CN107480695A - A kind of contour of object detection recognition method - Google Patents
A kind of contour of object detection recognition method Download PDFInfo
- Publication number
- CN107480695A CN107480695A CN201710555808.XA CN201710555808A CN107480695A CN 107480695 A CN107480695 A CN 107480695A CN 201710555808 A CN201710555808 A CN 201710555808A CN 107480695 A CN107480695 A CN 107480695A
- Authority
- CN
- China
- Prior art keywords
- dictionary
- anomalous event
- contour
- object detection
- recognition method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2134—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on separation criteria, e.g. independent component analysis
- G06F18/21345—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on separation criteria, e.g. independent component analysis enforcing sparsity or involving a domain transformation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/513—Sparse representations
Abstract
The invention belongs to frontier defense safety monitoring technology field, more particularly to a kind of contour of object detection recognition method.This method comprises the following steps：（1）KPCA Data Dimensionality Reductions；（2）Dictionary initializes；（3）Rarefaction representation is classified；（4）Anomalous event processing and online dictionary updating.The present invention gathers the contour feature of different objects by some groups of thermopile IR array of temperature sensor, wirelessly sends to computer, is identified using the Target Recognition Algorithms based on dictionary learning.The present invention has the features such as detection range is wide, disguised high, highly reliable, using anomalous event treatment mechanism, improves the scalability of system, and adaptivity.
Description
The application be Application No. 201510064774.5, the applying date be on 2 6th, 2015, invention and created name is a kind of thing
The divisional application of body contour detecting identifying system and outline identification method.
Technical field
The present invention relates to frontier defense safety monitoring technology field, especially a kind of contour of object detection recognition method.
Background technology
Realize that the real-time monitoring to border or depopulated zone has great importance to national security.In this across border activity
In the age to take place frequently, all countries are all in the effective method for improving boundary line safety of searching in the world.China's boundary line on land
Total length about 22, more than 000 kilometers, wherein many places are unoccupied, if automatic monitoring can be carried out these places, undoubtedly will
It is greatly enhanced national security prevention ability.
China is at present in existing monitoring system, and nearly 95% is all the Wu Fashi using being manually monitored by monitor
The now Intelligent Measurement to certain event and identification are big to the dependence of people.Due to the factor such as sleepy, tired, monitoring personnel is difficult to protect
Demonstrate,prove all the period of time effective monitoring, existence information utilization rate and the deficiencies of system operating efficiency is low, poor reliability.In addition, existing side
The early warning effect and real-time response performance of anti-safety monitoring system are also fallen flat, and wireless video monitoring system is adopted
The data bulk collected is very big, for very long border or the depopulated zone of the length and breadth of land, nets the interior number for transmitting, storing and receiving
It is too high according to considerable and cost, therefore also result in bulk information redundancy and be difficult to intelligentized application.
The content of the invention
In view of the above-mentioned problems, it is different using different classes of object appearance profile feature, with reference to compressive sensing theory, there is provided one
Kind contour of object detection recognition method.Front end data acquisition platform uses new thermopile infrared sensor array and FPGA processing
Device gathers the contour feature of object, has the features such as detection range is wide, disguised high, simple in construction, highly reliable.
To achieve the above object, present invention employs following technical scheme：
A kind of contour of object detects identifying system, including data acquisition device, data recording control apparatus, wireless transmitter and
Electric power controller.Data acquisition device, data recording control apparatus, wireless transmitter and electric power controller composition front end
Data acquisition platform.Described data acquisition device, its output end are connected with the input of data recording control apparatus.Described
Data recording control apparatus, its output end are connected with the input of wireless transmitter.Described wireless transmitter and profile
Wireless connection between the wireless transmitter of identifying system.
Described data acquisition device includes some groups of thermopile IR array of temperature sensor.
Described data recording control apparatus includes FPGA processor, crystal oscillating circuit, reset circuit and download configuration circuit.
Described wireless transmitter includes wireless transmitting terminals and wireless receiving end.
Described electric power controller is respectively that data acquisition device, data recording control apparatus and wireless transmitter supply
Electricity.
Described thermopile IR array of temperature sensor uses MLX90620 modules.
Described FPGA processor uses EP1C3T144 chips.
Described wireless transmitter uses nRF24L01 radio transmitting and receiving chips.
Described thermopile IR array of temperature sensor is arranged on fixed support, and thermopile IR temperature sensor
The angular field of view of array is 60 °.Thermopile IR array of temperature sensor is arranged on fixed support or tied up stable at some
On object, it is desirable to be that site visual angle is enough wide, sensor detection visual angle can reach 60 degree.
The invention further relates to a kind of contour of object detection recognition method, this method is that the target based on online dictionary learning is known
Other algorithm, including the initialization of KPCA Data Dimensionality Reductions, dictionary, rarefaction representation classification, anomalous event treatment mechanism and online dictionary are more
Newly.This method includes KPCA Data Dimensionality Reductions, dictionary initialization, rarefaction representation classification, anomalous event processing and online dictionary updating
Four steps.Specifically, described a kind of contour of object detection recognition method, first collect front-end acquisition device one
The different training sample of series is converted into the characteristic vector of same dimension by dimensionality reduction.Then it is eigenvector projection is empty to low-dimensional
Between construct dictionary, rarefaction representation and identification are carried out to test sample by the dictionary, can be sentenced according to sparse coefficient and residual values
Determine whether test sample belongs to anomalous event.If anomalous event, then anomalous event storehouse is classified to；If it is not, then according to residual
Poor size determines object generic.In anomalous event storehouse, anomalous event is classified by clustering, when same class is abnormal
Event occurs after reaching certain frequency, and this kind of event is removed from exception database, is put into dictionary, realizes the online of dictionary
Renewal.
From above technical scheme, front end data acquisition platform of the invention uses thermopile infrared sensor array pair
Data acquisition is carried out by the object in the region for being provided with the platform, it is defeated to obtain sensor array using data recording control apparatus
The contour of object data gone out, and the contour of object data of collection are sent to contour of object using wireless transmitter and identify system
System, contour of object identifying system judge object classification, this method according to a kind of Target Recognition Algorithms based on online dictionary learning
First to initializing dictionary after training sample dimensionality reduction, rarefaction representation and identification are carried out to test sample by the dictionary, and propose
A kind of anomalous event treatment mechanism, is first recorded to emerging anomalous event, deposit anomalous event storehouse, and is directed to abnormal thing
Part storehouse, when the frequency that certain a kind of anomalous event occurs is enough, change it to regular event, realize dictionary it is online more
Newly, and from anomalous event storehouse remove.This have the advantage that the different situation of test sample dimension is can adapt to, and energy
Abnormal conditions are handled and judged, improve the scalability of system, algorithm is adaptive to different situations.
The invention has the advantages that：
1st, it is wide with detection range using thermopile infrared sensor array acquisition data, low-power consumption, good concealment, integrated level
The features such as high.
2nd, system identification goes out automatic alarm after specific objective, reduces the human resources of monitoring personnel, is realizing intelligence inspection
Survey the effective monitoring with turn ensure that all the period of time while identification.
3rd, outline identification algorithm uses core principle component analysis method dimensionality reduction, will be no longer by linearly not by nonlinear transformation algorithm
Can merotype constraint, and can adapt to the different situation of test sample dimension.
4th, a kind of anomalous event treatment mechanism is proposed, abnormal conditions can be handled and judged, and in due course
Online updating dictionary, the scalability of system is improved, algorithm is adaptive to different situations.
Brief description of the drawings
Fig. 1 is the structural representation of the present invention；
Fig. 2 is data acquisition device and data recording control apparatus wiring schematic diagram；
Fig. 3 is the pin schematic diagram of FPGA processor in data recording control apparatus；
Fig. 4 is the circuit theory diagrams of reset circuit in data recording control apparatus；
Fig. 5 is the circuit theory diagrams of crystal oscillating circuit in data recording control apparatus；
Fig. 6 is the circuit theory diagrams of download configuration circuit in data recording control apparatus；
Fig. 7 is the circuit theory diagrams of wireless transmitter；
Fig. 8 is power circuit principle figure in electric power controller；
Fig. 9 is dictionary initialization flowchart；
Figure 10 is anomalous event treatment mechanism flow chart.
Wherein：
1st, data acquisition device, 2, data recording control apparatus, 3, wireless transmitter, 4, electric power controller, 5, profile knows
Other system.
Embodiment
The present invention will be further described below in conjunction with the accompanying drawings：
As shown in figure 1, a kind of contour of object detection identifying system, including front end data acquisition platform and outline identification system 5.Before
The contour of object data of end data acquisition platform collection are sent to outline identification system 5 by wireless transmitter, then use
A kind of contour of object recognition methods based on online dictionary learning judges object generic.Front end data acquisition platform includes number
According to harvester 1, data recording control apparatus 2, wireless transmitter 3 and electric power controller 4.Described data acquisition device
1, its output end is connected with the input of data recording control apparatus 2.Described data recording control apparatus 2, its output end with
The input of wireless transmitter 3 is connected.The wireless transmitter of described wireless transmitter 3 and outline identification system 5 it
Between wireless connection.The wireless transmitter of outline identification system is connected by USB interface with PC.
Described data acquisition device 1, is MLX90620 modules using thermopile IR array of temperature sensor, for catching
Obtain object two-dimensional silhouette image information.It is about 1 that described thermopile IR array of temperature sensor, which is arranged on apart from ground level,
On the fixed support of rice.MLX90620 is the accurate 164 pixel IR arrays of whole school, is integrated in the pin TO-39 of industrial standard four encapsulation,
Comprising 64 IR pixels, each pixel is corresponding with low noise chopper amplifier and high-speed ADC, exports the RAM inside existing,
And pass through I2C communications obtain.The letter that the non-contact temperature sensor that thermopile IR array of temperature sensor is 164 forms
Number perceive unit, there are tetra- pins of VDD, VSS, SDA and SCL.As shown in Fig. 2 data acquisition device includes chip U1, its
Power pins VDD meets 3.3VCC, and pin VSS is directly grounded, power pins VDD and grounding pin VSS indirect electric capacity C1, clock
Pin SCL divides two-way, meets 3.3VCC by resistance R1 all the way, and another way meets chip U2D the 141st pin, data pin SDA
Divide two-way, 3.3VCC is connect by resistance R2 all the way, another way connects chip U3D the 140th pin.
Described data recording control apparatus 2 includes FPGA processor, crystal oscillating circuit, reset circuit and download configuration electricity
Road.Described FPGA processor is using fpga chip EP1C3T144 chips serial CycloneI.Described crystal oscillating circuit and multiple
The output end of position circuit, is connected with the input of FPGA processor respectively.Data recording control apparatus, collection for data,
Storage, processing and transmitting-receiving.The utility model uses 20MHz crystal oscillator as global clock, when other circuits need different frequency
Clock when, divide to obtain by fpga chip.The structure of FPGA processor is shown in data recording control apparatus as shown in Figure 3
The structural representation of meaning FPGA processor.FPGA processor includes tetra- chips of U2A, U2B, U2C and U2D.
The circuit theory diagrams of reset circuit in data recording control apparatus as shown in Figure 4.Reset circuit includes resistance R3
In parallel with R4 to be followed by 3.3VCC, resistance R3 connects with button switch S1, and resistance R4 connects with button switch S2, button switch S1,
It is grounded after S2 parallel connections, diode D1 anode is connected between resistance R4 and button switch S2.Between resistance R3 and button switch S1
Lead-out wire RESET is connected with chip U2C the 92nd pin, lead-out wire nCONFIG and chip between resistance R4 and button switch S2
U2A the 14th pin is connected.Button switch S1 is resetted as software reset according to personal code work；Button switch S2 makees
For hardware reset, when pressing button switch S2, label nCONFIG leads are low level, now all FPGA codes again from
Read inside configuration chip A1 in FPGA processor, program restarts to run.
The circuit theory diagrams of crystal oscillating circuit in data recording control apparatus as shown in Figure 5.Described crystal oscillating circuit includes
Chip Y1, its 2nd pin ground connection, its 3rd pin connect chip U2C the 93rd pin, and its 4th pin divides two-way, passes through electricity all the way
Hold C2 ground connection, another way meets 3.3VCC, there is the global clock that source crystal oscillator provides 20MHz for system.
The circuit theory diagrams of download configuration circuit in data recording control apparatus as shown in Figure 6.Described download configuration
Circuit include configuration chip A1 and jtag interface JP1, JP2, jtag interface JP1 the 1st, 3,5,9 pins respectively with chip U2C
The 88th, 90,89,95 pins be connected, jtag interface JP2 the 1st, 5,7,9 pins respectively with chip U2A the 24th, 14,13,
25 pins are connected, and jtag interface JP2 the 3rd pin is connected with chip U2C the 86th pin, and the 1st, 2,5, the 6 of configuration chip A1
Pin respectively with chip U2A the 12nd, 13,25,24 pins be connected, configuration chip A1 the 3rd, 7,8 pins meet 3.3VCC respectively,
7th, 8 pins are grounded by electric capacity C3, its 4th pin ground connection, resistance R8, R9, R13 one end respectively with chip U2A the 22nd,
23rd, 21 pins are connected, and are grounded after resistance R8, R9, R13 other end parallel connection, resistance R10, R11 one end respectively with chip U2C
The 87th, 86 pins be connected, resistance R12 one end is connected with chip U2A the 14th pin, and resistance R10, R11, R12's is another
End parallel connection is followed by 3.3VCC.If downloaded by jtag interface JP1, needed after system power failure again under FPGA processor
Carry, if downloaded by jtag interface JP2, user program code will be stored in configuration chip A1, and system is upper electric every time
Afterwards, user program code will read in FPGA processor from configuration chip A1 automatically, then run.
Such as the circuit theory diagrams of Fig. 7 wireless transmitter, described wireless transmitter 3 includes chip U3, its 1st,
2nd, 3,4,5,6 pins respectively with chip U2B the 60th, 59,58,57,56,55 pins be connected, its 9th, Indirect Electro of 10 pins
Hold X1 and resistance R15, be then grounded respectively by electric capacity C10, C11, its 11st pin is grounded by electric capacity C12, connect by C4
Ground, it is connected by inductance L2 with chip U3 the 12nd pin, the 13rd pin phase that its 12nd pin passes through inductance L1 and chip U3
Even, its 13rd pin can provide stable RF outputs, its 15th pin by being used as RFI/O after inductance L3, electric capacity C5 to antenna
It is grounded with the 18th pin by C8, is grounded by C9, then meets power supply 3.3VCC, its 16th pin is grounded by R14, and it the 19th
Pin is grounded by electric capacity C7.Wireless transmitter by wirelessly by front-end data acquisition device collect information data transmission to
Outline identification system, outline identification system are identified further according to the data signal collected using the specific objective based on dictionary learning
Algorithm is handled, identified.
Position electric power controller as shown in Figure 8, described electric power controller 4 include chip U4, chip U5 and interface
J3, interface J3 the 2nd, 3 pins ground connection, interface J3 the 1st pin directly meets 5VCC, chip U4 the 1st, have electricity between 3 pins
Hold C14 and C13, wherein the 3rd pin meets 5VCC, chip U4 the 2nd pin is connected by electric capacity C15 with the 1st pin, wherein the 1st
Pin is grounded, and the 2nd pin meets 3.3VCC, and chip U5 the 3rd pin meets 3.3VCC all the way, and another way passes through C16 and the 1st pin phase
Even, chip U5 the 2nd pin meets 1.5VCC all the way, and another way is connected by C17 with the 1st pin, wherein the 1st pin is grounded, electricity
It is in parallel to hold C18, C19, C20, C21, a termination 1.5VCC, other end ground connection, electric capacity C22, C23, C24, C25, C26, C27,
C28, C29, C30 are in parallel, a termination 3.3VCC, other end ground connection.Described electric power controller 4 is respectively data acquisition device
1st, data recording control apparatus 2 and wireless transmitter 3 are powered.
Described front end data acquisition platform includes data acquisition device 1, data recording control apparatus 2, wireless receiving and dispatching dress
Put 3 and electric power controller 4.In the present invention, front end data acquisition platform is at least 1 group.When monitoring system of the present invention
When system includes multigroup thermopile IR array of temperature sensor, multigroup front end data acquisition platform is formed into wireless sensor network
Network, carry out the collection and transmission of data.
The invention further relates to a kind of contour of object recognition methods based on online dictionary learning to include KPCA Data Dimensionality Reductions, word
Allusion quotation initialization, rarefaction representation classification, anomalous event treatment mechanism and online dictionary updating.The detailed process of this method is as follows：
1st, KPCA Data Dimensionality Reductions：Front end data acquisition platform is mainly responsible for data acquisition, it is assumed that a people, a people are drawn respectively
Chest, a people squats down gathers training sample by these three situations, per a kind of sample number to be N number of, because everyone walks speed
The difference of degree, the sample dimension collected are different.Pass through core principle component analysis method（Specific visible bibliography：Kernel
PCA for Feature Extraction and De-Noising in Non-linear Regression）By 3N training
Sample extracts principal component to eliminate redundancy respectively by Nonlinear Mapping into nuclear space, while is translated into phase
With the characteristic vector of dimension, to facilitate construction dictionary, this method has good robustness to the speed of travel.
2nd, dictionary initializes：As shown in figure 9, respectively to a people, a people draws chest, and a people squats down this three classes event
Prepare several samples, carry out Data Dimensionality Reduction respectively, obtain the characteristic vector of same dimension in lower dimensional space, formed normalized
Vector is used as dictionary atom, dictionary atom is pressed into row permutation and combination, you can the dictionary initialized.
3rd, rarefaction representation is classified：When thering is object to pass through in sensor array visual field, front-end acquisition device gathered data, adopt
The data collected are test sample, and first carrying out Data Dimensionality Reduction to test sample obtains characteristic vector, and by dictionary to feature
Vector carries out rarefaction representation（Specific visible bibliography：Robust face recognition via sparse
representation）, primary signal is reconstructed according to sparse coefficient of all categories respectively and calculates residual error, if the residual error calculated is more than
Threshold value between class, it is determined as new events, and is included into anomalous event storehouse, is otherwise classified according to residual error minimum classification decision principle.
4th, anomalous event treatment mechanism and online dictionary updating：As shown in Figure 10, it is contemplated that the complexity of reality, with
When be likely to anomalous event occur.Anomalous event refers to being not comprised in the new events inside dictionary.For anomalous event
Processing, construct an anomalous event storehouse, if judged result is an anomalous event in previous step, this event be put into different
Normal event base；Otherwise by classification of the minimum kind judging of residual error for sample and output., will by clustering in anomalous event storehouse
Anomalous event is classified, after certain frequency occurs reaching in same class anomalous event, by this kind of event from exception database
Middle removal, is put into dictionary, realizes the online updating of dictionary.
When object is by one group of thermopile IR array of temperature sensor or by multigroup thermopile IR temperature sensor
During the wireless sensor network of array composition, the two-dimensional silhouette image letter of thermopile IR array of temperature sensor collection object
Breath.Data recording control apparatus gathers the output state of thermopile IR array of temperature sensor at regular intervals, and by this
A little states are wirelessly sent to contour of object identifying system.Outline identification system is based on online dictionary learning according to one kind
Target Recognition Algorithms judge object classification.In a word, the present invention has the features such as detection range is wide, disguised high, highly reliable,
Using anomalous event treatment mechanism, the scalability of system, and adaptivity are improved.
Embodiment described above is only that the preferred embodiment of the present invention is described, not to the model of the present invention
Enclose and be defined, on the premise of design spirit of the present invention is not departed from, technical side of the those of ordinary skill in the art to the present invention
The various modifications and improvement that case is made, it all should fall into the protection domain of claims of the present invention determination.
Claims (5)
- A kind of 1. contour of object detection recognition method, it is characterised in that：This method comprises the following steps：（1）KPCA Data Dimensionality Reductions；（2）Dictionary initializes；（3）Rarefaction representation is classified；（4）Anomalous event processing and online dictionary updating.
- A kind of 2. contour of object detection recognition method according to claim 1, it is characterised in that：Step（1）In, it is described KPCA Data Dimensionality Reductions are that a series of different training samples for collecting front-end acquisition device are empty to core by Nonlinear Mapping Between in, and extract principal component to eliminate redundancy, while be translated into the characteristic vector of same dimension.
- A kind of 3. contour of object detection recognition method according to claim 1, it is characterised in that：Step（2）In, it is described It will be to prepare several samples per a kind of event to carry out Data Dimensionality Reduction respectively in advance that dictionary initialization, which is, obtain phase in lower dimensional space With the characteristic vector of dimension, the dictionary that can be initialized by row permutation and combination.
- A kind of 4. contour of object detection recognition method according to claim 1, it is characterised in that：Step（3）In, it is described Rarefaction representation classification is will to carry out rarefaction representation by dictionary after test sample dimensionality reduction, reconstructed respectively according to sparse coefficient of all categories Primary signal simultaneously calculates residual error, and anomalous event and the other judgement of object type are realized according to residual error size and its correlation.
- A kind of 5. contour of object detection recognition method according to claim 1, it is characterised in that：Step（4）In, it is described Anomalous event processing and online dictionary updating are that first emerging anomalous event is recorded, deposit anomalous event storehouse, and pin To anomalous event storehouse, when the frequency that certain a kind of anomalous event occurs is enough, regular event is changed it to, realizes dictionary Online updating, and removed from anomalous event storehouse.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710555808.XA CN107480695A (en) | 2015-02-06 | 2015-02-06 | A kind of contour of object detection recognition method |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510064774.5A CN104616023A (en) | 2015-02-06 | 2015-02-06 | Object outline detection and recognition system and object outline recognition method |
| CN201710555808.XA CN107480695A (en) | 2015-02-06 | 2015-02-06 | A kind of contour of object detection recognition method |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510064774.5A Division CN104616023A (en) | 2015-02-06 | 2015-02-06 | Object outline detection and recognition system and object outline recognition method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107480695A true CN107480695A (en) | 2017-12-15 |
Family
ID=53150460
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510064774.5A Pending CN104616023A (en) | 2015-02-06 | 2015-02-06 | Object outline detection and recognition system and object outline recognition method |
| CN201710555808.XA Pending CN107480695A (en) | 2015-02-06 | 2015-02-06 | A kind of contour of object detection recognition method |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510064774.5A Pending CN104616023A (en) | 2015-02-06 | 2015-02-06 | Object outline detection and recognition system and object outline recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN104616023A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280417A (en) * | 2018-01-18 | 2018-07-13 | 苏州折衍光电科技有限公司 | A kind of finger vena method for quickly identifying |
| CN115629783A (en) * | 2022-10-27 | 2023-01-20 | 北方工业大学 | Model updating method for keeping privacy and resisting abnormal data in mobile crowd sensing |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11350264B2 (en) | 2018-07-25 | 2022-05-31 | Samsung Electronics Co., Ltd. | Method and apparatus for establishing device connection |
| CN113237567A (en) * | 2021-04-23 | 2021-08-10 | 广州碳思科技有限公司 | Device and method for recognizing object contour based on temperature difference |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101236605A (en) * | 2008-02-29 | 2008-08-06 | 重庆大学 | Human body intelligent recognition method and system based on infrared signal |
| CN102663425A (en) * | 2012-03-29 | 2012-09-12 | 浙江大学 | Combined target segmentation and identification method based on shape sparse representation |
| CN102938070A (en) * | 2012-09-11 | 2013-02-20 | 广西工学院 | Behavior recognition method based on action subspace and weight behavior recognition model |
| CN103020594A (en) * | 2012-12-03 | 2013-04-03 | 清华大学苏州汽车研究院（吴江） | Fatigue state detecting method for eliminating driver individual difference by utilizing online learning |
| CN104091151A (en) * | 2014-06-30 | 2014-10-08 | 南京信息工程大学 | Vehicle identification method based on Gabor feature extraction and sparse representation |
| CN104156979A (en) * | 2014-07-25 | 2014-11-19 | 南京大学 | Method for on-line detection of abnormal behaviors in videos based on Gaussian mixture model |
| CN104268594A (en) * | 2014-09-24 | 2015-01-07 | 中安消技术有限公司 | Method and device for detecting video abnormal events |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3581566B2 (en) * | 1998-06-10 | 2004-10-27 | 株式会社日立製作所 | Monitoring system |
| CN1166922C (en) * | 2002-07-18 | 2004-09-15 | 上海交通大学 | Multiple-sensor and multiple-object information fusing method |
| CN100446043C (en) * | 2007-02-06 | 2008-12-24 | 重庆大学 | Video safety prevention and monitoring method based on biology sensing and image information fusion |
| CN204423397U (en) * | 2015-02-06 | 2015-06-24 | 安徽大学 | A kind of contour of object detects recognition system |
-
2015
- 2015-02-06 CN CN201510064774.5A patent/CN104616023A/en active Pending
- 2015-02-06 CN CN201710555808.XA patent/CN107480695A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101236605A (en) * | 2008-02-29 | 2008-08-06 | 重庆大学 | Human body intelligent recognition method and system based on infrared signal |
| CN102663425A (en) * | 2012-03-29 | 2012-09-12 | 浙江大学 | Combined target segmentation and identification method based on shape sparse representation |
| CN102938070A (en) * | 2012-09-11 | 2013-02-20 | 广西工学院 | Behavior recognition method based on action subspace and weight behavior recognition model |
| CN103020594A (en) * | 2012-12-03 | 2013-04-03 | 清华大学苏州汽车研究院（吴江） | Fatigue state detecting method for eliminating driver individual difference by utilizing online learning |
| CN104091151A (en) * | 2014-06-30 | 2014-10-08 | 南京信息工程大学 | Vehicle identification method based on Gabor feature extraction and sparse representation |
| CN104156979A (en) * | 2014-07-25 | 2014-11-19 | 南京大学 | Method for on-line detection of abnormal behaviors in videos based on Gaussian mixture model |
| CN104268594A (en) * | 2014-09-24 | 2015-01-07 | 中安消技术有限公司 | Method and device for detecting video abnormal events |
Non-Patent Citations (2)
| Title |
|---|
| 刘洋 等: "基于线性动态系统稀疏编码的异常事件检测", 《计算机科学》 * |
| 谢锦生 等: "基于快速稀疏编码与惊奇计算的视频异常检测方法", 《中国科学技术大学学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280417A (en) * | 2018-01-18 | 2018-07-13 | 苏州折衍光电科技有限公司 | A kind of finger vena method for quickly identifying |
| CN115629783A (en) * | 2022-10-27 | 2023-01-20 | 北方工业大学 | Model updating method for keeping privacy and resisting abnormal data in mobile crowd sensing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104616023A (en) | 2015-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107480695A (en) | A kind of contour of object detection recognition method | |
| CN103325080B (en) | A kind of home for the aged based on technology of Internet of things intelligent safeguard system and method | |
| CN102283616B (en) | Domestic intelligent cleaning system based on machine vision | |
| CN102811343B (en) | Intelligent video monitoring system based on behavior recognition | |
| CN108564005A (en) | A kind of human body tumble discrimination method based on convolutional neural networks | |
| CN209186698U (en) | A kind of epilepsy early warning device based on FPGA | |
| CN208505501U (en) | A kind of intelligent carpet based on pliable pressure sensor | |
| CN103559508B (en) | A kind of based on continuous Adaboost video vehicle detection method | |
| CN104850841B (en) | Combination RFID and video identification a kind of old man abnormal behaviour monitoring method | |
| CN103413441A (en) | Road weather condition monitoring system and monitoring method | |
| CN206557859U (en) | A kind of enterprise's attendance management and long-range Fire Monitor System | |
| CN109558785A (en) | Safety defense monitoring system and the unmanned convenience store for applying it | |
| CN103456168B (en) | A kind of traffic intersection pedestrian behavior monitoring system and method | |
| Peng et al. | Design and development of the fall detection system based on point cloud | |
| CN101236605B (en) | Human body intelligent recognition method and system based on infrared signal | |
| CN113963192A (en) | Fall detection method and device and electronic equipment | |
| CN207663490U (en) | A kind of mixing recognition access control system management system based on neural calculation rod | |
| CN107724900B (en) | A kind of family security door based on personal recognition | |
| CN205179112U (en) | Host computer and next quick -witted communication system of reservoir detection of water level system | |
| CN110674717B (en) | Smoking monitoring system based on gesture recognition | |
| CN112668430A (en) | Smoking behavior detection method and system, computer equipment and storage medium | |
| CN204423397U (en) | A kind of contour of object detects recognition system | |
| CN201717992U (en) | Video passenger detector | |
| CN111626273A (en) | Fall behavior recognition system and method based on atomic action time sequence characteristics | |
| CN202404694U (en) | Adaptive disturbance signal identification module of distributing type optical fiber sensing application system |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171215 |
|
| RJ01 | Rejection of invention patent application after publication |