CN107273831A - A kind of Three-dimensional target recognition method based on spherical space - Google Patents
A kind of Three-dimensional target recognition method based on spherical space Download PDFInfo
- Publication number
- CN107273831A CN107273831A CN201710415042.5A CN201710415042A CN107273831A CN 107273831 A CN107273831 A CN 107273831A CN 201710415042 A CN201710415042 A CN 201710415042A CN 107273831 A CN107273831 A CN 107273831A
- Authority
- CN
- China
- Prior art keywords
- target
- dimensional
- image
- point
- spherical space
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
- G06V20/653—Three-dimensional objects by matching three-dimensional models, e.g. conformal mapping of Riemann surfaces
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of Three-dimensional target recognition method based on spherical space, including：Depth image is obtained using depth transducer, target three-dimensional information is obtained；Depth image is pre-processed, the partial noise in depth image is filtered out；Obtain the three-dimensional coordinate parameter of target image each point；Spherical space is built, log-polar histogram is built according to target point cloud coordinate, three-dimensional feature descriptor is formed；Matching degree is calculated, the matching angle value between different images is obtained.The present invention can carry out the extraction and effectively expression of feature to picture shape and depth information, with scale invariability, rotational invariance and translation invariance, improve the accuracy rate of identification.
Description
Technical field
The present invention relates to a kind of Three-dimensional target recognition method based on spherical space, belong to Shape Matching Technique field.
Background technology
Target identification is the hot issue in computer vision, and key effect is played in man-machine interaction, and it is studied and sent out
Exhibition makes man-machine interaction more flexible, makes the combination of smart machine and destination object even closer.These smart machines are answered
The various aspects of human lives and industrial development are used, therefore Target Recognition Algorithms are widely studied.
Three main periods of the development experience of Target Recognition Algorithms.First period is two-dimensional static target identification,
It is simplest target identification, some simple targets is recognized by obtaining two-dimensional signal.This technology can only recognize static state
Target, it is impossible to discover the lasting change of target.Second period is two-dimentional dynamic object recognition, by obtaining dynamic object
Information, can carry out more effective man-machine interaction.With the appearance of depth camera, the Kinect sensor of such as Microsoft, target
Three-dimensional information can be obtained well, therefore Three-dimensional target recognition has been widely studied.
There are many different Three-dimensional target recognition algorithms to be suggested, based on the type of feature used in them, three
Dimension target identification method can be divided into two classes：Method based on global characteristics and the method based on local feature.Scale invariant is special
It is a kind of classic algorithm in form fit to levy change scaling method, can be used for target identification.However, this method excessively dependence office
The gradient of portion's area pixel, therefore deficient in stability.HMM is another widely used target identification method,
This method is effectively but calculating is sufficiently complex.Can be while the target identification method of guaranteed efficiency and accuracy rate still requires study.
Therefore, for above-mentioned technical problem, it is necessary to provide a kind of Three-dimensional target recognition method based on spherical space.
The content of the invention
The goal of the invention of the present invention is to provide a kind of Three-dimensional target recognition method based on spherical space.
To achieve the above object of the invention, the technical solution adopted by the present invention is：A kind of objective based on spherical space is known
Other method, it is characterised in that methods described comprises the following steps：
S1, using depth transducer obtain target depth image, obtain target three-dimensional information, as shown in Figure 2；
S2, depth image is pre-processed, filter out the partial noise in depth image；
S3, the three-dimensional coordinate parameter for obtaining target image each point；
S4, structure spherical space, build log-polar histogram according to target point cloud coordinate, form three-dimensional feature descriptor, such as
Shown in Fig. 3, respectively target three-dimensional information figure, spherical space divide schematic diagram and log-polar histogram from left to right；
S5, by calculating matching degree, obtain the angle value that matches between target image and each template image, matching angle value is smaller then
Similarity is bigger.
Preferably, step S4 is specifically included：
S41, by certain point on target image and remaining there is a directed connection to form vector, calculate the Euclidean distance and phase of vector
For the angle of horizontal line and vertical line；
S42,12 parts will be divided into respectively relative to the angle of horizontal line and vertical line, every 30 ° of units, and obtain maximum
The logarithm of distance, and 5 parts are divided into, using angle as row, the logarithm of distance is row, forms the ball of one 12 × 12 × 5 dimension
Space matrix；
S43, calculate each point and remaining a little between angle and distance logarithm, and fallen in corresponding spherical space square
In grid array；
S44, statistics fall the points in each spherical space matrix lattice, you can obtain the log-polar histogram of this point, i.e., should
The characteristic vector of point.
Preferably, the step S5 is specifically included：
S51, it will be combined by the characteristic vector of the obtained each target points of step S4, altogether n profile point, you can obtain one
The dimensional feature matrix of individual n × 720；
S52, calculate certain point in target image and match angle value with certain point in template image；
Smallest match totle drilling cost between S53, two images of calculating；
S54, using the matching degree between T transformation calculations images, the matching smaller then target image of angle value is more similar to template image.
Because above-mentioned technical proposal is used, the present invention has following advantages compared with prior art：
The Three-dimensional target recognition method based on spherical space of the invention in Auto-matching and identifying system, can to target shape with
And depth information carries out the extraction and effectively expression of feature, with scale invariability, rotational invariance and translation invariance, improves
The accuracy rate of identification.
Brief description of the drawings
Fig. 1 is the particular flow sheet of the Three-dimensional target recognition method of the invention based on spherical space；
Fig. 2 be using depth transducer obtain comprising three-dimensional information diagram be intended to, from left to right respectively target artwork and
Target three-dimensional information figure；
Fig. 3 is the feature extraction schematic diagram in the three-dimensional information figure in the embodiment of the invention, is respectively from left to right
Target three-dimensional information figure, spherical space divide schematic diagram and log-polar histogram.
Embodiment
Below in conjunction with the accompanying drawings and embodiment the invention will be further described：
Embodiment one：Shown in Figure 1, a kind of Three-dimensional target recognition method based on spherical space, methods described includes following step
Suddenly：
S1, using depth transducer obtain target depth image, obtain target three-dimensional information, as shown in Figure 2；
S2, depth image is pre-processed, filter out the partial noise in depth image；
S3, the three-dimensional coordinate parameter for obtaining target image each point；
S4, structure spherical space, build log-polar histogram according to target point cloud coordinate, form three-dimensional feature descriptor, such as
Shown in Fig. 3, respectively target three-dimensional information figure, spherical space divide schematic diagram and log-polar histogram from left to right；
Step S4 is specifically included：
S41, by certain point on target image and remaining there is a directed connection to form vector, calculate the Euclidean distance and phase of vector
For the angle of horizontal line and vertical line；
S42,12 parts will be divided into respectively relative to the angle of horizontal line and vertical line, every 30 ° of units, and obtain maximum
The logarithm of distance, and 5 parts are divided into, using angle as row, the logarithm of distance is row, forms the ball of one 12 × 12 × 5 dimension
Space matrix；
S43, calculate each point and remaining a little between angle and distance logarithm, and fallen in corresponding spherical space square
In grid array；
S44, statistics fall the points in each spherical space matrix lattice, you can obtain the log-polar histogram of this point, i.e., should
The characteristic vector of point.
S5, by calculating matching degree, obtain the angle value that matches between target image and each template image, matching angle value is got over
Small then similarity is bigger.
Above-mentioned steps S5 is specifically included：
S51, it will be combined by the characteristic vector of the obtained each target points of step S4, altogether n profile point, you can obtain one
The dimensional feature matrix of individual n × 720；
S52, calculate certain point in target image and match angle value with certain point in template image；
Smallest match totle drilling cost between S53, two images of calculating；
S54, using the matching degree between T transformation calculations images, the matching smaller then target image of angle value is more similar to template image
Angle value D calculations are matched in step S5 is：
Assuming that target image points are n, template image points are m.Qj points on Pi points on target image and template image are carried out
Matching, formula Cij=C (Pi, Qj) represents the matching degree between the two points：
,
Wherein hi (k) and hj (k) are respectively Pi points and the corresponding histogram value of Qj points, K=60.
All paired matching degree C set is given, the minimum total cost of matching is calculated, is expressed as：
A corresponding relation has been found, is become using T bring the change for weighing different images afterwards.
Matching degree between image is calculated as follows：
Matching angle value is smaller, and target image and template image similarity are bigger.
As can be seen from the above technical solutions, the Three-dimensional target recognition method based on spherical space that the present invention is provided is automatic
In matching and identification, not only to the shape of image, the extraction and effectively expression of feature are also carried out to its depth information, three are formed
Dimensional feature descriptor, with scale invariability, rotational invariance and translation invariance, improves the accuracy rate of identification.
The foregoing description of the disclosed embodiments, enables those skilled in the art to realize or using the present invention.To this
A variety of modifications of a little embodiments will be apparent for a person skilled in the art, the general original defined in the present embodiment
Reason can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, the present invention will not
Be intended to be limited to the embodiments shown herein, and be to fit to it is consistent with principles disclosed herein and features of novelty most
Wide scope.
Claims (3)
1. a kind of Three-dimensional target recognition method based on spherical space, it is characterised in that methods described comprises the following steps：
S1, using depth transducer obtain target depth image, obtain target three-dimensional information；
S2, depth image is pre-processed, filter out the partial noise in depth image；
S3, the three-dimensional coordinate parameter for obtaining target image each point；
S4, structure spherical space, build log-polar histogram according to target point cloud coordinate, form three-dimensional feature descriptor；
S5, by calculating matching degree, obtain the angle value that matches between target image and each template image, matching angle value is smaller then
Similarity is bigger.
2. Three-dimensional target recognition method according to claim 1, it is characterised in that：Step S4 is specifically included：
S41, by certain point on target image and remaining there is a directed connection to form vector, calculate the Euclidean distance and phase of vector
For the angle of horizontal line and vertical line；
S42,12 parts will be divided into respectively relative to the angle of horizontal line and vertical line, every 30 ° of units, and obtain maximum
The logarithm of distance, and 5 parts are divided into, using angle as row, the logarithm of distance is row, forms the ball of one 12 × 12 × 5 dimension
Space matrix；
S43, calculate each point and remaining a little between angle and distance logarithm, and fallen in corresponding spherical space square
In grid array；
S44, statistics fall the points in each spherical space matrix lattice, obtain the log-polar histogram of this point, the i.e. point
Characteristic vector.
3. Three-dimensional target recognition method according to claim 1, it is characterised in that：The step S5 is specifically included：
S51, it will be combined by the characteristic vector of the obtained each target points of step S4, altogether n profile point, you can obtain one
The dimensional feature matrix of individual n × 720；
S52, calculate certain point in target image and match angle value with certain point in template image；
Smallest match totle drilling cost between S53, two images of calculating；
S54, using the matching degree between T transformation calculations images, the matching smaller then target image of angle value is more similar to template image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710415042.5A CN107273831A (en) | 2017-06-05 | 2017-06-05 | A kind of Three-dimensional target recognition method based on spherical space |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710415042.5A CN107273831A (en) | 2017-06-05 | 2017-06-05 | A kind of Three-dimensional target recognition method based on spherical space |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107273831A true CN107273831A (en) | 2017-10-20 |
Family
ID=60065473
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710415042.5A Pending CN107273831A (en) | 2017-06-05 | 2017-06-05 | A kind of Three-dimensional target recognition method based on spherical space |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107273831A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344750A (en) * | 2018-09-20 | 2019-02-15 | 浙江工业大学 | A kind of labyrinth three dimensional object recognition methods based on Structural descriptors |
| CN110276266A (en) * | 2019-05-28 | 2019-09-24 | 暗物智能科技(广州)有限公司 | A kind of processing method, device and the terminal device of the point cloud data based on rotation |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102208032A (en) * | 2011-07-04 | 2011-10-05 | 徐杰 | Spherical re-sampling-based three-dimensional face identification |
| CN103136520A (en) * | 2013-03-25 | 2013-06-05 | 苏州大学 | Shape matching and target recognition method based on PCA-SC algorithm |
| CN103519788A (en) * | 2013-10-18 | 2014-01-22 | 南京师范大学 | Attention scenario evaluation system based on Kinect interaction |
| CN103544492A (en) * | 2013-08-06 | 2014-01-29 | Tcl集团股份有限公司 | Method and device for identifying targets on basis of geometric features of three-dimensional curved surfaces of depth images |
| CN103777220A (en) * | 2014-01-17 | 2014-05-07 | 西安交通大学 | Real-time and accurate pose estimation method based on fiber-optic gyroscope, speed sensor and GPS |
| CN104318616A (en) * | 2014-11-07 | 2015-01-28 | 钟若飞 | Colored point cloud system and colored point cloud generation method based on same |
| CN104699234A (en) * | 2013-12-05 | 2015-06-10 | 浙江大学 | Three-dimensional space imaging interaction method and three-dimensional space imaging interaction based on laser |
| CN105678348A (en) * | 2016-01-07 | 2016-06-15 | 陕西师范大学 | Normative evaluation method and system for handwritten Chinese characters |
| CN105719278A (en) * | 2016-01-13 | 2016-06-29 | 西北大学 | Organ auxiliary positioning segmentation method based on statistical deformation model |
-
2017
- 2017-06-05 CN CN201710415042.5A patent/CN107273831A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102208032A (en) * | 2011-07-04 | 2011-10-05 | 徐杰 | Spherical re-sampling-based three-dimensional face identification |
| CN103136520A (en) * | 2013-03-25 | 2013-06-05 | 苏州大学 | Shape matching and target recognition method based on PCA-SC algorithm |
| CN103544492A (en) * | 2013-08-06 | 2014-01-29 | Tcl集团股份有限公司 | Method and device for identifying targets on basis of geometric features of three-dimensional curved surfaces of depth images |
| CN103519788A (en) * | 2013-10-18 | 2014-01-22 | 南京师范大学 | Attention scenario evaluation system based on Kinect interaction |
| CN104699234A (en) * | 2013-12-05 | 2015-06-10 | 浙江大学 | Three-dimensional space imaging interaction method and three-dimensional space imaging interaction based on laser |
| CN103777220A (en) * | 2014-01-17 | 2014-05-07 | 西安交通大学 | Real-time and accurate pose estimation method based on fiber-optic gyroscope, speed sensor and GPS |
| CN104318616A (en) * | 2014-11-07 | 2015-01-28 | 钟若飞 | Colored point cloud system and colored point cloud generation method based on same |
| CN105678348A (en) * | 2016-01-07 | 2016-06-15 | 陕西师范大学 | Normative evaluation method and system for handwritten Chinese characters |
| CN105719278A (en) * | 2016-01-13 | 2016-06-29 | 西北大学 | Organ auxiliary positioning segmentation method based on statistical deformation model |
Non-Patent Citations (1)
| Title |
|---|
| 张曾琪 等: "HeartVoice:基于Kinect的手语识别", 《南京航空航天大学第六届本科生学术论坛论文选集》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344750A (en) * | 2018-09-20 | 2019-02-15 | 浙江工业大学 | A kind of labyrinth three dimensional object recognition methods based on Structural descriptors |
| CN109344750B (en) * | 2018-09-20 | 2021-10-22 | 浙江工业大学 | Complex structure three-dimensional object identification method based on structure descriptor |
| CN110276266A (en) * | 2019-05-28 | 2019-09-24 | 暗物智能科技(广州)有限公司 | A kind of processing method, device and the terminal device of the point cloud data based on rotation |
| CN110276266B (en) * | 2019-05-28 | 2021-09-10 | 暗物智能科技(广州)有限公司 | Rotation-based point cloud data processing method and device and terminal equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106716450B (en) | Image-based feature detection using edge vectors | |
| CN109214282B (en) | A kind of three-dimension gesture critical point detection method and system neural network based | |
| WO2020259481A1 (en) | Positioning method and apparatus, electronic device, and readable storage medium | |
| Ma et al. | Kinect Sensor‐Based Long‐Distance Hand Gesture Recognition and Fingertip Detection with Depth Information | |
| Ren et al. | Depth camera based hand gesture recognition and its applications in human-computer-interaction | |
| CN105096377B (en) | A kind of image processing method and device | |
| KR101486543B1 (en) | Method and apparatus for recognition and segmentation object for 3d object recognition | |
| US20170236033A1 (en) | Method of providing image feature descriptors | |
| CN108629843B (en) | Method and equipment for realizing augmented reality | |
| CN108596193B (en) | Method and system for building deep learning network structure aiming at human ear recognition | |
| CN103177269A (en) | Equipment and method used for estimating object posture | |
| CN104298995B (en) | Three-dimensional face identifying device and method based on three-dimensional point cloud | |
| CN104050475A (en) | Reality augmenting system and method based on image feature matching | |
| CN102737235A (en) | Head posture estimation method based on depth information and color image | |
| CN110809785A (en) | Learning-based matching for active stereo systems | |
| US20180314909A1 (en) | Detection and recognition of objects lacking textures | |
| US20220130175A1 (en) | Method and apparatus for detecting liveness based on phase difference | |
| CN107330980A (en) | A kind of virtual furnishings arrangement system based on no marks thing | |
| CN107273831A (en) | A kind of Three-dimensional target recognition method based on spherical space | |
| Azis et al. | Substitutive skeleton fusion for human action recognition | |
| CN110705363B (en) | Commodity specification identification method and device | |
| CN116843867A (en) | Augmented reality virtual-real fusion method, electronic device and storage medium | |
| JP6393495B2 (en) | Image processing apparatus and object recognition method | |
| CN102110303A (en) | Method for compounding face fake portrait\fake photo based on support vector return | |
| WO2017179728A1 (en) | Image recognition device, image recognition method, and image recognition program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171020 |
|
| RJ01 | Rejection of invention patent application after publication |