CN106887059A - A kind of intelligent electronic lock system based on face recognition - Google Patents

Abstract

The present invention is on the basis of intelligent equipment, mainly use camera, obtain the image information of target, for the face-image that dynamic object extracts people, and itself and door lock link together by analyzing the facial characteristics of target and the image comparison of database, to judge whether invader is external unknown personnel, so as to realize the intelligent switch of door lock.The present invention has that high cost, environmental suitability are improved for current face recognition door control system, is designed using ARM Cortex M3 embedded chips, reduces hardware cost and operation power consumption, lifts the performance of hardware.MATLAB image processing platforms are based on simultaneously, it is proposed that improved 2DPCA face recognition algorithms.Simultaneously using features of skin colors algorithm locating human face's information is based on, lower environmental background influence, improve face identification rate.Compared with than traditional 2DPCA algorithms, set forth herein the innovatory algorithm used time it is shorter, while face the match is successful that rate reaches 87%, False Rate is less than 3%, and security of system reaches more than 97%.

Description

A kind of intelligent electronic lock system based on face recognition

Technical field

The present invention relates to a kind of intelligent identification device, and in particular to a kind of electronic door lock system based on face recognition.

Background technology

With the development of social economy, the innovation of technology, people attain the well-off standard the high standard of living of society, Er Qiean gradually Anti- consciousness is also being improved constantly.Thus, how safer, convenient, reliable safety guard system is designed using new technological means System, improves quality of residents'life, has become intelligent security guard field focus of attention.Wherein, door lock is used as intelligent safety and defence system The Zhong Shou roads outpost of the tax office, is the most important thing of whole safety-protection system.The authentication mode of conventional door lock system has：Key, password, Magnetic card etc., but these checking means all have separability with user, are easily replicated, decode and usurp, can not be complete Meet modern security protection theory.Now, human body biological characteristics such as fingerprint, face, iris etc. has a uniqueness because of it, consistency and not Separable characteristic, has been widely used in intelligent security guard field.Therefore design is recognized for the intelligent door lock of safety guard System, has important economic worth and Practical significance to the development of society.

The content of the invention

Today's society, the intelligent security guard equipment based on biological identification technology is becoming increasingly popular, due to biological identification technology The uniqueness that has, consistency and it is inseparable the features such as, be key character that other technologies are lacked.And face recognition is also One of topic being wherein widely studied.The intelligent door lock system based on face recognition also begins to heat up in recent years, but due to it The market price is sufficiently expensive, and recognition success rate is influenceed than larger, so promote also being stranded very much by factors such as environment, illumination Difficulty, compared to other safety-protection systems, its occupation rate of market is slightly lower, but development space is huge.

The present invention is achieved through the following technical solutions：Using ARM Cortex-M3 embedded chips design, reduce hardware into With operation power consumption, the performance of hardware is lifted.MATLAB image processing platforms are based on simultaneously, it is proposed that improved 2DPCA faces are known Other algorithm, in the sample training stage, builds different proper subspaces to each class personage respectively, and recognizes rank in face-image Section, then matched in the proper subspace that images to be recognized is projected to each class people respectively, is found maximum matching degree and is come true The ownership of fixed personage to be identified.Facial information is positioned using based on features of skin colors algorithm simultaneously, lowers environmental background influence, improved Face recognition rate.

On the basis of face recognition, using equipment personage outdoors is identified with analysis, then with P Passable Portrait photographs compare, then judge whether consistent, then determine whether that unlocking is allowed access into.

A kind of electronic door lock system based on face recognition, including stm32 main control modules, camera, serial communication module, Human body infrared induction module, memory module, electric control lock drive module, display module, acousto-optic hint module, power transfer module, Power switch, remote control reception module, sound and light alarm module, camera and motor drive module, it is characterised in that door-locking system work Main control module is communicated information to by other modules when making, modules is fed back to after being processed through main control module, and perform corresponding work Make.This hardware system has and stronger adaptability and transplanting characteristic, it is not necessary to change hardware system, you can carry out varying environment Transplanting, it is only necessary to the door-locking system under varying environment is expanded into by changing software parameters.

Preferably, in the hardware system, by camera module, serial port module, human body infrared induction module, TFT-LCD Module etc. communicates information to main control module, feeds back to modules after being processed through main control module, and perform relevant work.

Preferably, the hardware system includes electric control lock drive module, TFT-LCD display modules, acousto-optic hint module, DC Power transfer module, camera module, CH340G serial communication modulars, human body infrared induction module and SD card memory module etc. 9 Big hardware module composition.

Preferably, the software systems program part is broadly divided into two major parts, MATLAB programmings and STM32 master control journeys Sequence is designed.

Preferably, the MATLAB programmings, can be divided into serial communication, face detection, facial match, customer data base Four bulks such as management, make host computer be based on MATLAB softwares, design the monitoring software of face recognition intelligent door lock system, monitor Program interface simple and clear, several buttons can just realize the monitor in real time of system during user only needs to click on program interface, and Complete face recognition intelligent door lock system is constituted with slave computer STM32.

Preferably, the STM32 primary control programs design, each hardware module of itself and hardware system is corresponding, including face Positioning shooting head program, electromechanical locks drive module program, TFTLCD display modules driver, SD card memory module program, people Body infrared detection program, CH340 communication modules program and low-power consumption mode program etc..

Operation principle of the invention is：The work of each submodule is regulated and controled by main control module, and phase between module Mutual co-ordination.First by human body infrared induction module, whether someone is near door-locking system for real-time detection, if detecting people Body signal, then activation is in the main control chip under holding state, and then OV7670 positioning shootings head module completes initialization, prepares Face-image is obtained, is aided in using TFT-LCD LCDs, the facial picture position of display in real time facilitates user to position, and is easy to Head portrait is obtained.Simultaneously with TFT-LCD display screen Chinese prompts, user is reminded currently to perform task situation, every time activation face There are 3 distinguishing opportunities after identification door-locking system, if three identification fails, system can be braked and enter holding state, etc. Treat that human body infrared induction module reacquires trigger signal next time, electromechanical locks drive module primary recipient comes from STM32 master controls Instruction, perform and open the door and close door procedure, acousto-optic hint module then coordinates electromechanical locks drive module, performs command adapted thereto, plays User's effect is reminded, USB high-definition cameras are that face-image obtains main camera, and PC ends are sent to after shooting facial head piece MATLAB programs carry out face recognition, and SD memory modules meeting real-time storage recognizes the picture for getting every time, to retain evidence, It is convenient to check record at any time.

The present invention compared with prior art, with following beneficial effect：

1st, according to systematic isolation target, the Hardware Design based on ARM Cortex-M3 master controls is completed, performs face Portion's positioning, TFT-LCD intelligent prompts, and switch lock function.

2nd, based on MATLAB image processing platforms design intelligent monitoring software, complete based on features of skin colors face detection with And based on improved 2DPCA face recognition algorithms, while real-time management can also be carried out to customer data base, checking monitoring record.

3rd, good machine Interaction Interface Design, is designed by the GUI for learning MATLAB, completes monitor system software interface Design, including optimization button, simplify operation order, optimization display box etc..

4th, hardware system low-power consumption mode, only after human body infrared sensing is triggered, system can just be activated, three identification After failure, system is unresponsive automatically into the low-power consumption such as standby mode setting with 10 seconds automatically into low-power consumption standby pattern.

Brief description of the drawings

Fig. 1 is the main assembly framework of hardware system.

Fig. 2 is system hardware connection figure.

Fig. 3 is the overall program flow diagram of the intelligent door lock system monitoring software based on face recognition.

Fig. 4 is based on the intelligent door lock system monitoring software surface chart of face recognition.

Fig. 5 door-locking system service charts.

Specific embodiment

With reference to embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited In this.

Embodiment

Referring to Fig. 1,2, the main assembly framework and system hardware connection figure of hardware system, the enhanced system of STM32 single-chip microcomputers STM32F103ZET6 in row product, it is 32 miniature CPU, the processing frequency 72MHz for being based on ARM Cortex-M3 cores, With up to 64KB SRAM and 512KB FLASH, abundant timer resource, including 2 senior timers, 4 general timings Device, 2 basic timers, communication interface include 5 serial ports, 1 USB interface, 3 SPI interfaces, 2 IIC interfaces, and in addition 5 Also 1 12 DAC, 3 12 ADC, 1 FSMC interface etc., chip totally 144 pin, wherein 112 universal I/O ports, use LQFP-144 encapsulating structures.Electric control lock drive module uses TB6612FNG DC motor Driver chips, its MOSFET-H bridge knots Structure has high current drive capability, and with binary channels output, can simultaneously drive two-way direct current generator by PWM.Each passage Highest exports 1.2A Continuous Drive electric currents, can control direct current generator forward/reverse/stopping/braking, and highest supports 100kHZ's PWM frequency, with low pressure detection in piece and hot stoppage protection circuit, operating temperature：- 20~85 DEG C, sealed using SSOP24 pasters Dress.LCDs is that, using 2.8 cun of TFT-LCD of ALIENTEK, the liquid crystal display carries LCD driving chips IL9320,5V electricity Source powers, operating current 130mA~350mA, and resolution ratio shows up to 320*240 and 16 RGB.Therefore the display is used Screen power consumption is relatively low, meets the design demand.Dual camera gathered data is employed in the design, wherein Face location camera is adopted With ALIENTEK OV7670 camera modules, face acquisition camera is using USB high-definition cameras.Face location camera OV7670 is directly controlled by STM32 microprocessors, and its image for obtaining is displayed on TFT-LCD liquid crystal in real time, facilitates user to determine The position of oneself face of position.Face obtains camera and is controlled by the Matlab softwares of PC, and major function is to obtain face Image is identified treatment.Two camera installation requirements are that acquired extraneous picture should be identical, setting height(from bottom), inclination angle, Place environment is consistent.Acousto-optic hint module includes buzzer circuit and common anode red and green color LED lamp circuit, and hardware system is activated When, buzzer can short ring one sound prompting user, when face is identified as work(, buzzer also can one sound of short ring, while green light is lighted, Prompting is verified, system perform unlock order, when facial recognition failures, buzzer can it is short ring two sound, while red light can be bright Rise, point out authentication failed, it is necessary to user verifies again.The RCW-0506 human body infrared induction modules that the design is used, are to be based on The automatic control module of infrared technology, using German import probe LHI778, super low-power consumption work, small volume, sensitivity are high, It is widely used in security product, human body sensing light fixture, industrial automatic control etc..The design is also to use battery powered mode, RCW-0506 human body infrared induction modules standby current is less than 50uA, is very suitable for battery feed circuit and uses, hardware system Non-working time, now infrared module was using as the monitor of unique activity, its low power capabilities in holding state So that the low-power consumption standby performance of whole system is notable, make it possible that the design uses dry cell power supply.

Referring to Fig. 3,4, the overall program flow diagram of the intelligent door lock system monitoring software based on face recognition and based on face Portion identification intelligent door lock system monitoring software surface chart, under normal work, first door lock hardware system by data wire with Computer USB mouthful is connected, and then starts MATLAB ends monitoring programme, clicks on program initialization button, wait it is to be initiated finish, this When collection face-image frame occur camera real-time pictures, then click on operation monitoring, program will be into monitor in real time, so far PC Setting completed at end, next starts door-locking system hardware components, and STM32 can enter holding state, only human body infrared sensing mould Block maintains normal detection work, when someone enters infrared induction module detection range, and when being continuously detected more than 2 seconds, STM32 main control modules will be waken up, and now need the time of 2 seconds or so and carry out hardware system initialization, after initialization is finished, Lcd screen can point out user locating surface size picture, prepare the facial picture of acquisition and be identified, after user's face is positioned, OV7670 cameras and USB high-definition cameras can in time catch user's positioning head picture, and photo is captured after finishing, the short ring of buzzer One sound, LCD refreshes user's head portrait that display was captured just now, and now PC ends MATLAB will perform face detection and face recognition journey Sequence, LCD shows sentences such as " recognizing, please after a while ... ", the instruction after waiting the identification of PC ends to finish, if recognition result is to close Method user, then control electric control lock drive module performs door opening action after STM32 receives door open command, and otherwise STM32 will connect Instruction of locking a door is received, LCD shows that " recognition failures please be retried！", at this moment user is needed to reposition face, and taking pictures again is carried out Detection, if three identification fails, STM32 systems will enter standby mode, and human body infrared induction module can also enter 10 seconds blocking times, within this time, human body infrared induction module will no longer detect outer signals, and user can only wait next The arrival in secondary triggered time, could take pictures identification again.

Additionally, user can carry out various data base administrations in the intelligent door lock system monitoring software of the face recognition at PC ends Operation：Current Lawful user situation is for example checked, after clicking on " checking validated user ", automatic spring window shows current all conjunctions The head portrait picture of method user, when the new user of typing is needed, clicks on " the new user of typing " button, pop-up prompting " continuous typing 50 Facial picture is opened, ' confirmation ' continuation please be click on ", at this moment program can call USB high-definition cameras, and structure of taking pictures is carried out to new user Picture library, after about 30 seconds picture obtain finish can pop-up window prompting " new user profile typing is finished, please click on ' it is determined that ' after It is continuous ", this stylish user suffers in facial-recognition security systems, again taps on program initialization, it is possible to update validated user letter Breath, when needing to delete validated user, clicks on " deleting user " button, and system will eject user's picture library root, only need The personage " delete " of desired deletion.It is exactly above user data database management module function.Except user data Increase, delete, change, system is also provided with " checking record " button, be easy to check that user enters record, while can also be supervised as family Control record, when family is unfortunately forced open the door by robber, user can check record of forcing open the door by monitoring record, so can be used as detection One of capsule information of case.

Face recognition algorithm based on PCA：Mainly using PCA (PCA), the algorithm is this paper face recognitions A kind of method based on multivariate statistics.PCA mainly uses the thought of dimensionality reduction, research how using a few main component come The variance of multivariable is explained, wherein each principal component can reflect most information in original variable, and contained information is mutually not It is related.Because principal component analysis converts data to low-dimensional, people is allowed more to can be visually seen data structure, therefore PCA is also commonly used for Mode identification technology, data compression technique and data Feature Extraction Technology.

Traditional PCA method general principles are：Principal character composition, construction feature in face are extracted using Karhunen-Loeve transformation Face space matrix, in face recognition, images to be recognized is projected on feature space, obtains one group of projection coefficient, Ran Houtong Cross and compare identification with everyone facial characteristics.Using the coordinate transformation method of this decorrelation so that before compression of images Mean square deviation error is minimum afterwards, and still has good recognition capability after being transformed into lower dimensional space.

This paper face recognition methods are improved based on traditional 2DPCA algorithms, and traditional 2DPCA is in the training stage Different face-images is built into proper subspace jointly, in face recognition, face-image projection 23 to be identified is arrived It is compared on proper subspace.When Sample Storehouse has multiple different classes of personages, if different classes of personage is built into one jointly Euclidean distance is less than normal in individual proper subspace, so its class, and the Euclidean distance between inhomogeneity is larger with respect in class, and recognizes threshold Value T determines according to maximum Euclidean distance between sample again, so often cause erroneous judgement because recognition threshold T is bigger than normal.Thus, originally Text proposes a kind of 2DPCA face recognition algorithms based on multiple proper subspaces, in the sample training stage, respectively to each class Personage builds different proper subspaces, and in face-image cognitive phase, then images to be recognized is projected into each class respectively Matched in the proper subspace of people, the match is successful then images to be recognized belong to such personage, otherwise it is assumed that images to be recognized It is juridical-person thing.

Fig. 5 door-locking system service charts.

It is above-mentioned for the present invention preferably implementation method, but embodiments of the present invention are not limited by the above, its He any without departing from Spirit Essence of the invention and the change, modification, replacement made under principle, combine, simplify, should be The substitute mode of effect, is included within protection scope of the present invention.

Claims (7)

1. a kind of electronic door lock system based on face recognition, including stm32 main control modules, camera, serial communication module, people Body infrared induction module, memory module, electric control lock drive module, display module, acousto-optic hint module, power transfer module, electricity Source switch, remote control reception module, sound and light alarm module, camera and motor drive module, it is characterised in that door-locking system works When main control module is communicated information to by camera module, serial port module, human body infrared induction module, TFT-LCD modules etc., through master Modules are fed back to after control resume module, and relevant work is performed, this system have and stronger adaptability and transplants characteristic, Hardware system need not be changed, you can carry out the transplanting of varying environment, it is only necessary to be expanded into difference by changing software parameters Door-locking system under environment.

2. the electronic door lock system based on face recognition according to claim 1, it is characterised in that use with kernel ARM Cortex-M3 is the STM32F103ZET6 chips of framework, is compared with embedded chip, and it is performed, and code efficiency is higher, and support is outer The advantages of extension of portion's memory, support low-power consumption mode.

3. the electronic door lock system based on face recognition according to claim 2, it is characterised in that what the system was selected is Automatically controlled sun lock, because this door-locking system is related to a large amount of hardware modules and ancillary equipment, it is necessary to unify to be integrated into door lock, using electricity The positive lock scheme of control is designed, it is to avoid wiring, installation procedure it is cumbersome, portable stronger, embedded system is fully-integrated Only need to change the lock body of script in door lock, during user installation.

4. the electronic door lock system based on face recognition according to claim 1, it is characterised in that this system employs double Camera gathered data, wherein Face location camera use ALIENTEK OV7670 camera modules, face to obtain camera Using USB high-definition cameras.

5. the electronic door lock system based on face recognition according to claim 1, it is characterised in that the RCW-0506 of use Human body infrared induction module, is the automatic control module based on infrared technology, using German import probe LHI778, ultralow work( Work consuming work, small volume, sensitivity are high.And acousto-optic hint module includes buzzer circuit and common anode red and green color LED lamp circuit, firmly When part system is activated, buzzer can it is short ring one sound prompting user, when face is identified as work(, buzzer also can it is short ring one sound, Simultaneously green light light, prompting is verified, system perform unlock order, when facial recognition failures, buzzer can it is short ring two sound, Red light can be lighted simultaneously, point out authentication failed, it is necessary to user verifies again.

6. the electronic door lock system based on face recognition according to claim 1, it is characterised in that based on features of skin colors Type of face detection method, analyzes by color characteristic, and discovery uses restraint under YCgCr and YCgCb color spaces, can be fine Meet under Asian Face Detection, and YCgCr and YCgCb color spaces, influenceed smaller by brightness, at face In also can rapidly detect facial positions under complex background, a large amount of non-features of skin colors, fabulous reservation flesh tone portion can be excluded Image.

7. the electronic door lock system based on face recognition according to claim 4, it is characterised in that propose improved 2DPCA face recognition algorithms, in the sample training stage, build different proper subspaces to each class personage respectively, and in face Portion's image recognition stage, then matched in the proper subspace that images to be recognized is projected to each class people respectively, found most Big matching degree determines the ownership of personage to be identified.