Summary of the invention
The object of the present invention is to provide the based reminding method and system of a kind of navigation object travelling route, realizes to remind in time and lead
Whether object travelling route of navigating is correct, reduces the purpose of user time waste.
Technical solution provided by the invention is as follows:
The present invention provides a kind of based reminding method of navigation object travelling route, comprising steps of
S1000 obtains the current location of navigation object and solicited message of asking the way;The solicited message of asking the way includes target position
It sets;
S2000 generates navigation routine according to the current location and target position;
The travelling route of S3000 acquisition navigation object;The travelling route is navigation object according to the navigation circuit, with
The current location is that starting point starts mobile route;
S4000 judges whether the travelling route matches with the navigation routine;If it is not, executing step S5000;
S5000 generates traveling error message.
Further, the step S3000 comprising steps of
S3010 obtains target image frame;The target image frame includes navigation object;
The progress feature extraction of corresponding target image frame is obtained target block diagram by several classifiers by S3020;
S3030 calculates the response of each target block diagram, and the corresponding position of the confirmation maximum target block diagram of response is to lead
The spatial position of boat object;
S3040 generates the travelling route of navigation object according to all spatial positions.
Further, before the step S3010 comprising steps of
S3001 obtains the first video image according to current detection range;
S3002 is when obtaining first video image according to the current detection range, when navigation object is not lost, to institute
It states the first video image progress image procossing and obtains the target image frame;
S3003 when navigation object is lost, expands inspection when obtaining first video image according to the current detection range
Range is surveyed, until being extended to specified detection range;
Detection range of S3004 after according to expansion is examined in preset duration according to the apparent information of navigation object again
When measuring navigation object, the second video image is obtained according to the detection range after expansion, to first video image and described
Second video image carries out image procossing and obtains the target image frame;
Detection range of S3005 after according to expansion is in preset duration, not again according to the apparent information of navigation object
When detecting navigation object, continue to expand detection range, until be extended to after specified detection range in preset duration, according to leading
When the apparent information of boat object does not detect navigation object again yet, stop the tracking to the navigation object.
Further, the step S4000 comprising steps of
The movement routine s1 of S4100 acquisition navigation object;The travelling route S=s1+s2+ ... ..., si, i ∈ N,
In, the s1 is using the current location as the first segment mobile route of starting point in the travelling route, and the s2 is the row
Second segment mobile route in route line, the si are i-th section of mobile route in the travelling route;
S4200 judges the movement routine s1 and compares gun parallax between the d1 of path whether within the scope of preset difference value;
The navigation circuit D=d1+d2+ ... ..., di, i ∈ N, wherein the d1 is in the navigation routine with the current location
For the first segment movement routine of starting point, the d2 is the second segment movement routine in the navigation routine, and the di is described leads
I-th section of movement routine in the line of air route;If it is not, executing step S5000.
Further, before the step S1000 comprising steps of
S0100 obtains user speech information, and identifies that the voice messaging obtains critical field;
S0200 judges whether the critical field includes preset path inquiry field;If so, executing step S0300;It is no
Then, return step S0100;
The direction of S0300 rotating camera is to target direction;The target direction is that the preset path inquires field pair
The direction of the user for the user speech information answered;
S0400 is determined in the corresponding default acquisition range of the camera, and captures the collected user images ruler of frame
Very little maximum user is navigation object.
The present invention also provides a kind of system for prompting of navigation object travelling route, comprising:
Data obtaining module obtains the current location of navigation object and solicited message of asking the way;The request packet of asking the way
Include target position;
Route Generation module generates navigation routine according to the current location and target position;
Route acquiring module obtains the travelling route of navigation object;The travelling route is that navigation object is led according to
Course line road starts mobile route as starting point using the current location;
Matching judgment module, judges whether the travelling route matches with the navigation routine;
Information generating module generates traveling error message when the travelling route and navigation routine mismatch.
Further, the route acquiring module includes:
Image acquisition unit obtains target image frame;The target image frame includes navigation object;
The progress feature extraction of corresponding target image frame is obtained target frame by several classifiers by block diagram acquiring unit
Figure;
Response acquiring unit, calculates the response of each target block diagram, and the maximum target block diagram of confirmation response is corresponding
Position be navigation object spatial position;
Travelling route generation unit generates the travelling route of navigation object according to all spatial positions.
Further, the route acquiring module further include:
Acquisition unit obtains the first video image according to current detection range;Also in preset duration, according to navigation pair
When the apparent information of elephant does not detect navigation object again, the second video image is obtained according to the detection range after expansion;
Control unit when navigation object is lost, is controlled when obtaining first video image according to the current detection range
It makes the acquisition unit and expands detection range, until being extended to specified detection range;
Image processing unit, when obtaining first video image according to the current detection range, navigation object is not lost
When mistake, image procossing is carried out to first video image and obtains the target image frame;
Described image processing unit, also detection range after according to expansion is in preset duration, according to navigation object
When apparent information detects navigation object again, the second video image is obtained according to the detection range after expansion, to described first
Video image and second video image carry out image procossing and obtain the target image frame;
Described image processing unit, also detection range after according to expansion is in preset duration, according to navigation object
When apparent information does not detect navigation object again, continue to expand detection range, until being extended to after specified detection range pre-
If in duration, when not detecting navigation object again yet according to the apparent information of navigation object, stop to the navigation object with
Track.
Further, the matching judgment module includes:
Path acquiring unit obtains the target movement routine s1 of navigation object;The travelling route S=s1+s2+ ... ...,
Si, i ∈ N, wherein the s1 be using the current location as the first segment mobile route of starting point in the travelling route, it is described
S2 is the second segment mobile route in the travelling route, and the si is i-th section of mobile route in the travelling route;
Comparison judgment unit judges the movement routine s1 and whether compares gun parallax between the d1 of path in preset difference value
In range;The navigation circuit D=d1+d2+ ... ..., di, i ∈ N, wherein the d1 is to be worked as in the navigation routine with described
Anteposition is set to the first segment movement routine of starting point, and the d2 is the second segment movement routine in the navigation routine, and the di is
I-th section of movement routine in the navigation routine;
The information generating module as the movement routine s1 and compares gun parallax between the d1 of path in preset difference value model
When enclosing outer, traveling error message is generated.
Further, further includes:
Voice obtains module, obtains user speech information, and identify that the voice messaging obtains critical field;
Identification module judges whether the critical field includes preset path inquiry field;
The voice obtains module, when the critical field does not include preset path inquiry field, reacquires new
User speech information;
Rotary module, when the critical field includes preset path inquiry field, the direction of rotating camera to target
Direction;The target direction is the direction for the user that the preset path inquires the corresponding user speech information of field;
Navigation object determining module determines in the corresponding default acquisition range of the camera, and captures frame size most
The collected user of big correspondence is navigation object.
The based reminding method and system of a kind of navigation object travelling route provided through the invention, can bring it is following at least
It is a kind of the utility model has the advantages that
1) of the invention since the travelling route to navigation object is compared with navigation routine, navigation pair can be reminded in time
As whether travelling route is correct, user time waste is reduced, the navigation usage experience of user is prompted.
2) the progress feature extraction of corresponding target image frame is obtained target block diagram, root by several classifiers by the present invention
The spatial position of navigation object is determined according to the response of each target block diagram to generate the travelling route of navigation object, Neng Gouke
The interference for taking external environment promotes the reliability and accuracy of the tracing detection of navigation object.
3) present invention obtain a multilevel iudge navigation routine and travelling route using current location as before starting point first
Section can judge whether the direction of travel of navigation object and travel path are correct, can accelerate the efficiency of discriminatory analysis, thus
It reduces and generates the time that navigation object is reminded in traveling error message, further speed up prompt efficiency.
Specific embodiment
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, Detailed description of the invention will be compareed below
A specific embodiment of the invention.It should be evident that drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing, and obtain other embodiments.
To make simplified form, part related to the present invention is only schematically shown in each figure, they are not represented
Its practical structures as product.In addition, there is identical structure or function in some figures so that simplified form is easy to understand
Component only symbolically depicts one of those, or has only marked one of those.Herein, "one" is not only indicated
" only this ", can also indicate the situation of " more than one ".
A kind of one embodiment of the based reminding method of navigation object travelling route of the present invention, as shown in Figure 1, comprising:
S1000 obtains the current location of navigation object and solicited message of asking the way;The solicited message of asking the way includes target position
It sets;
S2000 generates navigation routine according to the current location and target position;
The travelling route of S3000 acquisition navigation object;The travelling route is navigation object according to the navigation circuit, with
The current location is that starting point starts mobile route;
S4000 judges whether the travelling route matches with the navigation routine;If it is not, executing step S5000;
S5000 generates traveling error message.
Specifically, the navigation equipments such as navigating robot obtain the current location of navigation object, navigation object in the present embodiment
Current location acquisition modes can be GPS positioning acquisition, can also multiple camera shooting head nodes by three-point fox method acquisition,
Any indoor positioning mode is in protection scope of the present invention, also, the navigation equipments such as navigating robot also obtain navigation pair
As the solicited message of asking the way of input, solicited message of asking the way can be voice input and obtain, and be also possible to user in human-computer interaction circle
Face is manually entered acquisition, after navigation equipment gets the current location of navigation object and solicited message of asking the way, according to current location
Path planning, which is carried out, with target position generates navigation routine, navigation object starts to be moved to target position according to navigation routine, when
When navigation object leaves from current location and moved along the guide direction of navigation routine, navigation equipment acquires navigation in real time
The travelling route of object is compared analysis with navigation routine to the travelling route of navigation object, judges travelling route and navigation
Whether route matches, and navigation equipment can not generate any prompt information if travelling route is matched with navigation routine, can also
To generate traveling correct information, inform that the direction of travel of navigation object and travel path are correct;If travelling route and navigation road
Line mismatch then generates traveling error message, to remind the direction of travel and travel path mistake of navigation object.The present invention can
It reminds navigation object travelling route whether correct in time, reduces user time waste, prompt the navigation usage experience of user.
A kind of another embodiment of the based reminding method of navigation object travelling route of the present invention, as shown in Figure 2, comprising:
S1000 obtains the current location of navigation object and solicited message of asking the way;The solicited message of asking the way includes target position
It sets;
S2000 generates navigation routine according to the current location and target position;
S3010 obtains target image frame;The target image frame includes navigation object;
The progress feature extraction of corresponding target image frame is obtained target block diagram by several classifiers by S3020;
S3030 calculates the response of each target block diagram, and the corresponding position of the confirmation maximum target block diagram of response is to lead
The spatial position of boat object;
S3040 generates the travelling route of navigation object according to all spatial positions;
S4000 judges whether the travelling route matches with the navigation routine;If it is not, executing step S5000;
S5000 generates traveling error message.
Specifically, in the present embodiment, it is complete to carry out navigation object inspection based on YOLO V2 algorithm and KCF algorithm
It surveys and tracks, the purpose of trace detection is realized in navigation equipment, specifically:
1. navigation object detecting and tracking
Navigation object detection is carried out to the picture frame by YOLO V2 algorithm, i.e. YOLO V2 uses 32 layers of neural network
Structure (including convolutional layer and pond layer) carries out pre-training detection to picture frame using the Web portal of 416 × 416 resolution ratio,
Pre-training detection is carried out, (frame of prolate is less, and tall and thin frame is more, to meet the spy of people using the box dimensions of 5 kinds of sizes
Sign) location prediction is carried out, the target block diagram (such as resolution ratio is 26*26) of shallow-layer is connected to the target block diagram (resolution ratio of deep layer
13*13), height Resolutions are obtained characteristic pattern and link by this connection type, are superimposed adjacent feature to different channels
The feature map of 26*26*512 is become 13*13* similar to " the quick connection " of ResNet by (rather than being spatial position)
Then 2048 feature map is connected with original further feature figure, it is defeated just to change model every several wheels (10batches)
Enter size, due to the method using multiple dimensioned training detection, has good robustness.After changing mode input size, continue
Model is trained, this trained mechanism forces how 32 layers of Structural Learning of Neural Networks is carried out in various input dimensions
Prediction, it means that identical network can predict the testing result under different resolution.Because YOLO V2 model is in small scale
Quickly, therefore, YOLO V2 provides a kind of in the side that speed and accuracy is weighed between the two the speed of service in input
Case in the case that YOLO V2 is able to maintain accuracy rate in the detection of small image in different resolution, promotes processing speed.
2. navigation object detecting and tracking
It is discriminate tracking by KCF algorithm, the training classifier in tracing process removes detection next frame using classifier
Whether predicted position is navigation object, then reuses new testing result and goes to update training set and then update classifier.And it is instructing
It is positive sample that navigation object region is generally chosen when practicing classifier, and the peripheral region of navigation object is negative sample, certainly closer to
A possibility that region of navigation object is positive sample is bigger, acquires positive negative sample, benefit using the circular matrix of target peripheral region
With ridge regression training classifier, and successfully using circular matrix Fourier space diagonalizable property by the operation of matrix
It is converted into the dot product of element, greatly reduces operand, improves arithmetic speed, algorithm is made to meet requirement of real-time.It will be linear
The ridge regression in space is mapped to non-linear space by kernel function, in non-linear space by solving a dual problem and certain
Common constraint, can also use the diagonalization of circular matrix Fourier space can also simplify calculating.
3. navigation object location determination
Judge that is traced into is the background information of navigation object or surrounding, it is main that sample is adopted using rotation matrix
Collection, using fast Fourier variation acceleration calculating is carried out to algorithm, tracked based on the navigation object detected, first with
Navigation object is detected before track, the position of navigation object is obtained, then navigation object is learnt, is tracked, then
The response of each target block diagram is calculated, the corresponding position of the confirmation maximum target block diagram of response is the space bit of navigation object
It sets.Start to move with current location according to navigation routine as shown in Fig. 2, obtaining navigation object, navigation equipment shooting starts to obtain
Image/video, the image on the left side Fig. 3 are current image frame P1, confine navigation object using dotted line frame 6 to current image frame P1, obtain
Pixel coordinate Q1 of the navigation object on image is taken, solid box 3 is exactly the sample object block diagram for including navigation object, other
The corresponding block diagram of solid box (such as solid box 1, solid box 2, solid box 4, solid box 5) is exactly to recycle sample object block diagram
The sample that alignment navigation object obtains after displacement, goes out a classifier by these sample trainings, after classifier training is good,
Next image frame P2, that is, the image on the right of Fig. 3 have been come, it is corresponding in sample object block diagram i.e. solid box 3 first
Region is sampled, and then carries out cyclic shift to the sampling, is aligned sub as the image on the right of Fig. 3 is shown after target
(being easy to understand herein, be misaligned in practice) calculates response to target block diagram respectively using classifier, it is clear that solid box 1 is right
The corresponding response of the target picture frame answered is maximum, is calculated by the position of the corresponding target picture frame of solid box 1 and obtains navigation object
Then pixel coordinate Q2 on image switches next image frame Pj, j ∈ N, continues above-mentioned steps measure and navigation pair
As the pixel coordinate Q3 on image is navigated according to the transformational relation between world coordinate system and image coordinate system
Coordinate position (including the corresponding coordinate position M1, that is, current location pixel coordinate Q1, the corresponding coordinate of pixel coordinate Q2 of object
The corresponding coordinate position Mj of position M2, pixel coordinate Qj), the traveling road for generating navigation object is drawn according to all coordinate positions
Line.The present embodiment under tracking-detection framework, using YOLO V2 and KCF algorithm improve follow-up investigations accuracy, efficiency and
Reliability while can naturally utilize multiple feature channels, detect in network every layer since KCF has the characteristics that fireballing
Feature port number it is all more, since abstractness of the different layers in neural network structure for the description of target is different, bottom
The feature of layer is simple, and the high-rise feature of semanteme is more suitable positioning more, and due to during tracking target do not stop to change, with
The degree of difficulty of track can change, and low-level image feature may be just inaccurate, cascade multiple trackers at this time;Certainly, if shallow-layer
Tracker can track well, and tracking effect is preferable, then just not carry out it is subsequent calculate, the time can be saved, with
It is to see the response currently calculated that whether track device, which does very well, and the more big then tracking effect of response is better, therefore selects neural network
The multiple KCF of different layer buildings are cascaded in structure, are established multiple Individual classifiers respectively using multiple network layers, are mentioned
The reliability and accuracy of high navigation object tracking.
A kind of another embodiment of the based reminding method of navigation object travelling route of the present invention, as shown in Figure 4, comprising:
S1000 obtains the current location of navigation object and solicited message of asking the way;The solicited message of asking the way includes target position
It sets;
S2000 generates navigation routine according to the current location and target position;
S3001 obtains the first video image according to current detection range;
S3002 is when obtaining first video image according to the current detection range, when navigation object is not lost, to institute
It states the first video image progress image procossing and obtains the target image frame;
S3003 when navigation object is lost, expands inspection when obtaining first video image according to the current detection range
Range is surveyed, until being extended to specified detection range;
Detection range of S3004 after according to expansion is examined in preset duration according to the apparent information of navigation object again
When measuring navigation object, the second video image is obtained according to the detection range after expansion, to first video image and described
Second video image carries out image procossing and obtains the target image frame;
Detection range of S3005 after according to expansion is in preset duration, not again according to the apparent information of navigation object
When detecting navigation object, continue to expand detection range, until be extended to after specified detection range in preset duration, according to leading
When the apparent information of boat object does not detect navigation object again yet, stop the tracking to the navigation object.
S3010 obtains target image frame;The target image frame includes navigation object;
The progress feature extraction of corresponding target image frame is obtained target block diagram by several classifiers by S3020;
S3030 calculates the response of each target block diagram, and the corresponding position of the confirmation maximum target block diagram of response is to lead
The spatial position of boat object;
S3040 generates the travelling route of navigation object according to all spatial positions;
S4000 judges whether the travelling route matches with the navigation routine;If it is not, executing step S5000;
S5000 generates traveling error message.
Specifically, the camera of navigation equipment carries out the first video image acquisition with current detection range in the present embodiment,
Judge the case where whether navigation object will appear loss when obtaining the first video image in real time, the reason of navigation object loss has more
Kind, it may be possible to because of classifier missing inspection, it is also possible to the navigation caused by the blocking of the background object (such as wall, trees etc.) in scene
Image tracing is lost.When navigation object is not lost, image procossing directly is carried out to the first video image and obtains target image frame;
When active out, expand detection range when navigation object is lost, according to the apparent information of navigation object, apparent information is to lead
Distinguishing characteristics between object and the background of navigating, apparent information include navigation object, head position, volume size, height size,
The colour of skin, hair style, clothes color, clothes Texture eigenvalue judge whether weigh in preset duration according to the detection range after expansion
Newly navigation object is detected, if detecting the navigation object of loss again in preset duration, according to the detection after expansion
Range obtains the second video image, carries out image procossing to the first video image and the second video image and obtains target image frame;
If the navigation object of loss is also not detected in preset duration, continue to expand detection range, until the inspection after expanding
Until survey range reaches specified detection range, if after reaching specified detection range, can't detect and lose in preset duration
The navigation object of mistake does not then carry out tracing detection to the navigation object.Usually within a certain period of time, the navigation object meeting of loss
It is reappeared in a certain range near the disappearance position, so the purpose in this stage is exactly to pick up the navigation pair lost
As processing method is to retain the navigation object lost for a period of time, and compare in a certain range of region in disappearance position and lose
Apparent information similarity degree between the navigation object of mistake and the object detected again, if similitude is greater than certain threshold value,
Illustrate that the object detected again within the scope of the detection block is the navigation object to disappear before.The present invention is avoided due to navigation pair
After losing, it can not continue to carry out tracing detection problem to navigation object, improve the robustness of navigation object travel line detection.
A kind of another embodiment of the based reminding method of navigation object travelling route of the present invention, as shown in Figure 5, comprising:
S1000 obtains the current location of navigation object and solicited message of asking the way;The solicited message of asking the way includes target position
It sets;
S2000 generates navigation routine according to the current location and target position;
S3001 obtains the first video image according to current detection range;
S3002 is when obtaining first video image according to the current detection range, when navigation object is not lost, to institute
It states the first video image progress image procossing and obtains the target image frame;
S3003 when navigation object is lost, expands inspection when obtaining first video image according to the current detection range
Range is surveyed, until being extended to specified detection range;
Detection range of S3004 after according to expansion is examined in preset duration according to the apparent information of navigation object again
When measuring navigation object, the second video image is obtained according to the detection range after expansion, to first video image and described
Second video image carries out image procossing and obtains the target image frame;
Detection range of S3005 after according to expansion is in preset duration, not again according to the apparent information of navigation object
When detecting navigation object, continue to expand detection range, until be extended to after specified detection range in preset duration, according to leading
When the apparent information of boat object does not detect navigation object again yet, stop the tracking to the navigation object.
S3010 obtains target image frame;The target image frame includes navigation object;
The progress feature extraction of corresponding target image frame is obtained target block diagram by several classifiers by S3020;
S3030 calculates the response of each target block diagram, and the corresponding position of the confirmation maximum target block diagram of response is to lead
The spatial position of boat object;
S3040 generates the travelling route of navigation object according to all spatial positions;
The movement routine s1 of S4100 acquisition navigation object;The travelling route S=s1+s2+ ... ..., si, i ∈ N,
In, the s1 is using the current location as the first segment mobile route of starting point in the travelling route, and the s2 is the row
Second segment mobile route in route line, the si are i-th section of mobile route in the travelling route;
S4200 judges the movement routine s1 and compares gun parallax between the d1 of path whether within the scope of preset difference value;
The navigation circuit D=d1+d2+ ... ..., di, i ∈ N, wherein the d1 is in the navigation routine with the current location
For the first segment movement routine of starting point, the d2 is the second segment movement routine in the navigation routine, and the di is described leads
I-th section of movement routine in the line of air route;If it is not, executing step S5000;
S5000 generates traveling error message.
Specifically, above-described embodiment is the complete travelling route of tracing detection navigation object always, traveling road herein
Line is the travelling route of the collected user in effective acquisition range of the image capture device (such as camera) of navigation equipment,
Matching is compared with navigation routine to complete travelling route, and the present embodiment, as shown in fig. 6, obtaining the row of navigation object
It first segment route, that is, movement routine s1 in route line and the first segment route of navigation circuit and compares path d1 and is compared and sentence
It is disconnected, it is capable of capable of sentencing by first segment before starting point of current location for a multilevel iudge navigation routine and travelling route in this way
Whether the direction of travel and travel path of disconnected navigation object are correct, can accelerate the efficiency of discriminatory analysis, to reduce generation
The time of navigation object is reminded in traveling error message, further speeds up prompt efficiency, and navigation object travelling route is reminded to be in time
It is no correct, user time waste is reduced, the navigation usage experience of user is prompted.
A kind of one embodiment of the system for prompting of navigation object travelling route of the present invention, as shown in fig. 7, comprises:
Data obtaining module 100 obtains the current location of navigation object and solicited message of asking the way;The solicited message of asking the way
Including target position;
Route Generation module 200 generates navigation routine according to the current location and target position;
Route acquiring module 300 obtains the travelling route of navigation object;The travelling route is navigation object according to
Navigation circuit starts mobile route as starting point using the current location;
Matching judgment module 400, judges whether the travelling route matches with the navigation routine;
Information generating module 500 generates mistake letter of advancing when the travelling route and navigation routine mismatch
Breath.
Specifically, the present embodiment is the corresponding system embodiment of above method embodiment, specific effect is referring to above-mentioned correspondence
Embodiment of the method, this is no longer going to repeat them.
A kind of another embodiment of the system for prompting of navigation object travelling route of the present invention, as shown in Figure 8, comprising:
Data obtaining module 100 obtains the current location of navigation object and solicited message of asking the way;The solicited message of asking the way
Including target position;
Route Generation module 200 generates navigation routine according to the current location and target position;
Route acquiring module 300 obtains the travelling route of navigation object;The travelling route is navigation object according to
Navigation circuit starts mobile route as starting point using the current location;
Matching judgment module 400, judges whether the travelling route matches with the navigation routine;
Information generating module 500 generates mistake letter of advancing when the travelling route and navigation routine mismatch
Breath;
Route acquiring module 300 includes:
Image acquisition unit 310 obtains target image frame;The target image frame includes navigation object;
The progress feature extraction of corresponding target image frame is obtained mesh by several classifiers by block diagram acquiring unit 320
Mark block diagram;
Response acquiring unit 330 calculates the response of each target block diagram, confirms the maximum target block diagram pair of response
The position answered is the spatial position of navigation object;
Travelling route generation unit 340 generates the travelling route of navigation object according to all spatial positions.
Preferably, route acquiring module 300 further include:
Acquisition unit 350 obtains the first video image according to current detection range;Also in preset duration, according to leading
When the apparent information of boat object does not detect navigation object again, the second video image is obtained according to the detection range after expansion;
Control unit 360, when obtaining first video image according to the current detection range, navigation object is lost
When, it controls the acquisition unit and expands detection range, until being extended to specified detection range;
Image processing unit 370, when obtaining first video image according to the current detection range, navigation object is not
When loss, image procossing is carried out to first video image and obtains the target image frame;
Described image processing unit 370, also detection range after according to expansion is in preset duration, according to navigation object
Apparent information when detecting navigation object again, the second video image is obtained according to the detection range after expansion, to described the
One video image and second video image carry out image procossing and obtain the target image frame;
Described image processing unit 370, also detection range after according to expansion is in preset duration, according to navigation object
Apparent information when not detecting navigation object again, continue to expand detection range, until after being extended to specified detection range
In preset duration, when not detecting navigation object again yet according to the apparent information of navigation object, stop to the navigation object
Tracking.
Preferably, the matching judgment module 400 includes:
Path acquiring unit 410 obtains the target movement routine s1 of navigation object;The travelling route S=s1+s2
+ ... ..., si, i ∈ N, wherein the s1 is in the travelling route using the current location as the mobile road of the first segment of starting point
Line, the s2 are the second segment mobile route in the travelling route, and the si is i-th section of mobile road in the travelling route
Line;
Whether comparison judgment unit 420, the gun parallax for judging the movement routine s1 and comparing between the d1 of path are being preset
In difference range;The navigation circuit D=d1+d2+ ... ..., di, i ∈ N, wherein the d1 in the navigation routine with institute
The first segment movement routine that current location is starting point is stated, the d2 is the second segment movement routine in the navigation routine, described
Di is i-th section of movement routine in the navigation routine;
The information generating module 500 as the movement routine s1 and compares gun parallax between the d1 of path default poor
When being worth outside range, traveling error message is generated.
Preferably, further includes:
Voice obtains module 600, obtains user speech information, and identify that the voice messaging obtains critical field;
Identification module 700 judges whether the critical field includes preset path inquiry field;
The voice obtains module 600, when the critical field does not include preset path inquiry field, reacquires new
User speech information;
Rotary module 800, when the critical field includes preset path inquiry field, the direction of rotating camera to mesh
Mark direction;The target direction is the direction for the user that the preset path inquires the corresponding user speech information of field;
Navigation object determining module 900 determines in the corresponding default acquisition range of the camera, and captures frame size
The maximum collected user of correspondence is navigation object.
Specifically, the present embodiment is the corresponding system embodiment of above method embodiment, specific effect is referring to above-mentioned correspondence
Embodiment of the method, this is no longer going to repeat them.
According to above-described embodiment, example cites an actual example, as shown in figure 9,
In the more complex indoor spaces of the environment such as market, hospital, station, airport, navigating robot can be navigation
The path that object provides specified target position indicates service.
The main contents of the task are summarized are as follows:
1) navigation object inquires locality (such as some outlet, some dining room, toilet nearby to navigating robot
Deng) track route;
2) navigating robot provides the navigation routine that target position is reached from local i.e. current location according to the map;
3) navigating robot is to navigation object, that is, navigation object modeling tracking a distance, if navigation object this section away from
The comparison path that robot provides is not met from interior movement routine, to navigation object will correct mentioning using voice prompting
Show.
That is the process of event development is that navigation object inquires route, is walked according to the navigation routine that navigating robot provides,
Navigating robot tracks navigation object, compares the track i.e. travelling route and the provided navigation of navigation instrument of navigation object
The consistency of route, wherein the difficult point of target tracking stage is: closer close to navigating robot when navigation object is inquired
And be front, it is thereafter then the tracking of rear whole body to navigation object tracking, it is therefore desirable to cope with the variation of target scale, target
Rotation, the variation of target appearance also need to overcome the interference of other pedestrians due to being the place more intensive in crowd,
Detecting and tracking can be carried out using KCF algorithm combination YOLO V2 algorithm herein.
In practice, navigation routine usually will include multistage, and in order to promote the efficiency of navigating robot work, model is corrected in tracking
It encloses for movement routine s1, as when navigation object, direction of travel and travel path do not malfunction in movement routine s1, will no longer
Tracking correction task is carried out to the navigation object.
It should be noted that above-described embodiment can be freely combined as needed.The above is only of the invention preferred
Embodiment, it is noted that for those skilled in the art, in the premise for not departing from the principle of the invention
Under, several improvements and modifications can also be made, these modifications and embellishments should also be considered as the scope of protection of the present invention.