A kind of driving behavior early warning car-mounted terminal based on information fusion
Technical field
The utility model relates to a kind of vehicle active safety prior-warning device, particularly relates to a kind of driving behavior early warning car-mounted terminal based on information fusion, belongs to intelligent transportation field.
Background technology
Along with the continuous increase of motor vehicles, traffic safety accident problem highlights.According to Ministry of Public Security's analysis of statistical data, the leading reason causing traffic accidents is the dangerous driving behavior of driver, comprises the unsafe acts such as furious driving, fatigue driving, in violation of rules and regulations driving.
Department of Transportation proposes to the dangerous driving behavior early warning in public transportation system, danger kinds of goods vehicles operation system thinking of determining of requesting for conciliation accordingly especially, and market derives plurality of devices or the systems such as a series of driving behavior identification terminal as vehicle satellite locating terminal, automobile video frequency monitor terminal, vehicle operation characteristic acquisition terminal, video analysis thus.
But most of product carries out monitoring and early warning for video in vehicle driving trace and railway carriage, substance analysis is not carried out to driving behavior; Another part product relates to the safe early warning technology of driving behavior, but due to technical bottlenecks such as single piece of information source analysis algorithm difficulty are large, be difficult to make for driving behavior judge comparatively accurately, or driving behavior analysis index is incomplete.
In the patent achievement in similar technique field, patent ZL200910110643.0 proposes vehicle early warning analysis and processing method and system, describes the method for vehicle monitoring early warning under the prerequisite that vehicle risk state is clear and definite; Patent ZL200810040704.6 utilizes GPS intelligent vehicle mounted terminal to realize vehicle monitoring and dispatching system, realizes vehicle monitoring early warning; Patent ZL201110210735.3 utilizes camera collection to comprise the image of driver's hand and bearing circle, then judges driving behavior to the process of image with identification; Patent ZL201010244609.5 relates to driver's operation behavior monitoring technology, and a kind of driver behavior characteristic analysis method is described; These similar patents do not relate to the driving behavior early warning car-mounted terminal and method that merge based on driver's face image and driver's operation information.
Therefore, overcome the above problems and can take measures from following three aspects:
(1) make full use of mobile unit multi-source data, from driver's face image and driver's operating characteristics two kinds of detection meanss, extract the parameters that driving behavior is relevant;
(2) take multi-sources Information Fusion Method, overcome the problems such as the low accuracy of data mapping analysis, promote the reliability of driving behavior analysis early warning;
(3) strengthen interface compatibility and intercommunity, make terminal not only become driving behavior and detect monitoring and early warning equipment in the car analyzed, and can be remote vehicle supervision and upload early warning message in real time.
Therefore, the utility model proposes a kind of driving behavior early warning car-mounted terminal based on information fusion and method, and be different from existing end product or scheme with this.
Utility model content
For solving the deficiencies in the prior art, the utility model provides a kind of driving behavior early warning car-mounted terminal based on information fusion, by extracting the feature of driver's face image and driver behavior two aspects, feature is carried out convergence analysis, thus obtain the safety evaluatio of driving behavior and perform warning action, realize accurate, comprehensive early warning of vehicle driving safety.
The technical scheme that the utility model adopts is:
Based on a driving behavior early warning car-mounted terminal for information fusion, it is characterized in that: comprise prewarning unit, remote alarms unit in face image collection analysis unit, driving performance information collection analysis unit, information convergence analysis device, early warning judgement controller, car; Wherein, described face image collection analysis unit is communicated with information convergence analysis device respectively with driving performance information collection analysis unit; Described information convergence analysis device and early warning are adjudicated controller and are communicated with; Described early warning judgement controller is communicated with remote alarms unit respectively with prewarning unit in car again.
Wherein, described face image collection analysis unit comprises the high-definition image camera, the image analysis processor that are linked in sequence; Described driving performance information collection analysis unit comprises that pedal gathers submodule, bearing circle controls to gather submodule, and gather submodule with described pedal, bearing circle controls to gather the driving performance information analysis processor that submodule is connected.
Further, described information convergence analysis device comprises embedded processing submodule; And feature-based fusion algorithm and decision level fusion algorithm is preset in described embedded processing submodule; Described early warning judgement controller comprises the micro-control unit submodule, the button that are linked in sequence; And the set model of early warning judgement controller comprises artificial set model and auto-set mode; Further, above-mentioned information convergence analysis device and early warning judgement controller are made up of one piece of flush bonding processor, and described flush bonding processor has multiple control interface.
Further, in described car, prewarning unit is alarm, and described alarm is at least one in voice guard, word icon prompting alarm, light flash alarm; Described remote alarms unit carries out remote transmission by the wireless Ubiquitous Network of 2G or 3G.
Utilize the method for early warning of the above-mentioned driving behavior early warning car-mounted terminal based on information fusion, step is as follows:
(1) image collecting function of face image collection analysis unit is started, the facial picture rich in detail of Real-time Obtaining driver; The facial characteristics extract the facial characteristics of the driver being used for fatigue detecting, detecting for driver distraction; Main gene quantification is carried out to each eigenwert, and the quantization weight of assigned characteristics value, obtain driver fatigue, driver distraction's testing result by the evaluation index of various feature; Represent testing result by grade quantizing value, and result and each are detected eigenwert output to information convergence analysis device with digital signal;
(2) start the driver behavior collection apparatus function of driving performance information collection analysis unit, Real-time Obtaining driver is to the multiple operating characteristics of pedal, bearing circle; Operating characteristics is carried out main gene quantification, and the quantization weight of assigned characteristics value, obtain the security testing result based on driver behavior by the evaluation index of various feature; Represent testing result by grade quantizing value, and result and each are detected eigenwert output to information convergence analysis device with digital signal;
(3) information convergence analysis device receives the multi-source driving behavior information of face image collection analysis unit and driving performance information collection analysis unit, carries out data be correlated with and fusion reasoning according to the convergence analysis algorithm preset; Result after the analysis of information convergence analysis device is driving dangerousness behavior classification and value of safety factor value, result is outputted to early warning judgement controller.
(4), after early warning judgement controller receives the driving dangerousness behavior relevant information of information convergence analysis device output, secure threshold safety coefficient set with it is made comparisons; If lower than secure threshold, then perform prewarning unit action in car; In described car, prewarning unit drives by the control of early warning judgement controller, and in described car, prewarning unit adopts at least one mode in voice guard, word icon prompting alarm, light flash alarm to report to the police;
(5) after early warning judgement controller receives the driving dangerousness behavior relevant information of information convergence analysis device output, if value of safety factor value is lower than secure threshold, and require to upload early warning message according to safety grades, then early warning information is uploaded to car operation supervision department.Described remote alarms unit drives by the control of early warning judgement controller, and described remote alarms unit utilizes the wireless Ubiquitous Network of 2G or 3G, by the message transmissions of early warning information to remote vehicle supervising platform server, realizes the remote supervisory of vehicle.
Wherein, face image collection analysis unit described in above-mentioned steps () extract the driver being used for fatigue detecting facial characteristics comprises eyes size, eyes blink dynamic frequency, head pose, pupil diameter, Rotation of eyeball speed, eyes closed time length; Extract and be used for that facial characteristics that driver distraction detects comprises feature triangle that eyes and face form, head pose, head position, sight line depart from time, sight line deviation area.
Further, driving performance information collection analysis unit described in above-mentioned steps (two) extracts the operating characteristics of driver to pedal, bearing circle and comprises that driver's lower limb are trampled throttle, clutch coupling is trampled, the dynamics of brake pedal, speed and frequency, and driver's upper limbs is to the dynamics of wheel steering, bearing circle rotating speed, rotating of steering wheel, speed and frequency.
Further, information convergence analysis device internal preset described in above-mentioned steps (three) has feature-based fusion algorithm and decision level fusion algorithm, comprises Bayes theoretical analysis, the fusion method of D-S argument, the fusion method of fuzzy neural network, the fusion method of Kalman filtering, the fusion method of rough set theory, the fusion method of fuzzy theory.
Further, the set model of the early warning judgement controller described in above-mentioned steps (four) comprises artificial set model and auto-set mode; Described artificial set model setting threshold value is by pressing key assignments or rear end management software set; Described auto-set mode setting threshold value is determined automatically according to convergence analysis result.
Further, the alarm mode of voice guard to early warning information described in above-mentioned steps (four) comprises number of times, the sound of broadcasting, the volume of broadcasting of broadcasting; The described alarm mode of word icon prompting alarm to early warning information comprises the roll display mode of word, the display brightness of icon, displaying time and word or icon; The described alarm mode of light flash alarm to early warning information comprises the mechanical periodicity mode of intensity of light, color, flicker.
Further, the message of the early warning information described in above-mentioned steps (five) comprises frame head, function code, information code, check code, postamble.
The beneficial effects of the utility model are: can carry out active prevention to car accident hidden danger, especially carry out monitoring for dangerous driving behavior, analyze, early warning; Multi-source driving behavior information is excavated in abundant collection, provide after convergence analysis and comparatively accurately, comprehensively differentiate and early warning conclusion, improve vehicle self drive safety, for vehicle operating provides safety guarantee, also can be a kind of effective means that the supervision of car operation supervision department to safe driving of vehicle behavior provides supervision enforce foundation; Each work package modularization work that the utility model relates to, compatibility is good, extendability is strong, can easily with other information of vehicles register system, vehicle monitoring system Integrated predict model, thus effectively improve vehicle active safety prevention level, support vehicles safe driving, improves traffic safety level.
accompanying drawing illustrates:
Fig. 1 is the utility model structural representation;
Fig. 2 is utility model works process flow diagram;
Fig. 3 is the utility model information fusion process flow diagram;
Fig. 4 is a kind of implementation concrete structure figure of the present utility model.
embodiment:
Below in conjunction with accompanying drawing, concrete introduction is done to the utility model.
Embodiment one: a kind of driving behavior early warning car-mounted terminal based on information fusion
As Figure 1-3: the face image collection analysis unit comprising the facial picture rich in detail in order to obtain driver, and described face image collection analysis unit comprises the high-definition image camera, the image analysis processor that are linked in sequence; In order to obtain the driving performance information collection analysis unit of driver to the multiple operating characteristics of pedal, bearing circle, and described driving performance information collection analysis unit comprises, and pedal gathers submodule, bearing circle controls to gather submodule, and gather submodule with described pedal, bearing circle controls to gather the driving performance information analysis processor that submodule is connected; Simultaneously, the present embodiment also comprises and gathers to above-mentioned two kinds of modes the information come and to merge and controller is adjudicated in the convergence analysis device that process and early warning, and described convergence analysis device comprises the embedded processing submodule of interior default feature-based fusion algorithm and decision level fusion algorithm; The set model of described early warning judgement controller comprises artificial set model and auto-set mode; And above-mentioned information convergence analysis device and early warning judgement controller are made up of one piece of flush bonding processor, and described flush bonding processor has multiple control interface; The present embodiment also comprises prewarning unit and remote alarms unit in car; In described car, prewarning unit is alarm, and described alarm is at least one in voice guard, word icon prompting alarm, light flash alarm; Described remote alarms unit carries out remote transmission by the wireless Ubiquitous Network of 2G or 3G.Wherein, described face image collection analysis unit is communicated with information convergence analysis device respectively with driving performance information collection analysis unit; Described information convergence analysis device and early warning are adjudicated controller and are communicated with; Described early warning judgement controller is communicated with remote alarms unit respectively with prewarning unit in car again.
Embodiment two: a kind of method for early warning of the driving behavior early warning car-mounted terminal based on information fusion, step is as follows:
Start the image collecting function of face image collection analysis unit, the facial picture rich in detail of Real-time Obtaining driver; First, the facial characteristics extracting the driver being used for fatigue detecting, the facial characteristics detected for driver distraction; Wherein, described face image collection analysis unit extract the driver being used for fatigue detecting facial characteristics comprises eyes size, eyes blink dynamic frequency, head pose, pupil diameter, Rotation of eyeball speed, eyes closed time length; Secondly, extract and be used for that facial characteristics that driver distraction detects comprises feature triangle that eyes and face form, head pose, head position, sight line depart from time, sight line deviation area; After above-mentioned information acquisition completes, by image acquisition analytic unit, main gene quantification is carried out to each eigenwert, and the quantization weight of assigned characteristics value, obtain driver fatigue, driver distraction's testing result by the evaluation index of various feature; Represent testing result by grade quantizing value, and result and each are detected eigenwert output to information convergence analysis device with digital signal.
Start the driver behavior collection apparatus function of driving performance information collection analysis unit, Real-time Obtaining driver is to the multiple operating characteristics of pedal, bearing circle; And the multiple operating characteristics of described driver to pedal, bearing circle comprises that driver's lower limb are trampled throttle, clutch coupling is trampled, the dynamics of brake pedal, speed and frequency, and driver's upper limbs is to the dynamics of wheel steering, bearing circle rotating speed, rotating of steering wheel, speed and frequency; After above-mentioned information acquisition completes, by driving performance information collection analysis unit, operating characteristics is carried out main gene quantification, and the quantization weight of assigned characteristics value, obtain the security testing result based on driver behavior by the evaluation index of various feature; Represent testing result by grade quantizing value, and result and each are detected eigenwert output to information convergence analysis device with digital signal.
Information convergence analysis device receives the multi-source driving behavior information of face image collection analysis unit and the output of driving performance information collection analysis unit, wherein, there are feature-based fusion algorithm and decision level fusion algorithm in information convergence analysis device internal preset, comprise Bayes theoretical analysis, the fusion method of D-S argument, the fusion method of fuzzy neural network, the fusion method of Kalman filtering, the fusion method of rough set theory, the fusion method of fuzzy theory; Described information convergence analysis device carries out data according to the convergence analysis algorithm preset and is correlated with and fusion reasoning; And the result after analyzing is driving dangerousness behavior classification and value of safety factor value, meanwhile, result is outputted to early warning judgement controller by information convergence analysis device.
After early warning judgement controller receives the driving dangerousness behavior relevant information of information convergence analysis device output, the decision threshold set with it is made comparisons; Wherein, the set model of described early warning judgement controller comprises artificial set model and auto-set mode; Described artificial set model setting threshold value is by pressing key assignments or rear end management software set; Described auto-set mode setting threshold value is determined automatically according to convergence analysis result; In described car, prewarning unit drives by the control of early warning judgement controller, and in described car, prewarning unit adopts at least one mode in following three kinds of type of alarms to report to the police, with the careful operation of alerting drivers; Three kinds of type of alarms are as follows:
1, voice guard, and the alarm mode of described voice guard to early warning information comprises number of times, the sound of broadcasting, the volume of broadcasting of broadcasting;
2, word icon prompting alarm, and the described alarm mode of word icon prompting alarm to early warning information comprises the roll display mode of word, the display brightness of icon, displaying time and word or icon;
3, light flash alarm, and the described alarm mode of light flash alarm to early warning information comprises the mechanical periodicity mode of intensity of light, color, flicker;
If early warning judgement controller receives the driving dangerousness behavior safety coefficient value of information convergence analysis device output lower than secure threshold, then perform early warning action in car; If early warning judgement controller receives the driving dangerousness behavior safety coefficient value of information convergence analysis device output lower than secure threshold, and requires to upload early warning message according to safety grades, then need early warning information to be uploaded to car operation supervision department; Described remote alarms unit drives by the control of early warning judgement controller, and described remote alarms unit utilizes the wireless Ubiquitous Network of 2G or 3G, by comprise frame head, function code, information code, check code, postamble the message transmissions of early warning information to remote vehicle supervising platform server, finally realize the remote supervisory of vehicle.
Embodiment three: the device embodiments that a kind of each cell mesh of the driving behavior early warning car-mounted terminal based on information fusion is selected
As shown in Figure 4:
1, the high-definition image camera of face image collection analysis unit adopts the ccd sensor of 8,000,000 pixels, and filming frequency 50Hz, selects TMS320DM642 Digital Image Processing chip to be image analysis processor; The SPI interface of TMS320DM642 is connected by SPI interface with information convergence analysis device;
2, in driving performance information collection analysis unit, pedal gathers dynamics and the speed that submodule adopts brake pedal force and displacement transducer detection pedal, output voltage signal; Bearing circle controls to gather submodule and adopts relative value rotary angle transmitter, exports square-wave signal; Driving performance information analysis processor adopts low cost S TM32F101C8T6 process chip; STM32F101C8T6 is connected by SPI interface with information convergence analysis device;
3, information convergence analysis device and early warning judgement controller adopt one piece of flush bonding processor to realize, flush bonding processor is ARM chip s3c2440, configuration dominant frequency is 400MHz, and memory expansion is to 128M, and chip periphery circuit comprises stereo audio output interface circuit and LCD Interface circuit; S3c2440 is all connected by SPI interface with driving performance information collection analysis unit with face image collection analysis unit; S3c2440 is connected by stereo audio output interface, LCD Interface, GPIO interface with prewarning unit in car; S3c2440 is connected by serial ports with remote alarms unit;
4, in car, prewarning unit comprises the speech chip and loudspeaker, 3.5 cun of TFT full color displays for writings and image display, highlighted red, yellow, and green LED for glimmering reported for sound, and these three assemblies are connected with the stereo audio output interface of s3c2440, LCD Interface, GPIO interface respectively;
5, remote alarms unit comprises the wireless data transmission module for sending message, realizes 2G or 3G wireless communication transmissions, supports by note or TCP/IP mode transparent transmission; Wireless data transmission module is connected with the serial ports of s3c2440;
6, above all modules adopt 12V DC power supply, for chip and peripheral circuit provide power supply after voltage stabilizing.
The above is only the preferred implementation of the utility model patent; should be understood that; for those skilled in the art; under the prerequisite not departing from the utility model patent principle; can also make some improvements and modifications, these improvements and modifications also should be considered as the protection domain of the utility model patent.