition with convolutional neural networks | |
| CN104715470B (en) | A kind of klt Corner Detections device and method | |
| CN114549765A (en) | Three-dimensional reconstruction method and device and computer-readable storage medium | |
| CN108184062A (en) | High speed tracing system and method based on multi-level heterogeneous parallel processing | |
| CN110633630B (en) | Behavior identification method and device and terminal equipment | |
| Dominguez-Morales et al. | Image matching algorithms in stereo vision using address-event-representation: A theoretical study and evaluation of the different algorithms | |
| CN205692214U (en) | A kind of monocular vision pose measurement system | |
| WO2023109086A1 (en) | Character recognition method, apparatus and device, and storage medium | |
| CN109710562A (en) | A kind of configurable and high speed FPGA configuration circuit and implementation method based on SELECTMAP | |
| CN116246343A (en) | Light human body behavior recognition method and device |
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: 20180515 |
|
| WD01 | Invention patent application deemed withdrawn after publication |