CN105335377B - Information processing method and equipment - Google Patents
Information processing method and equipment Download PDFInfo
- Publication number
- CN105335377B CN105335377B CN201410291261.3A CN201410291261A CN105335377B CN 105335377 B CN105335377 B CN 105335377B CN 201410291261 A CN201410291261 A CN 201410291261A CN 105335377 B CN105335377 B CN 105335377B
- Authority
- CN
- China
- Prior art keywords
- feature descriptor
- node
- feature
- increased
- information processing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Processing Or Creating Images (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Disclose information processing method and equipment.The information processing method, applied to an information processing equipment, three-dimensional map in the information processing equipment comprising particular space environment is pre-created, terminal device in the particular space environment can determine the position of oneself according to the three-dimensional map, and described method includes following steps: obtain the image shot by the terminal device；The characteristic point in described image is extracted, the feature descriptor for characterizing the characteristic point is obtained；Depth information in position and described image based on the terminal device obtains the three-dimensional position of the characteristic point；Determine the minimum space node in the particular space environment；It according to the three-dimensional position of each characteristic point obtained, determines that minimum space node is belonging to each characteristic point to be clustered, obtains all feature descriptors that minimum space inter-node is included；And each minimum space node is managed based on specific data structure.
Description
Technical field
The present invention relates to information processing method and equipment, more specifically to easily and fast capable of managing and update
The information processing method and equipment of feature database.
Background technique
It, can be by positioning immediately and map structuring (Simultaneous Localization for a circumstances not known
And Mapping, SLAM) technology etc. builds up three-dimensional map in advance.In the environment for having built up figure, mobile terminal device is (e.g.,
Robot) match with the characteristic point of global map obtaining itself by the characteristic point in the image that shoots itself
Location information.In robot self-localization, since robot scene is continually changing, and data of the feature database of global map
Amount is very big, therefore the characteristic point progress in the feature database of the characteristic point and global map in the image that robot itself is shot
With very time-consuming.In addition, the operation that the feature database of global map is updated according to the image of robot itself shooting nor
It is often time-consuming.It may relate to complicated data manipulation in addition, being extended to the feature database of global map.
Summary of the invention
In view of above situation, it is intended to provide fast and convenient feature database management and update method.That is, the application to be solved
The problem of be how easily and fast to manage and update feature database.
According to an aspect of the invention, there is provided a kind of information processing method, is applied to an information processing equipment, in institute
It states comprising the three-dimensional map that particular space environment is pre-created in information processing equipment, in the particular space environment
Terminal device can determine the position of oneself according to the three-dimensional map, and described method includes following steps: obtain by the end
One image of end equipment shooting；The characteristic point in described image is extracted, the feature descriptor for characterizing the characteristic point is obtained；
Depth information in position and described image based on the terminal device obtains the three-dimensional position of the characteristic point；Determine institute
State the minimum space node in particular space environment；According to the three-dimensional position of each characteristic point obtained, each characteristic point institute is determined
The minimum space node of category obtains all feature descriptors that minimum space inter-node is included to be clustered；And it is based on
Specific data structure manages each minimum space node.
Preferably, in information processing method according to an embodiment of the present invention, the specific data structure is Octree knot
Structure, and wherein the external cube of minimum of the particular space environment is the root node of the octree structure, then this
Minimum external cube is divided into eight first order sub- sections of identical first sub-cube of eight sizes as the root node
Then each of eight first sub-cubes are divided into identical second sub-cube of eight sizes as institute respectively by point
Eight second level child nodes of first order child node are stated, repeat such segmentation until minimum space node.
Preferably, information processing method according to an embodiment of the present invention may further include: in all features of acquisition
Among descriptor, the feature descriptor of first threshold quantity is selected.
Preferably, in information processing method according to an embodiment of the present invention, each feature descriptor has corresponding
Feature weight is used to indicate the characteristic strength of feature descriptor.
Preferably, information processing method according to an embodiment of the present invention may further include: judge whether there is for
Specific node needs increased feature descriptor；If the judgment is Yes, then calculate separately feature descriptor to be increased with it is described
Similarity between the existing each feature descriptor of specific node, wherein if similarity is greater than second threshold, then it is assumed that the two
It is similar；If there is feature descriptor similar with feature descriptor to be increased in existing each feature descriptor, it is determined that
Without increasing.
Preferably, information processing method according to an embodiment of the present invention may further include: judge whether there is for
Specific node needs increased feature descriptor；If the judgment is Yes, then calculate separately feature descriptor to be increased with it is described
Similarity between the existing each feature descriptor of specific node, wherein if similarity is greater than second threshold, then it is assumed that the two
It is similar；If feature descriptor similar with feature descriptor to be increased is not present in existing each feature descriptor, sentence
Whether the quantity of the feature descriptor at the specific node of breaking is less than first threshold；If the judgment is Yes, it is determined that increased
Add；If the judgment is No, then it is replaced with feature descriptor to be increased special in the existing each feature descriptor of specific node
The smallest feature descriptor of weight is levied, and the weight of the feature descriptor newly increased is set as initial value.
According to another aspect of the present invention, a kind of information processing equipment is provided, for being pre-created three to specific environment
Map is tieed up, in order to which the terminal device in the particular space environment can determine the position of oneself according to the three-dimensional map
It sets, the equipment includes: storage unit, for storing the three-dimensional map being pre-created；Communication unit, for obtaining by the end
One image of end equipment shooting；Extraction unit is obtained for extracting the characteristic point in described image for characterizing the characteristic point
Feature descriptor；Three-dimensional position determination unit, for the depth letter in position and described image based on the terminal device
Breath, obtains the three-dimensional position of the characteristic point；Cluster cell, for determining the minimum space knot in the particular space environment
Point, and the three-dimensional position of each characteristic point obtained according to the three-dimensional position determination unit, determine minimum belonging to each characteristic point
Space node obtains all feature descriptors that minimum space inter-node is included to be clustered；And control unit, so that
The storage unit is based on specific data structure and manages each minimum space node.
Preferably, in information processing equipment according to an embodiment of the present invention, the specific data structure is Octree knot
Structure, and wherein the external cube of minimum of the particular space environment is the root node of the octree structure, then this
Minimum external cube is divided into eight first order sub- sections of identical first sub-cube of eight sizes as the root node
Then each of eight first sub-cubes are divided into identical second sub-cube of eight sizes as institute respectively by point
Eight second level child nodes of first order child node are stated, repeat such segmentation until minimum space node.
Preferably, information processing equipment according to an embodiment of the present invention may further include: selecting unit, in institute
Among all feature descriptors for stating cluster cell acquisition, the feature descriptor of first threshold quantity is selected.
Preferably, information processing equipment according to an embodiment of the present invention may further include:
Weighted units are used to indicate feature description for assigning corresponding feature weight for each feature descriptor
The characteristic strength of symbol.
Preferably, in information processing equipment according to an embodiment of the present invention, described control unit is judged whether there is pair
Increased feature descriptor is needed in specific node；If the judgment is Yes, then described control unit calculates separately spy to be increased
The similarity between descriptor and the existing each feature descriptor of the specific node is levied, wherein if similarity is greater than the second threshold
Value, then it is assumed that the two is similar；If there is feature similar with feature descriptor to be increased in existing each feature descriptor
Descriptor, then described control unit is determined without increasing.
Preferably, in information processing equipment according to an embodiment of the present invention, described control unit is judged whether there is pair
Increased feature descriptor is needed in specific node；If the judgment is Yes, then described control unit calculates separately spy to be increased
The similarity between descriptor and the existing each feature descriptor of the specific node is levied, wherein if similarity is greater than the second threshold
Value, then it is assumed that the two is similar；If spy similar with feature descriptor to be increased is not present in existing each feature descriptor
Descriptor is levied, then described control unit judges whether the quantity of the feature descriptor at the specific node is less than first threshold；
If the judgment is Yes, then described control unit determination is increased；If the judgment is No, then described control unit with to be increased
Feature descriptor replaces the smallest feature descriptor of feature weight in the existing each feature descriptor of specific node, and will be new
The weight of increased feature descriptor is set as initial value.
In information processing method according to an embodiment of the present invention and information processing equipment, proposed based on Octree applicable
In the data structure of 3d space feature database management.Data structure in this way more easily and fast can be managed and be updated
Feature database, and self-positioning algorithm can be enhanced to the adaptability of environmental change.
Detailed description of the invention
Fig. 1 is to show the flow chart of the process of information processing method according to an embodiment of the present invention；
Fig. 2A and 2B respectively illustrates the space structure and tree-like storage form of Octree；
Fig. 3 is to show the first exemplary flow chart for determining processing；
Fig. 4 is to show the second exemplary flow chart for determining processing；And
Fig. 5 is to show the functional block diagram of the configuration of information processing equipment according to an embodiment of the present invention.
Specific embodiment
Each preferred embodiment of the invention is described below with reference to accompanying drawings.It provides referring to the drawings
Description, to help the understanding to example embodiment of the invention as defined by appended claims and their equivalents.It includes side
The various details of assistant's solution, but they can only be counted as illustratively.Therefore, it would be recognized by those skilled in the art that
Embodiment described herein can be made various changes and modifications, without departing from scope and spirit of the present invention.Moreover, in order to
Keep specification more clear succinct, by omission pair it is well known that the detailed description of function and construction.
Firstly, information processing method according to an embodiment of the present invention will be described referring to Fig.1.The information processing method application
Three-dimensional map in an information processing equipment, in the information processing equipment comprising particular space environment is pre-created.Place
Terminal device (e.g., robot) in the particular space environment can determine the position of oneself according to the three-dimensional map.
As shown in Figure 1, described method includes following steps:
Firstly, obtaining the image shot by the terminal device in step S101.Here image can be grayscale image
Picture, naturally it is also possible to colored (RGB) image.
Then, in step S102, the characteristic point in described image is extracted, acquisition is retouched for characterizing the feature of the characteristic point
State symbol.In general, characteristic point is typically selected to have the point of position characteristics, such as vertex or endpoint.There are many Feature Points Extraction,
Such as Harris angle point, DOG extreme value etc..Since this is not directly related to the present invention, its details is not described in detail.
Once it is determined that characteristic point, be in next step to the corresponding descriptor of each feature point extraction, with distinguish different characteristic point and
With identical characteristic point.Ideal feature descriptor, which will meet, has centainly constant to scale, rotation, even affine etc. transformation
Property；To insensitive for noise；There is good selectivity, that is, the feature descriptor correlation for corresponding to different characteristic points wants small, in this way
Different characteristic points can just be efficiently differentiated.As example that is a kind of most simple and being easy to understand, the feature descriptor can be with
It is the gray value of characteristic point.
Next, in step S103, depth information in position and described image based on the terminal device obtains institute
State the three-dimensional position of characteristic point.Here, it should be noted that premise of the invention is that specific environment has been completed dimensionally
Figure and the further creation of feature database, the terminal device can pass through the characteristic point and feature database in the image of itself shooting
In characteristic point matching to determine the position where itself.Since this is not directly related to the present invention, to its details
It is not described in detail.Here the position of the terminal device obtained is the three-dimensional coordinate under global coordinate system.It is set according to terminal
The described image of standby shooting and depth information wherein included, can obtain each characteristic point using the terminal device as coordinate
Position under the local coordinate system of origin.Herein on basis, in conjunction with the terminal device global coordinate system under three-dimensional sit
Mark, can be by three-dimensional coordinate of each characteristic point in the case where the three-dimensional coordinate transformation under local coordinate system is global coordinate system.
Then, it in step S104, determines the minimum space node in the particular space environment, that is, determines the specific sky
Between spatial resolution in environment.That is, dividing the particular space environment with minimum space node.Certainly, here
Minimum space node be the space with the designated volume of corresponding three-dimensional coordinate.Details in relation to minimum space node will
It is described later on.
After step S104 determines minimum space node, processing proceeds to step S105.In step S105, according to
The three-dimensional position of step S103 each characteristic point obtained determines that minimum space node is belonging to each characteristic point to be clustered,
Obtain all feature descriptors that minimum space inter-node is included.Specifically, it is assumed that minimum space node a is eight endpoints
Respectively cube of (0,0,0), (0,0,1), (0,1,0), (0,1,1), (1,0,0), (1,1,0), (1,1,1), (1,0,1)
Body, i.e. side length are 1 and an endpoint is located at the cube of coordinate origin, if that there are three characteristic points, three-dimensional position
Respectively (1,1,0), (1/2,1/2,1/4), (0,1,1), then these three characteristic points are all fallen in minimum space node a, i.e., this
Minimum space node belonging to three characteristic points is minimum space node a, and by these three feature points clusterings to minimum space
Node a.At this time, it may be necessary to, it is noted that although the position of these three characteristic points is not identical, after cluster, by these three
The three-dimensional position of characteristic point sees an identical three-dimensional position as, for example, regarding the central point of minimum space node a as
(1/2,1/2,1/2).Also, feature descriptor corresponding to these three characteristic points regards the feature of minimum space node a as
Descriptor.
Finally, managing each minimum space node based on specific data structure in step S106.For example, pipe described here
Reason may include retrieval, maintenance, increase, deletion etc..
In a flow diagram in figure 1, it illustrates only in a characteristic point of one place and the clustering processing of feature descriptor.
The feature database of the three-dimensional map of entire space environment can by multiple and different positions, repeatedly step S101~S106 and
It is continuously updated.
For example, the specific data structure can be tree.By managing each minimum space node with tree,
Compared with ordinary construction, desired locations can be retrieved more quickly, to extract feature descriptor therein or update therein
Feature descriptor, and existing feature database can be extended simplerly.
More specifically, the specific data structure is octree structure.The data structure of Octree indicates method by sky
Between position enumerative technique develop, be a kind of hierarchical data structure.Fig. 2A -2B respectively illustrate Octree space structure and
Tree-like storage form.The external square of minimum of space environment is constructed first.As shown in Figure 2 A, maximum cube can regard sky as
Between environment the external square of minimum.It corresponds to Fig. 2 B, the external cube of minimum of the particular space environment is eight fork
The root node of tree construction.Then the external cube of the minimum is divided into identical first sub-cube of eight sizes as described in
Each of eight first sub-cubes, are then divided into eight size phases respectively by eight first order child nodes of root node
Eight second level child nodes of the second same sub-cube as the first order child node repeat such segmentation until minimum
Until the node of space.In fig. 2, the small cubes for being denoted as black are minimum space node.
In addition, for child nodes at different levels, it is understood that there may be three kinds of states.The first state, as indicated in Fig. 2 B with black
Shown in box, to include the one or more features point clustered in the step S105 being described above.Second of state is such as schemed
With shown in the box of blank mark in 2B, to be not included in any one feature clustered in the above step S105
Point.The third state, as with shown in stain, being undeveloped space node in Fig. 2 B.
The case where for ease of description and drawing, illustrating only three-level segmentation in figs. 2a, 2b.But the skill of this field
For art personnel it should be understood that for actual particular space environment, the series of segmentation may be far longer than three-level.
As mentioned above it is possible, the tree based on Octree, the present invention devises feature database storage method, quickly to obtain
The feature of local space is taken, convenient for retrieval and maintenance.In the feature database based on Octree, it may include at each space node
Multiple feature descriptors.In order to adapt to diversified space characteristics demand, while the memory loss of control structure, by each node
The quantity of the feature descriptor at place is limited to for example, N.That is, embodiment more preferably, after step S105
Further comprise following steps: among all feature descriptors of acquisition, selecting the feature descriptor of first threshold quantity.
In addition, way of example more preferably, each feature descriptor can have corresponding feature weight, use
In the characteristic strength of indicative character descriptor, and as the foundation for updating feature descriptor.For example, the spy being described below
When levying library update, the feature weight can be used for rejecting weaker feature descriptor.
For example, setting 0 for the weight of all feature descriptors when initial.As mentioned above it is possible, in robot self-localization
In, it needs to match the characteristic point in the image of robot shooting with the characteristic point in the feature database of global map.If
The successful number of Feature Points Matching in feature database is more, then this feature point turns out more useful, therefore by the spy of this feature point
Sign weight increases with the increase of successful match number.Alternatively, distinction it is strong the corresponding weight of feature descriptor it is higher, i.e.,
Weight corresponding with the big feature descriptor of surrounding point difference is higher.Further, since longer there are the time, then its is corresponding reliable
Property is lower, so that the feature weight that time longer feature descriptor will be present is arranged smaller.
As mentioned above it is possible, being only completed the primary building of feature database in step S101~S106.However, in order to adapt to
The actual conditions of large scene and scene change, it is far from being enough that this feature library, which once constructs, it is also necessary to continuous to update and expand
Exhibition.
On the basis of the feature database once constructed, can based on the image data newly obtained come to existing feature database into
Row increases and/or deletes.Specifically, in the new close-perspective recording of terminal device for the another position being located in the particular space environment
An image is taken the photograph.The processing similar with the above step S101~S105 is executed for the image, i.e. extraction characteristic point
And characteristic point is clustered.Whether the description of feature corresponding to this feature point is increased for space node belonging to characteristic point
Symbol, that is to say, that in the case where judging to exist for the specific node increased feature descriptor of needs, need to carry out as follows
Determine processing.The first example of determination processing will be described in detail referring to Fig. 3.As shown in figure 3, firstly, judging in step S301
Increased feature descriptor is needed with the presence or absence of for specific node.If be judged as YES in step S301, processing proceeds to
Step S302.In step S302, calculates separately feature descriptor to be increased and the existing each feature of the specific node describes
Similarity between symbol.Wherein, if similarity is greater than second threshold, then it is assumed that the two is similar.Then, in step S303, sentence
Break in existing each feature descriptor with the presence or absence of the spy for being greater than second threshold with the similarity of feature descriptor to be increased
Levy descriptor.If be judged as YES in step S303, processing proceeds to step S304.In step S304, determine without increasing
Add, to avoid invalid operation.On the other hand, if be judged as NO in step S303, it may be assumed that existing each feature descriptor not
In the presence of feature descriptor similar with feature descriptor to be increased, then processing proceeds to step S305.In step S305, determine
Increased.
However, there are in the case where the upper limit for the feature descriptor for including in each minimum space node, it is also necessary to further
Determine whether the feature descriptor after increasing is more than the upper limit.If it exceeds the upper limit, then need to reject a part of feature descriptor.
Next, the second example that determination processing will be described in detail referring to Fig. 4.
Specifically, as shown in figure 4, step S401~S404 is similar with step S301~S304 in Fig. 3.With Fig. 3 institute
Unlike, having determined existing each feature descriptor, there is no feature similar with feature descriptor to be increased descriptions
In the case where symbol, processing proceeds to step S405.In step S405, the feature descriptor at the specific node is further judged
Quantity whether be less than first threshold, i.e. upper limit value.
If be judged as YES in step S405, that is, the quantity of the feature descriptor at the specific node has not yet been reached
Limit, then processing proceeds to step S406.In step S406, determination is increased.
On the other hand, if be judged as NO in step S405, that is, the quantity of the feature descriptor at the specific node is
Reached the upper limit, then processing proceeds to step S407.In step S407, the feature descriptor of the quantity beyond upper limit value is rejected.
Here it is possible to randomly reject the feature descriptor of the quantity beyond upper limit value among all feature descriptors.
Certainly, this is not optimal embodiment.Embodiment more preferably, as mentioned above it is possible, each feature descriptor
In the case where with corresponding feature weight, the existing each feature of specific node is replaced with feature descriptor to be increased and is retouched
The smallest feature descriptor of feature weight in symbol is stated, and the weight of the feature descriptor newly increased is set as initial value.
As mentioned above it is possible, the detailed process for how realizing the real-time update of feature database is described in detail.Next, will
Information processing equipment according to an embodiment of the present invention is described referring to Fig. 5.Information processing equipment according to an embodiment of the present invention is used for
Three-dimensional map is pre-created to specific environment, in order to which the terminal device in the particular space environment can be according to described
Three-dimensional map determines the position of oneself.
As shown in figure 5, the information processing equipment 500 includes: storage unit 501, communication unit 502, extraction unit
503, three-dimensional position determination unit 504, cluster cell 505 and control unit 506.
Storage unit 501 is for storing the three-dimensional map being pre-created.
Communication unit 502 obtains the image shot by the terminal device.
Extraction unit 503 extracts the characteristic point in described image, obtains the feature descriptor for characterizing the characteristic point.
In general, characteristic point is typically selected to have the point of position characteristics, such as vertex or endpoint.Once it is determined that characteristic point, in next step
It is to the corresponding descriptor of each feature point extraction, to distinguish different characteristic points and match identical characteristic point.
Depth information in position and described image of the three-dimensional position determination unit 504 based on the terminal device obtains
The three-dimensional position of the characteristic point.
Cluster cell 505 determines the minimum space node in the particular space environment, and true according to the three-dimensional position
The three-dimensional position for each characteristic point that order member obtains, determines that minimum space node belonging to each characteristic point is obtained to be clustered
All feature descriptors that minimum space inter-node is included.
Control unit 506 makes the storage unit 501 be based on each minimum space node of specific data structure management.Example
Such as, management described here may include retrieval, maintenance, increase, deletion etc..
For example, the specific data structure can be tree.By managing each minimum space node with tree,
Compared with ordinary construction, desired locations can be retrieved more quickly, to extract feature descriptor therein or update therein
Feature descriptor, and existing feature database can be extended simplerly.
More specifically, the tree is octree structure, and wherein outside the minimum of the particular space environment
The root node that cube is the octree structure is connect, the external cube of the minimum is then divided into eight sizes identical the
Eight first order child nodes of one sub-cube as the root node, then respectively each in eight the first sub-cubes
A eight second level child nodes for being divided into identical second sub-cube of eight sizes as the first order child node repeat
Such segmentation is until minimum space node.
In the feature database based on Octree, it may include multiple feature descriptors at each space node.In order to adapt to
Diversified space characteristics demand, while the memory loss of control structure, the quantity of the feature descriptor at each node is limited
It is made as example, N.Therefore, embodiment more preferably, the information processing equipment 500 may further include: selection is single
Member 507, for selecting the feature of first threshold quantity to retouch among all feature descriptors that the cluster cell 505 obtains
State symbol.
In addition, way of example more preferably, the information processing equipment 500 be may further include: weighting is single
Member 508, for assigning corresponding feature weight for each feature descriptor, the feature for being used to indicate feature descriptor is strong
Degree.For example, weighted units 508 by the weight of all feature descriptors set 0 when initial.As mentioned above it is possible, in robot
Characteristic point progress in self-positioning, in the feature database for the characteristic point and global map in image for needing to shoot in robot
Match.If the successful number of Feature Points Matching in feature database is more, this feature point turns out more useful, therefore weighted units
508 increase the feature weight of this feature point with the increase of successful match number.Alternatively, weighted units 508 are by distinction
The corresponding weight of strong feature descriptor is arranged higher, it may be assumed that by weight corresponding with the big feature descriptor of surrounding point difference
It is arranged higher.Further, since longer there are the time, then its corresponding reliability is lower, thus when weighted units 508 will be present
Between the feature weight of longer feature descriptor be arranged smaller.
In order to adapt to the actual conditions of large scene and scene change, this feature library once construct be it is far from being enough, also need
It constantly to update and extend.It, can be based on the image data newly obtained come on the basis of the feature database once constructed
Some feature databases are increased and/or are deleted.Specifically, at the end for the another position being located in the particular space environment
End equipment has taken an image recently.It is single by extraction unit 503, three-dimensional position determination unit 504 and cluster for the image
Member 505 is extracted characteristic point and is clustered to characteristic point.Whether this feature point institute is increased for space node belonging to characteristic point
Corresponding feature descriptor, that is to say, that judging there is the case where increased feature descriptor is needed for specific node
Under, control unit 506 needs to carry out following determine and handles: having calculated separately feature descriptor to be increased and the specific node
The similarity between each feature descriptor having, wherein if similarity is greater than second threshold, then it is assumed that the two is similar.
If there is feature descriptor similar with feature descriptor to be increased in existing each feature descriptor, really
Determine without increasing, to avoid invalid operation.On the other hand, if existing each feature descriptor be not present with it is to be increased
The similar feature descriptor of feature descriptor, it is determined that increased.
However, the feature descriptor for including in each minimum space node is there are in the case where the upper limit, control unit 506 is also
Need to further determine that whether the feature descriptor after increasing is more than the upper limit.If it exceeds the upper limit, then need to reject a part
Feature descriptor.
Specifically, having determined existing each feature descriptor, there is no similar with feature descriptor to be increased
In the case where feature descriptor, control unit 506 further judge the feature descriptor at the specific node quantity whether
Less than first threshold, i.e. upper limit value.
If the judgment is Yes, that is, the upper limit has not yet been reached in the quantity of the feature descriptor at the specific node, it is determined that into
Row increases.
On the other hand, if the judgment is No, that is, the quantity of the feature descriptor at the specific node has reached
Limit then needs to reject the feature descriptor of the quantity beyond upper limit value.
Here it is possible to randomly reject the feature descriptor of the quantity beyond upper limit value among all feature descriptors.
Certainly, this is not optimal embodiment.Embodiment more preferably, as mentioned above it is possible, each feature descriptor
In the case where with corresponding feature weight, the existing each feature of specific node is replaced with feature descriptor to be increased and is retouched
The smallest feature descriptor of feature weight in symbol is stated, and the weight of the feature descriptor newly increased is set as initial value.
Due to being handled in information processing equipment 500, performed by each unit and the above reality according to the present invention
Each step applied in the information processing method of example is completely corresponding.Therefore, in order to avoid redundancy, no longer its details is unfolded to describe.
So far, information processing method and letter according to an embodiment of the present invention is described in detail referring to figs. 1 to Fig. 5
Cease processing equipment.In information processing method according to an embodiment of the present invention and information processing equipment, proposed based on Octree
Data structure suitable for 3d space feature database management.Data structure in this way, can more easily and fast manage and
Feature database is updated, and self-positioning algorithm can be enhanced to the adaptability of environmental change.
It should be noted that in the present specification, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence " including ... ", it is not excluded that including
There is also other identical elements in the process, method, article or equipment of the element.
Finally, it is to be noted that, it is above-mentioned it is a series of processing not only include with sequence described here in temporal sequence
The processing of execution, and the processing including executing parallel or respectively rather than in chronological order.
Through the above description of the embodiments, those skilled in the art can be understood that the present invention can be by
Software adds the mode of required hardware platform to realize, naturally it is also possible to all be implemented by software.Based on this understanding,
Technical solution of the present invention can be embodied in the form of software products in whole or in part to what background technique contributed,
The computer software product can store in storage medium, such as ROM/RAM, magnetic disk, CD, including some instructions are to make
It obtains a computer equipment (can be personal computer, server or the network equipment etc.) and executes each embodiment of the present invention
Or method described in certain parts of embodiment.
The present invention is described in detail above, specific case used herein is to the principle of the present invention and embodiment party
Formula is expounded, and the above description of the embodiment is only used to help understand the method for the present invention and its core ideas；Meanwhile it is right
In those of ordinary skill in the art, according to the thought of the present invention, change is had in specific embodiments and applications
Place, in conclusion the contents of this specification are not to be construed as limiting the invention.
Claims (12)
1. a kind of information processing method is applied to an information processing equipment, comprising to specific sky in the information processing equipment
Between the three-dimensional map that is pre-created of environment, the terminal device in the particular space environment can be according to the three-dimensional map
Determine the position of oneself, described method includes following steps:
Obtain the image shot by the terminal device；
The characteristic point in described image is extracted, the feature descriptor for characterizing the characteristic point is obtained；
Depth information in position and described image based on the terminal device obtains the three-dimensional position of the characteristic point；
Determine the minimum space node in the particular space environment；
According to the three-dimensional position of each characteristic point obtained, determine that minimum space node is belonging to each characteristic point to be clustered,
Obtain all feature descriptors that minimum space inter-node is included；And
Each minimum space node is managed based on specific data structure.
2. information processing method according to claim 1, wherein the specific data structure is octree structure, and
Wherein the external cube of minimum of the particular space environment is the root node of the octree structure, then the minimum
External cube is divided into eight first order child nodes of identical first sub-cube of eight sizes as the root node, so
Each of eight first sub-cubes are divided into identical second sub-cube of eight sizes as described respectively afterwards
Eight second level child nodes of level-one child node repeat such segmentation until minimum space node.
3. information processing method according to claim 1, further comprises:
Among all feature descriptors of acquisition, the feature descriptor of first threshold quantity is selected.
4. information processing method according to claim 3, wherein
Each feature descriptor has corresponding feature weight, is used to indicate the characteristic strength of feature descriptor.
5. further comprising to information processing method described in 4 any one according to claim 1:
It judges whether there is and increased feature descriptor is needed for specific node；
If the judgment is Yes, then feature descriptor and the existing each feature descriptor of the specific node to be increased are calculated separately
Between similarity, wherein if similarity be greater than second threshold, then it is assumed that the two is similar；
If there is feature descriptor similar with feature descriptor to be increased in existing each feature descriptor, it is determined that no
Increased.
6. information processing method according to claim 4, further comprises:
It judges whether there is and increased feature descriptor is needed for specific node；
If the judgment is Yes, then feature descriptor and the existing each feature descriptor of the specific node to be increased are calculated separately
Between similarity, wherein if similarity be greater than second threshold, then it is assumed that the two is similar；
If feature descriptor similar with feature descriptor to be increased is not present in existing each feature descriptor, judge
Whether the quantity of the feature descriptor at the specific node is less than first threshold；
If the judgment is Yes, it is determined that increased；
If the judgment is No, then it is replaced with feature descriptor to be increased special in the existing each feature descriptor of specific node
The smallest feature descriptor of weight is levied, and the weight of the feature descriptor newly increased is set as initial value.
7. a kind of information processing equipment, described specific in order to be in for three-dimensional map to be pre-created to particular space environment
Terminal device in space environment can determine the position of oneself according to the three-dimensional map, and the equipment includes:
Storage unit, for storing the three-dimensional map being pre-created；
Communication unit, for obtaining the image shot by the terminal device；
Extraction unit obtains the feature descriptor for characterizing the characteristic point for extracting the characteristic point in described image；
Three-dimensional position determination unit obtains institute for the depth information in position and described image based on the terminal device
State the three-dimensional position of characteristic point；
Cluster cell is determined for determining the minimum space node in the particular space environment, and according to the three-dimensional position
The three-dimensional position for each characteristic point that unit obtains, determines that minimum space node belonging to each characteristic point obtains most to be clustered
All feature descriptors that small space inter-node is included；And
Control unit, so that the storage unit is based on specific data structure and manages each minimum space node.
8. information processing equipment according to claim 7, wherein the specific data structure is octree structure, and
Wherein the external cube of minimum of the particular space environment is the root node of the octree structure, then the minimum
External cube is divided into eight first order child nodes of identical first sub-cube of eight sizes as the root node, so
Each of eight first sub-cubes are divided into identical second sub-cube of eight sizes as described respectively afterwards
Eight second level child nodes of level-one child node repeat such segmentation until minimum space node.
9. information processing equipment according to claim 7, further comprises:
Selecting unit, for selecting the spy of first threshold quantity among all feature descriptors that the cluster cell obtains
Levy descriptor.
10. information processing equipment according to claim 9, further comprises:
Weighted units are used to indicate feature descriptor for assigning corresponding feature weight for each feature descriptor
Characteristic strength.
11. according to information processing equipment described in claim 7 to 10 any one, wherein
Described control unit, which is judged whether there is, needs increased feature descriptor for specific node；
If the judgment is Yes, then described control unit calculates separately feature descriptor to be increased and the specific node is existing
Similarity between each feature descriptor, wherein if similarity is greater than second threshold, then it is assumed that the two is similar；
If there is feature descriptor similar with feature descriptor to be increased, the control in existing each feature descriptor
Unit processed is determined without increasing.
12. information processing equipment according to claim 10, wherein
Described control unit, which is judged whether there is, needs increased feature descriptor for specific node；
If the judgment is Yes, then described control unit calculates separately feature descriptor to be increased and the specific node is existing
Similarity between each feature descriptor, wherein if similarity is greater than second threshold, then it is assumed that the two is similar；
It is described if feature descriptor similar with feature descriptor to be increased is not present in existing each feature descriptor
Control unit judges whether the quantity of the feature descriptor at the specific node is less than first threshold；
If the judgment is Yes, then described control unit determination is increased；
If the judgment is No, then described control unit with feature descriptor to be increased replaces the existing each spy of specific node
The smallest feature descriptor of feature weight in descriptor is levied, and the weight of the feature descriptor newly increased is set as initial value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410291261.3A CN105335377B (en) | 2014-06-25 | 2014-06-25 | Information processing method and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410291261.3A CN105335377B (en) | 2014-06-25 | 2014-06-25 | Information processing method and equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105335377A CN105335377A (en) | 2016-02-17 |
| CN105335377B true CN105335377B (en) | 2019-03-29 |
Family
ID=55285918
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410291261.3A Active CN105335377B (en) | 2014-06-25 | 2014-06-25 | Information processing method and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105335377B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108268514A (en) * | 2016-12-30 | 2018-07-10 | 乐视汽车（北京）有限公司 | High in the clouds map map rejuvenation equipment based on Octree |
| CN114022721A (en) * | 2021-11-23 | 2022-02-08 | 浙江商汤科技开发有限公司 | Image feature point selection method, related device, equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101996420A (en) * | 2009-08-21 | 2011-03-30 | 索尼公司 | Information processing device, information processing method and program |
| CN102147812A (en) * | 2011-03-31 | 2011-08-10 | 中国科学院自动化研究所 | Three-dimensional point cloud model-based landmark building image classifying method |
| CN103398717A (en) * | 2013-08-22 | 2013-11-20 | 成都理想境界科技有限公司 | Panoramic map database acquisition system and vision-based positioning and navigating method |
| CN103712617A (en) * | 2013-12-18 | 2014-04-09 | 北京工业大学 | Visual-content-based method for establishing multi-level semantic map |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8857703B2 (en) * | 2012-02-15 | 2014-10-14 | International Business Machines Corporation | Mapping an image to an object using a matrix code |
-
2014
- 2014-06-25 CN CN201410291261.3A patent/CN105335377B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101996420A (en) * | 2009-08-21 | 2011-03-30 | 索尼公司 | Information processing device, information processing method and program |
| CN102147812A (en) * | 2011-03-31 | 2011-08-10 | 中国科学院自动化研究所 | Three-dimensional point cloud model-based landmark building image classifying method |
| CN103398717A (en) * | 2013-08-22 | 2013-11-20 | 成都理想境界科技有限公司 | Panoramic map database acquisition system and vision-based positioning and navigating method |
| CN103712617A (en) * | 2013-12-18 | 2014-04-09 | 北京工业大学 | Visual-content-based method for establishing multi-level semantic map |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105335377A (en) | 2016-02-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Doersch et al. | What makes paris look like paris? | |
| US9911340B2 (en) | Real-time system for multi-modal 3D geospatial mapping, object recognition, scene annotation and analytics | |
| Workman et al. | Wide-area image geolocalization with aerial reference imagery | |
| CN112802204B (en) | Target semantic navigation method and system for three-dimensional space scene prior in unknown environment | |
| CN103745498B (en) | A kind of method for rapidly positioning based on image | |
| EP3274964B1 (en) | Automatic connection of images using visual features | |
| CN105224582B (en) | Information processing method and equipment | |
| Han et al. | Development of a hashing-based data structure for the fast retrieval of 3D terrestrial laser scanned data | |
| CN106469190A (en) | Three-dimensional scenic management method and three-dimensional scenic management system | |
| Wang et al. | Automatic segmentation of urban point clouds based on the Gaussian map | |
| CN105335377B (en) | Information processing method and equipment | |
| She et al. | 3D building model simplification method considering both model mesh and building structure | |
| CN116266359A (en) | Target tracking method, device, computer equipment and storage medium | |
| CN105512194A (en) | Game scene management method and device | |
| CN108205820A (en) | Method for reconstructing, fusion method, device, equipment and the storage medium of plane | |
| CN105677843B (en) | A kind of automatic acquisition ancestor four to attribute method | |
| CN117036653A (en) | Point cloud segmentation method and system based on super voxel clustering | |
| CN111415406A (en) | Hierarchical and block-divided three-dimensional model data classification compression method | |
| JP2018156458A (en) | Creation device, creation method, and creation program | |
| CN114586075A (en) | Visual object instance descriptor for location identification | |
| CN109697464A (en) | Method and system based on the identification of the precision target of object detection and signature search | |
| CN106469437B (en) | Image processing method and image processing apparatus | |
| Yuan et al. | Automatic foreground extraction based on difference of Gaussian | |
| Xiao et al. | Confidence map based 3D cost aggregation with multiple minimum spanning trees for stereo matching | |
| CN114742995B (en) | Indoor positioning method based on digital twin building and heterogeneous feature fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |