CN107392953A - Depth image recognition methods based on contour - Google Patents
Depth image recognition methods based on contour Download PDFInfo
- Publication number
- CN107392953A CN107392953A CN201710854890.6A CN201710854890A CN107392953A CN 107392953 A CN107392953 A CN 107392953A CN 201710854890 A CN201710854890 A CN 201710854890A CN 107392953 A CN107392953 A CN 107392953A
- Authority
- CN
- China
- Prior art keywords
- depth image
- contour
- closed curve
- depth
- image recognition
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to field of image recognition, discloses a kind of depth image recognition methods based on contour, solves the problems, such as that traditional characteristic extracting method can not be stated directly the feature of depth image very well.The present invention is equal by pixel value in depth image, or point of the gap in certain threshold range connects, and forms depth image contour, you can image is identified by information such as the shape of contour, differences in height.The present invention is applied to depth image and identified.
Description
Technical field
The present invention relates to field of image recognition, the depth image recognition methods more particularly to based on contour.
Background technology
With the proposition of the concepts such as smart city, intelligent domestic, society's science and technology is fast-developing towards intelligent direction, tradition
Machine vision can not meet the needs of people are for Three-dimension object recognition.Each point is relative to video camera in acquisition scene
Distance is one of vital task of computer vision system, and each pixel represents certain point and video camera in scene in image
Distance, this image is depth image.Compared with gray level image, depth image has the three-dimensional feature information of object, and
And depth image can accurately show information on target object not by illumination effect.Depth image is with three-dimensional visual sensor institute
Based on obtained figure, image, by recovering the 3D shape of object and carrying out image procossing, possess at high speed, high efficiency
Feature, need to carry out somatometric industry suitable for medical science, archaeology, clothes manufacture, engraving, artificial limb etc..Depth image
Processing is different with gray-scale map, coloured image from identifying, depth image processing needs to consider gradation of image, depth information, transports
Carried out completing depth image processing and identification with new image-recognizing method.
The present invention proposes a kind of depth image recognition methods based on contour.Contour is geography noun, is referred to
The closed curve that the equal adjacent spots of elevation are linked to be on topographic map.It is bent that height above sea level identical point on ground is linked to be closure
Line, and upright projection is on a horizontal plane, and contracting is painted on drawing in proportion, just obtains contour.It can also regard as not
With the horizontal plane of height above sea level and the intersection of actual ground, the numeral marked on contour is the height above sea level of the contour.Depth map
Each pixel value represents the distance of object and camera position as in, similar with height above sea level concept, therefore can be by contour concept
Combined with depth image, for identifying image information, the depth image recognition methods as of the present invention based on contour.
Image recognition refers to handle image using computer, analyzed and understood, various without pattern to identify
The technology of target and object, based on the principal character of image, the characteristic information amount of image is larger for image recognition, people often with
Characteristics of image differentiates different objects and shape, and such as alphabetical " A " has a point, and alphabetical " B " has two circles to the right etc., conventional
Image characteristic extracting method has the methods of grey level histogram, Fourier transformation, principal component analysis Linear feature extraction, but in depth
In image, due to graphical representation be object steric information, the conventional feature extracting method of some of the above can not be directly to depth
The feature for spending image carries out statement well, it is necessary to find a kind of feature extracting method that can directly state depth image.
The content of the invention
The technical problem to be solved in the present invention is：A kind of depth image recognition methods based on contour is provided, solves to pass
The problem of system feature extracting method can not be stated directly the feature of depth image very well.
To solve the above problems, the technical solution adopted by the present invention is：Depth image recognition methods based on contour, bag
Include step：
S1. depth image is obtained using depth image video camera；
S2. the minimum pixel value A and max pixel value M in depth map is found；
S3. set depth threshold k；
S4. since A pixel values, the pixel value that height difference is less than or equal to threshold k is classified as one kind, and connects such
All pixels point forms closed curve；
S5. according to the size, sum and distribution situation of closed curve, judge whether closed curve meets the requirements,
If so, then jump procedure S6, otherwise removes closed curve, and jump procedure S3；
S6. template matches, ratio match are performed to closed curve, completes image recognition, complete image recognition.
Further, it is undesirable if situations below occurs in closed curve in step S5：Closed curve is excessive or mistake
Small, closed curve sum is small, and distribution is more sparse.
Further, in step S1, the depth image video camera is Microsoft Kinect.
Further, if step S6 carries out human bioequivalence, by judging whether X/Y value is in the range of Z to sentence
Whether the contour body that breaks is human body, wherein, X is that the maximum in depth image apart from video camera minimum distance similar round closed curve is straight
Footpath, Y are that Z is human body head width and shoulder apart from the maximum gauge of video camera time minimum distance closed curve in depth image
The proportion of width.
The beneficial effects of the invention are as follows：The present invention applies to geography contour in depth image, can be to depth map
Depth information as in is comprehensively stated, and pixel value in depth image is equal, or gap is in certain threshold range
Point connects, and forms depth image contour, you can image is known by information such as the shape of contour, differences in height
Not.
Brief description of the drawings
Fig. 1 is the flow chart of embodiment.
Embodiment
Embodiment provides a kind of depth image recognition methods based on contour, and it is special to extract image using contour line method
Sign, that is, profile information, the range information of objects in images are extracted, so as to which image be identified, specific implementation method is as follows：
S1. depth image video camera is used, such as Microsoft Kinect, obtains depth image.
S2. the minimum pixel value A and max pixel value M in depth map is found.
S3. set depth threshold k.
S4. depth image contour map is formed by interval of K：Since A pixel values, height difference is less than or equal to threshold value
K pixel value is classified as one kind, and connects such all pixels point and form closed curve, and the closed curve is depth image etc.
High line.
S5. according to the size, sum and distribution situation of closed curve, judge whether closed curve meets the requirements,
If meeting, jump procedure S6, if do not meet, such as occur closed curve area is excessive or too small, closed curve it is total it is small, point
The more sparse situation of cloth, then remove closed curve, and jump procedure S3 percentage regulation threshold ks；
S6. the image recognition operations such as template matches, ratio match are performed to closed curve, completes image recognition.
Such as the human body in image is identified using the above method, then implementation steps are：
(1) picture of human body is included from the shooting of people crown oblique upper using depth camera.
(2) pixel value that depth is minimum in image is found out, it is assumed that the value is a, and finds out max pixel value in figure, it is assumed that value
For m.
(3) set depth threshold value k.
(4) depth image contour map is formed by interval of K, i.e., by a to a+k, a+k to a+2k, a+2k to a+3k....a
Pixel in each sections of+nk to m connects into closed curve respectively.
(5) repeat step (3) (4), adjustment threshold value k to image closed curve are clearly presented.
(6) from structoure of the human body, human body head width and the proportional scope Z of shoulder width, then depth map is judged
It is maximum straight with time minimum distance similar round closed curve apart from video camera minimum distance similar round closed curve maximum gauge X as in
Whether footpath Y ratio is in the range of Z, you can judges whether the contour body is human body.
The general principle of the present invention and main feature are the foregoing described, the description of specification simply illustrates the original of the present invention
Reason, without departing from the spirit and scope of the present invention, various changes and modifications of the present invention are possible, these changes and improvements
It all fall within the protetion scope of the claimed invention.
Claims (3)
1. the depth image recognition methods based on contour, it is characterised in that including step：
S1. depth image is obtained using depth image video camera；
S2. the minimum pixel value A and max pixel value M in depth map is found；
S3. set depth threshold k；
S4. since A pixel values, the pixel value that height difference is less than or equal to threshold k is classified as one kind, and connects such all
Pixel forms closed curve；
S5. according to the size, sum and distribution situation of closed curve, judge whether closed curve meets the requirements, if so,
Then jump procedure S6, otherwise remove closed curve, and jump procedure S3；
S6. template matches, ratio match are performed to closed curve, completes image recognition.
2. the depth image recognition methods based on contour as claimed in claim 1, it is characterised in that described in step S1
Depth image video camera is Microsoft Kinect.
3. the depth image recognition methods based on contour as claimed in claim 1, it is characterised in that if step S6 enters pedestrian
Body identifies, then by judging whether X/Y value is in the range of Z to judge whether contour body is human body, wherein, X is depth
Apart from the maximum gauge of video camera minimum distance similar round closed curve in image, Y is secondary most apart from video camera in depth image
The closely maximum gauge of closed curve, Z are the proportion of human body head width and shoulder width.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710854890.6A CN107392953B (en) | 2017-09-20 | 2017-09-20 | Depth image identification method based on contour line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710854890.6A CN107392953B (en) | 2017-09-20 | 2017-09-20 | Depth image identification method based on contour line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107392953A true CN107392953A (en) | 2017-11-24 |
| CN107392953B CN107392953B (en) | 2020-11-10 |
Family
ID=60351047
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710854890.6A Active CN107392953B (en) | 2017-09-20 | 2017-09-20 | Depth image identification method based on contour line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107392953B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108052912A (en) * | 2017-12-20 | 2018-05-18 | 安徽信息工程学院 | A kind of three-dimensional face image recognition methods based on square Fourier descriptor |
| CN110824726A (en) * | 2019-12-06 | 2020-02-21 | 成都工业学院 | Three-dimensional display device with pi-shaped pixel arrangement |
| CN111881733A (en) * | 2020-06-17 | 2020-11-03 | 艾普工华科技（武汉）有限公司 | Worker operation step specification visual identification judgment and guidance method and system |
| CN112132785A (en) * | 2020-08-25 | 2020-12-25 | 华东师范大学 | Transmission electron microscope image recognition and analysis method and system for two-dimensional material |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894254A (en) * | 2010-06-13 | 2010-11-24 | 南开大学 | Contouring method-based three-dimensional face recognition method |
| US20120170832A1 (en) * | 2010-12-31 | 2012-07-05 | Industrial Technology Research Institute | Depth map generation module for foreground object and method thereof |
| US20140118335A1 (en) * | 2012-10-30 | 2014-05-01 | Primesense Ltd. | Depth mapping with enhanced resolution |
| CN105631852A (en) * | 2015-11-03 | 2016-06-01 | 四川长虹电器股份有限公司 | Depth image contour line-based indoor human body detection method |
-
2017
- 2017-09-20 CN CN201710854890.6A patent/CN107392953B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894254A (en) * | 2010-06-13 | 2010-11-24 | 南开大学 | Contouring method-based three-dimensional face recognition method |
| US20120170832A1 (en) * | 2010-12-31 | 2012-07-05 | Industrial Technology Research Institute | Depth map generation module for foreground object and method thereof |
| US20140118335A1 (en) * | 2012-10-30 | 2014-05-01 | Primesense Ltd. | Depth mapping with enhanced resolution |
| CN105631852A (en) * | 2015-11-03 | 2016-06-01 | 四川长虹电器股份有限公司 | Depth image contour line-based indoor human body detection method |
Non-Patent Citations (1)
| Title |
|---|
| 匿名用户: "gis怎么用dem生成等高线", 《HTTPS://WENWEN.SOGOU.COM/Z/Q549492873.HTM》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108052912A (en) * | 2017-12-20 | 2018-05-18 | 安徽信息工程学院 | A kind of three-dimensional face image recognition methods based on square Fourier descriptor |
| CN110824726A (en) * | 2019-12-06 | 2020-02-21 | 成都工业学院 | Three-dimensional display device with pi-shaped pixel arrangement |
| CN110824726B (en) * | 2019-12-06 | 2024-05-28 | 成都工业学院 | Three-dimensional display device with pi-type pixel arrangement |
| CN111881733A (en) * | 2020-06-17 | 2020-11-03 | 艾普工华科技（武汉）有限公司 | Worker operation step specification visual identification judgment and guidance method and system |
| CN111881733B (en) * | 2020-06-17 | 2023-07-21 | 艾普工华科技（武汉）有限公司 | Method and system for judging and guiding worker operation step standard visual identification |
| CN112132785A (en) * | 2020-08-25 | 2020-12-25 | 华东师范大学 | Transmission electron microscope image recognition and analysis method and system for two-dimensional material |
| CN112132785B (en) * | 2020-08-25 | 2023-12-15 | 华东师范大学 | Transmission electron microscope image identification and analysis method and system for two-dimensional material |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107392953B (en) | 2020-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021022970A1 (en) | Multi-layer random forest-based part recognition method and system | |
| CN105069466B (en) | Pedestrian's dress ornament color identification method based on Digital Image Processing | |
| CN104850850B (en) | A kind of binocular stereo vision image characteristic extracting method of combination shape and color | |
| CN107392953A (en) | Depth image recognition methods based on contour | |
| CN102663411B (en) | Recognition method for target human body | |
| CN105046206B (en) | Based on the pedestrian detection method and device for moving prior information in video | |
| CN105678318B (en) | The matching process and device of traffic sign | |
| CN103914699A (en) | Automatic lip gloss image enhancement method based on color space | |
| CN104715238A (en) | Pedestrian detection method based on multi-feature fusion | |
| CN107705288A (en) | Hazardous gas spillage infrared video detection method under pseudo- target fast-moving strong interferers | |
| CN106991370B (en) | Pedestrian retrieval method based on color and depth | |
| CN102096823A (en) | Face detection method based on Gaussian model and minimum mean-square deviation | |
| CN105335725A (en) | Gait identification identity authentication method based on feature fusion | |
| CN111160291B (en) | Human eye detection method based on depth information and CNN | |
| CN106204594A (en) | A kind of direction detection method of dispersivity moving object based on video image | |
| CN102147867B (en) | Method for identifying traditional Chinese painting images and calligraphy images based on subject | |
| CN108470178B (en) | Depth map significance detection method combined with depth credibility evaluation factor | |
| CN110032932B (en) | Human body posture identification method based on video processing and decision tree set threshold | |
| CN104143077B (en) | Pedestrian target search method and system based on image | |
| Tung et al. | Elliptical density shape model for hand gesture recognition | |
| CN109948461A (en) | A kind of sign language image partition method based on center coordination and range conversion | |
| CN109766850A (en) | Fingerprint image matching method based on Fusion Features | |
| Hatzidimos | Automatic traffic sign recognition in digital images | |
| CN106446832B (en) | Video-based pedestrian real-time detection method | |
| CN112381747A (en) | Terahertz and visible light image registration method and device based on contour feature points |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |