A kind of intelligent wiper starts method
Technical field
Starting method, especially one kind the present invention relates to a kind of rain brush, can not have to user manually opened and fully meet use
The intelligent wiper of family actual demand starts method.
Background technology
Rain brush is device indispensable on various vehicles, is related to Exploration on Train Operation Safety of the vehicle in the rainy day, is hindered for eliminating
Sleet, dust on gear glass etc., the user visual field is ensured to ensure glass transparent clearly.Existing most of rain brush are all
User is needed to be adjusted manually, this undoubtedly causes the scattered so as to be easy to cause accident of user's notice in driving, and
The gear of adjusting is limited, can not well adapt to the demand of user.Also exist by rainfall inductor to detect rainwater so as to certainly
The dynamic rain brush being controlled, then rain sensor is typically to detect water droplet using principles such as infrared ray, capacitance, thermocouples
In the presence of, it is necessary to all install many hardware devices additional in each car, causing the increase of cost, and for southern user and
Speech, the control of this principle can not often detect the presence of drizzle, cause user to still need hand in drizzle
It is dynamic to be operated.
Also by the way of image is controlled, by the feature for identifying picture, extracting picture, it is filtered etc. multiple
It can just judge whether open rain brush after processing, but this mode needs each car to be all equipped with the very powerful processing of function
Device, significantly increases the cost of operation, and correctness influence of the selection of algorithm on result is very big, if algorithm is not suitable with certain
The characteristic of the rainwater in area, then the result identified is likely to wrong.
The content of the invention
Therefore, in view of the above-mentioned problems, starting method the present invention provides a kind of intelligent wiper, it can be achieved that rain brush automatically turns on
And fully meet user's actual need, it is not necessary to which the cost of excessive increase vehicle, greatly improves user experience.
In order to achieve the above object, the present invention proposes a kind of intelligent wiper startup method, including:
1. video acquisition;
2. data acquisition;
3. obtain image;
4. intelligence learning;
5. intelligent starting.
Above-mentioned intelligent wiper starts method, it further also meets condition:Step 1 is specifically, using being placed in in-car forward sight
What the collection in real time of camera near mirror position was obtained by front windshield consistent with the driver visual field carries time shaft mark
Video is simultaneously sent to entire car controller by the video of note in real time.
Above-mentioned intelligent wiper starts method, it further also meets condition:Camera is the camera of automobile data recorder.
Above-mentioned intelligent wiper starts method, it further also meets condition:Camera is powered by in-car cigarette lighter plug.
Above-mentioned intelligent wiper starts method, it further also meets condition:Step 2 is specifically, detection stepless time adjustment device
Whether it is rotated, initial time will be rotated, the rotation end time is sent to entire car controller.
Above-mentioned intelligent wiper starts method, it further also meets condition:Camera uses super clear camera.
Above-mentioned intelligent wiper starts method, it further also meets condition:Video format is MPEG-4.
The intelligent wiper of the present invention starts method, it is only necessary to realizes video acquisition, data acquisition and is tied according to collection in car side
Fruit obtains image and then the process for uploading image, it is not necessary to which car side carries out excessive information and collects, handles, analyzes work, subtracts
The light burden of car side, while Cloud Server is based on big data and carries out intelligent depth study to image, draws and meets user demand
Neural model, and neural model software implementation is handed down to car side, and then only need to install software in car side and run can be real
Show fully automated and there is the rain brush intelligent starting of high user experience, rain can not only be opened in time when user needs
Brush and can stablize the speed needed in user.
Brief description of the drawings
A kind of intelligent wiper that Fig. 1 is the present invention starts method flow diagram.
Embodiment
Embodiment one.
Refer to Fig. 1.
A kind of intelligent wiper starts method, including:
1. video acquisition;
2. data acquisition;
3. obtain image;
4. intelligence learning;
5. intelligent starting.
Step 1 is specifically, collection passes through front windshield in real time using the camera being placed near in-car front-view mirror position
Video is simultaneously sent to entire car controller by obtained consistent with the driver visual field video with time shaft mark in real time.
In order to cost-effective so that scheme has more feasibility, and camera can be installed in most of vehicle
The camera of automobile data recorder.In order to improve the precise degrees of image recognition, camera at least uses high-definition camera, preferably adopts
With super clear camera.For the ease of transmitting and handling, video should reach balance between clarity and size, therefore should
Using mpeg format, especially MPEG-4 forms.
Camera is usually powered by in-car cigarette lighter plug, can also use storage battery, the solar cell additionally set
Power Deng device.
Step 2 specifically, detection stepless time adjustment device whether be rotated, will rotate initial time, rotate the end time hair
Give entire car controller.
Embodiment two.
A kind of intelligent wiper starts method, it is characterised in that it comprises the following steps:
Obtained using the camera being placed near in-car front-view mirror position collection in real time by front windshield and driver
Video is simultaneously sent to entire car controller by the consistent video with time shaft mark in the visual field in real time;
Whether detection stepless time adjustment device is rotated, and will rotate initial time, the rotation end time is sent to full-vehicle control
Device;
Entire car controller is based on rotation initial time one predetermined reaction time of rollback and obtains rotating required time, in band sometimes
The rotation required time corresponding image is intercepted to obtain to image during rotation demand, in band in the video of countershaft mark
Sometimes the rotation end time corresponding image is intercepted into the image at the end of being rotated in the video of countershaft mark, will
There are image during rotation demand, the image at the end of rotation to be packaged into a file, sent out this document by telecommunication mode
Give cloud server;
Cloud server receives the multiple files for uploading to come from multiple entire car controllers, will have rotation in All Files
Image during demand is grouped together into rotating order training picture set, by the image sets at the end of being rotated in All Files
It is combined to form constant speed order training picture set;Cloud server will rotate each figure ordered in training picture set
As resolving into the subgraph overlapped according to certain tactic some, each subgraph is input to a nervelet net
In network, the output of the nervelet network is stored in a small ordered series of numbers, all small ordered series of numbers are arranged according to certain order
Row obtain a big ordered series of numbers, which is input in one big neutral net, so obtain can according to image judgement be
The no first nerves controller there are rotation demand;Each image in picture set is trained in constant speed order by cloud server
The subgraph overlapped according to certain tactic some is resolved into, each subgraph is input to a nervelet network
In, the output of the nervelet network is stored in a small ordered series of numbers, all small ordered series of numbers are arranged according to certain order
A big ordered series of numbers is obtained, which is input in one big neutral net, and then obtain to judge rotating speed according to image
Whether the nervus opticus controller of user demand has been met;
First nerves controller, nervus opticus controller difference software are melted into the first software, second soft by cloud server
Part, and issue the first software, the second software to all entire car controllers;
Each entire car controller receives the first software, the second software and is installed;Entire car controller will carry time shaft
The video of mark is input to the first software in real time, when the first software judges that there are export to rotate requirement command to whole during rotation demand
Vehicle controller, entire car controller receives controls rain brush to start with a predetermined acceleration raising speed turn immediately when rotating requirement command
It is dynamic, while the video marked with time shaft is no longer input to the first software by entire car controller in real time, and by the band having time
The video of axis mark is input to the second software in real time;It is fixed to be exported when the second software judges that rain brush rotating speed has met user demand
Speed is instructed to entire car controller, and entire car controller controls rain brush persistently to be rotated with present speed immediately when receiving constant speed instruction,
And entire car controller continues the video with time shaft mark being input to the first software and the second software in real time respectively;When
Output reduction of speed is instructed to entire car controller when first software judges rotation demand is not present, and entire car controller receives reduction of speed instruction
When control rain brush to be rotated with a predetermined acceleration reduction of speed immediately, when the second software judges that rain brush rotating speed no longer meets user demand
When, instruction of raising speed is exported to entire car controller, and entire car controller controls rain brush to make a reservation for add with one immediately when receiving speed-raising instruction
Speed raising speed rotates.
Embodiment three.
A kind of intelligent wiper starts method, it is characterised in that it comprises the following steps:
Obtained using the camera being placed near in-car front-view mirror position collection in real time by front windshield and driver
Video is simultaneously sent to entire car controller by the consistent video with time shaft mark in the visual field in real time,
The emphasis for adjusting the video that camera is taken and the side in the driver visual field that the position of camera should be suitably
Emphasis is consistent, therefore, can not use the camera of automobile data recorder, and in addition sets a camera to enable its position
Flexibly adjusted;
Whether detection stepless time adjustment device is rotated, and will rotate initial time, the rotation end time is sent to full-vehicle control
Device,
Stepless time adjustment device be used for by artificial rotation process come realize rain brush by be stopped to work at a slow speed it is last to
The process of rapid operation, its control to rain brush rotary speed is electrodeless;
Entire car controller is based on rotation initial time one predetermined reaction time of rollback and obtains rotating required time, in band sometimes
The rotation required time corresponding image is intercepted to obtain to image during rotation demand, in band in the video of countershaft mark
Sometimes the rotation end time corresponding image is intercepted into the image at the end of being rotated in the video of countershaft mark, will
There are image during rotation demand, the image at the end of rotation to be packaged into a file, sent out this document by telecommunication mode
Give cloud server,
Due to the presence of human response's time, it is not that user is actually needed the time that rain brush is opened to rotate initial time,
Therefore the rotation initial time for needing to obtain by being easy to detection subtracts the predetermined reaction time of characterization human response's time and obtains
To rotate required time accurately embody user actually want to rain brush open at the time of,
Stepless time adjustment is different from traditional stage speed governing, and stage speed governing is substantially instantaneous, and stepless time adjustment is to deposit
In the governing time that one can not ignore, i.e. user starts speed governing and is no longer adjusted to after speed is adjusted the speed being satisfied with to oneself
Fast this period, shows that user is satisfied with the clarity brought after the rain brush work of present speed when user's no longer speed governing,
Due to needing according to rotation required time and rotating the end time to video progress sectional drawing so that image can
The visual field of user at that time is accurately embodied, therefore should be when starting by the time of camera and the time of stepless time adjustment device
It is synchronous,
Image when entire car controller will only have a rotation demand, rotate at the end of image transmitting to cloud server, and
Can be abandoned for the image of other times, it is not necessary to store, not only reduce the burden of communication, reduce communication congestion can
Energy property, and the processing of cloud server is more succinctly facilitated,
Common telecommunication mode includes 4G, 5G, wifi;
Cloud server receives the multiple files for uploading to come from multiple entire car controllers, will have rotation in All Files
Image during demand is grouped together into rotating order training picture set, by the image sets at the end of being rotated in All Files
It is combined to form constant speed order training picture set;Cloud server will rotate each figure ordered in training picture set
As resolving into the subgraph overlapped according to certain tactic some, each subgraph is input to a nervelet net
In network, the output of the nervelet network is stored in a small ordered series of numbers, all small ordered series of numbers are arranged according to certain order
Row obtain a big ordered series of numbers, which is input in one big neutral net, so obtain can according to image judgement be
The no first nerves controller there are rotation demand;Each image in picture set is trained in constant speed order by cloud server
The subgraph overlapped according to certain tactic some is resolved into, each subgraph is input to a nervelet network
In, the output of the nervelet network is stored in a small ordered series of numbers, all small ordered series of numbers are arranged according to certain order
A big ordered series of numbers is obtained, which is input in one big neutral net, and then obtain to judge rotating speed according to image
Whether the nervus opticus controller of user demand has been met,
For each subgraph, using identical neutral net weight;
First nerves controller, nervus opticus controller difference software are melted into the first software, second soft by cloud server
Part, and the first software, the second software are issued to all entire car controllers once with cloud server communication,
The process of machine learning needs identification and the matching process of great amount of images, and conventional car side controller can not be realized,
And the cost of vehicle can be greatly improved if it can realize this process in car side upgrade controller, it is unfavorable for technology
Popularization and application, therefore, the cloud server for being alternatively provided at a distant place realizes this process, and cloud server can be according to region
Set, such as one cloud server is set in south, another cloud server is set in the north, can also be according to car type
Set, such as one cloud server be set for truck, another cloud server is set for car,
The judgement precise degrees of first, second neuron controller should be limited, i.e., only when its accuracy rate is in a fixed limit
Just by software implementation and then entire car controller can be handed down to during the value above;
Entire car controller receives the first software, the second software and is installed;Entire car controller will carry time shaft and mark
Video be input to the first software in real time, when the first software judge there are export during rotation demand rotation requirement command give vehicle control
Device processed, entire car controller are controlled rain brush to start and are rotated with a predetermined acceleration raising speed immediately when receiving rotation requirement command, together
When entire car controller the video marked with time shaft is no longer input to the first software in real time, and this is marked with time shaft
Video be input to the second software in real time, when the second software judges that rain brush rotating speed has met user demand export constant speed instruction
To entire car controller, entire car controller controls rain brush persistently to be rotated with present speed immediately when receiving constant speed instruction, and whole
Vehicle controller continues the video with time shaft mark being input to the first software and the second software in real time respectively;When first soft
Output reduction of speed is instructed to entire car controller when part judges rotation demand is not present, and entire car controller is received when reduction of speed instructs immediately
Control rain brush is rotated with a predetermined acceleration reduction of speed, when the second software judges that rain brush rotating speed no longer meets user demand, output
To entire car controller, entire car controller controls rain brush with a predetermined acceleration raising speed immediately when receiving speed-raising instruction for speed-raising instruction
Rotate, constant speed is exported when the second software judges that rain brush rotating speed has met user demand and is instructed to entire car controller, vehicle control
Device processed controls rain brush persistently to be rotated with present speed immediately when receiving constant speed instruction,
By the way that entire car controller will be handed down to after first, second neuron controller software implementation so that entire car controller is not required to
Powerful processing, calculation function is wanted to realize the identification function to image, so that automatic decision goes out to need to carry out rain brush control
The situation of system.
It should be noted that above content is that to combine specific embodiment made for the present invention further specifically
It is bright, it is impossible to assert that the embodiment of the present invention is only limitted to this, under the above-mentioned guidance of the present invention, those skilled in the art can
To carry out various improvement and deformation on the basis of above-described embodiment, and these are improved or deformation falls in protection model of the invention
In enclosing.